id	sid	tid	token	lemma	pos
ejpam-515	1	1	5_xxx_qian.dvi	5_xxx_qian.dvi	NUM
ejpam-515	1	2	european	european	ADJ
ejpam-515	1	3	journal	journal	NOUN
ejpam-515	1	4	of	of	ADP
ejpam-515	1	5	pure	pure	ADJ
ejpam-515	1	6	and	and	CCONJ
ejpam-515	1	7	applied	apply	VERB
ejpam-515	1	8	mathematics	mathematic	NOUN
ejpam-515	1	9	vol	vol	NOUN
ejpam-515	1	10	.	.	PUNCT
ejpam-515	2	1	3	3	NUM
ejpam-515	2	2	,	,	PUNCT
ejpam-515	2	3	no	no	INTJ
ejpam-515	2	4	.	.	NOUN
ejpam-515	2	5	3	3	NUM
ejpam-515	2	6	,	,	PUNCT
ejpam-515	2	7	2010	2010	NUM
ejpam-515	2	8	,	,	PUNCT
ejpam-515	2	9	417	417	NUM
ejpam-515	2	10	-	-	SYM
ejpam-515	2	11	434	434	NUM
ejpam-515	2	12	issn	issn	PROPN
ejpam-515	2	13	1307	1307	NUM
ejpam-515	2	14	-	-	SYM
ejpam-515	2	15	5543	5543	NUM
ejpam-515	2	16	–	–	PUNCT
ejpam-515	2	17	www.ejpam.com	www.ejpam.com	X
ejpam-515	2	18	special	special	ADJ
ejpam-515	2	19	issue	issue	NOUN
ejpam-515	2	20	on	on	ADP
ejpam-515	2	21	granger	granger	PROPN
ejpam-515	2	22	econometrics	econometric	NOUN
ejpam-515	2	23	and	and	CCONJ
ejpam-515	2	24	statistical	statistical	ADJ
ejpam-515	2	25	modeling	modeling	NOUN
ejpam-515	2	26	dedicated	dedicate	VERB
ejpam-515	2	27	to	to	ADP
ejpam-515	2	28	the	the	DET
ejpam-515	2	29	memory	memory	NOUN
ejpam-515	2	30	of	of	ADP
ejpam-515	2	31	prof	prof	NOUN
ejpam-515	2	32	.	.	PUNCT
ejpam-515	3	1	sir	sir	PROPN
ejpam-515	3	2	clive	clive	PROPN
ejpam-515	3	3	w.j	w.j	PROPN
ejpam-515	3	4	.	.	PROPN
ejpam-515	4	1	granger	granger	PROPN
ejpam-515	4	2	law	law	PROPN
ejpam-515	4	3	of	of	ADP
ejpam-515	4	4	iterated	iterated	ADJ
ejpam-515	4	5	logarithm	logarithm	NOUN
ejpam-515	4	6	and	and	CCONJ
ejpam-515	4	7	strong	strong	ADJ
ejpam-515	4	8	consistency	consistency	NOUN
ejpam-515	4	9	in	in	ADP
ejpam-515	4	10	poisson	poisson	PROPN
ejpam-515	4	11	regression	regression	NOUN
ejpam-515	4	12	model	model	NOUN
ejpam-515	4	13	selection	selection	PROPN
ejpam-515	4	14	guogi	guogi	PROPN
ejpam-515	4	15	qian	qian	PROPN
ejpam-515	4	16	department	department	PROPN
ejpam-515	4	17	of	of	ADP
ejpam-515	4	18	mathematics	mathematics	PROPN
ejpam-515	4	19	and	and	CCONJ
ejpam-515	4	20	statistics	statistic	NOUN
ejpam-515	4	21	,	,	PUNCT
ejpam-515	4	22	university	university	NOUN
ejpam-515	4	23	of	of	ADP
ejpam-515	4	24	melbourne	melbourne	PROPN
ejpam-515	4	25	,	,	PUNCT
ejpam-515	4	26	vic	vic	PROPN
ejpam-515	4	27	3010	3010	NUM
ejpam-515	4	28	,	,	PUNCT
ejpam-515	4	29	australia	australia	PROPN
ejpam-515	4	30	abstract	abstract	NOUN
ejpam-515	4	31	.	.	PUNCT
ejpam-515	5	1	in	in	ADP
ejpam-515	5	2	this	this	DET
ejpam-515	5	3	paper	paper	NOUN
ejpam-515	5	4	we	we	PRON
ejpam-515	5	5	first	first	ADV
ejpam-515	5	6	derive	derive	VERB
ejpam-515	5	7	a	a	DET
ejpam-515	5	8	law	law	NOUN
ejpam-515	5	9	of	of	ADP
ejpam-515	5	10	iterated	iterated	ADJ
ejpam-515	5	11	logarithm	logarithm	NOUN
ejpam-515	5	12	for	for	ADP
ejpam-515	5	13	the	the	DET
ejpam-515	5	14	maximum	maximum	ADJ
ejpam-515	5	15	likelihood	likelihood	NOUN
ejpam-515	5	16	estimator	estimator	NOUN
ejpam-515	5	17	of	of	ADP
ejpam-515	5	18	the	the	DET
ejpam-515	5	19	parameters	parameter	NOUN
ejpam-515	5	20	in	in	ADP
ejpam-515	5	21	a	a	DET
ejpam-515	5	22	poisson	poisson	NOUN
ejpam-515	5	23	regression	regression	NOUN
ejpam-515	5	24	model	model	NOUN
ejpam-515	5	25	.	.	PUNCT
ejpam-515	6	1	we	we	PRON
ejpam-515	6	2	then	then	ADV
ejpam-515	6	3	use	use	VERB
ejpam-515	6	4	this	this	DET
ejpam-515	6	5	result	result	NOUN
ejpam-515	6	6	to	to	PART
ejpam-515	6	7	establish	establish	VERB
ejpam-515	6	8	the	the	DET
ejpam-515	6	9	strong	strong	ADJ
ejpam-515	6	10	consistency	consistency	NOUN
ejpam-515	6	11	of	of	ADP
ejpam-515	6	12	a	a	DET
ejpam-515	6	13	class	class	NOUN
ejpam-515	6	14	of	of	ADP
ejpam-515	6	15	model	model	NOUN
ejpam-515	6	16	selection	selection	NOUN
ejpam-515	6	17	criteria	criterion	NOUN
ejpam-515	6	18	in	in	ADP
ejpam-515	6	19	poisson	poisson	PROPN
ejpam-515	6	20	regression	regression	NOUN
ejpam-515	6	21	model	model	NOUN
ejpam-515	6	22	selection	selection	NOUN
ejpam-515	6	23	.	.	PUNCT
ejpam-515	7	1	we	we	PRON
ejpam-515	7	2	show	show	VERB
ejpam-515	7	3	that	that	SCONJ
ejpam-515	7	4	under	under	ADP
ejpam-515	7	5	some	some	DET
ejpam-515	7	6	general	general	ADJ
ejpam-515	7	7	conditions	condition	NOUN
ejpam-515	7	8	,	,	PUNCT
ejpam-515	7	9	a	a	DET
ejpam-515	7	10	model	model	NOUN
ejpam-515	7	11	selection	selection	NOUN
ejpam-515	7	12	criterion	criterion	NOUN
ejpam-515	7	13	,	,	PUNCT
ejpam-515	7	14	which	which	PRON
ejpam-515	7	15	consists	consist	VERB
ejpam-515	7	16	of	of	ADP
ejpam-515	7	17	a	a	DET
ejpam-515	7	18	minus	minus	CCONJ
ejpam-515	7	19	maximum	maximum	ADJ
ejpam-515	7	20	loglikelihood	loglikelihood	NOUN
ejpam-515	7	21	and	and	CCONJ
ejpam-515	7	22	a	a	DET
ejpam-515	7	23	penalty	penalty	NOUN
ejpam-515	7	24	term	term	NOUN
ejpam-515	7	25	,	,	PUNCT
ejpam-515	7	26	will	will	AUX
ejpam-515	7	27	select	select	VERB
ejpam-515	7	28	the	the	DET
ejpam-515	7	29	simplest	simple	ADJ
ejpam-515	7	30	correct	correct	ADJ
ejpam-515	7	31	model	model	NOUN
ejpam-515	7	32	almost	almost	ADV
ejpam-515	7	33	surely	surely	ADV
ejpam-515	7	34	if	if	SCONJ
ejpam-515	7	35	the	the	DET
ejpam-515	7	36	penalty	penalty	NOUN
ejpam-515	7	37	term	term	NOUN
ejpam-515	7	38	increases	increase	VERB
ejpam-515	7	39	with	with	ADP
ejpam-515	7	40	model	model	NOUN
ejpam-515	7	41	dimension	dimension	NOUN
ejpam-515	7	42	and	and	CCONJ
ejpam-515	7	43	has	have	VERB
ejpam-515	7	44	an	an	DET
ejpam-515	7	45	order	order	NOUN
ejpam-515	7	46	in	in	ADP
ejpam-515	7	47	between	between	ADP
ejpam-515	7	48	o(log	o(log	PROPN
ejpam-515	7	49	log	log	PROPN
ejpam-515	7	50	n	n	CCONJ
ejpam-515	7	51	)	)	PUNCT
ejpam-515	7	52	and	and	CCONJ
ejpam-515	7	53	o(n	o(n	NUM
ejpam-515	7	54	)	)	PUNCT
ejpam-515	7	55	.	.	PUNCT
ejpam-515	8	1	2000	2000	NUM
ejpam-515	8	2	mathematics	mathematic	NOUN
ejpam-515	8	3	subject	subject	NOUN
ejpam-515	8	4	classifications	classification	NOUN
ejpam-515	8	5	:	:	PUNCT
ejpam-515	8	6	62f12	62f12	NUM
ejpam-515	8	7	,	,	PUNCT
ejpam-515	8	8	62j12	62j12	NUM
ejpam-515	8	9	,	,	PUNCT
ejpam-515	8	10	60f15	60f15	DET
ejpam-515	8	11	key	key	ADJ
ejpam-515	8	12	words	word	NOUN
ejpam-515	8	13	and	and	CCONJ
ejpam-515	8	14	phrases	phrase	NOUN
ejpam-515	8	15	:	:	PUNCT
ejpam-515	8	16	law	law	NOUN
ejpam-515	8	17	of	of	ADP
ejpam-515	8	18	iterated	iterated	ADJ
ejpam-515	8	19	logarithm	logarithm	NOUN
ejpam-515	8	20	;	;	PUNCT
ejpam-515	8	21	poisson	poisson	NOUN
ejpam-515	8	22	regression	regression	NOUN
ejpam-515	8	23	;	;	PUNCT
ejpam-515	8	24	maximum	maximum	ADJ
ejpam-515	8	25	likelihood	likelihood	NOUN
ejpam-515	8	26	estimator	estimator	NOUN
ejpam-515	8	27	;	;	PUNCT
ejpam-515	8	28	model	model	NOUN
ejpam-515	8	29	selection	selection	NOUN
ejpam-515	8	30	;	;	PUNCT
ejpam-515	8	31	strong	strong	ADJ
ejpam-515	8	32	consistency	consistency	NOUN
ejpam-515	8	33	1	1	NUM
ejpam-515	8	34	.	.	PUNCT
ejpam-515	9	1	introduction	introduction	NOUN
ejpam-515	9	2	poisson	poisson	PROPN
ejpam-515	9	3	regression	regression	NOUN
ejpam-515	9	4	model	model	NOUN
ejpam-515	9	5	is	be	AUX
ejpam-515	9	6	a	a	DET
ejpam-515	9	7	widely	widely	ADV
ejpam-515	9	8	used	use	VERB
ejpam-515	9	9	econometric	econometric	ADJ
ejpam-515	9	10	and	and	CCONJ
ejpam-515	9	11	statistical	statistical	ADJ
ejpam-515	9	12	tool	tool	NOUN
ejpam-515	9	13	for	for	ADP
ejpam-515	9	14	studying	study	VERB
ejpam-515	9	15	the	the	DET
ejpam-515	9	16	relationship	relationship	NOUN
ejpam-515	9	17	between	between	ADP
ejpam-515	9	18	a	a	DET
ejpam-515	9	19	poisson	poisson	ADJ
ejpam-515	9	20	-	-	PUNCT
ejpam-515	9	21	type	type	NOUN
ejpam-515	9	22	response	response	NOUN
ejpam-515	9	23	variable	variable	NOUN
ejpam-515	9	24	and	and	CCONJ
ejpam-515	9	25	a	a	DET
ejpam-515	9	26	set	set	NOUN
ejpam-515	9	27	of	of	ADP
ejpam-515	9	28	explanatory	explanatory	ADJ
ejpam-515	9	29	variables	variable	NOUN
ejpam-515	9	30	.	.	PUNCT
ejpam-515	10	1	a	a	DET
ejpam-515	10	2	familiar	familiar	ADJ
ejpam-515	10	3	example	example	NOUN
ejpam-515	10	4	is	be	AUX
ejpam-515	10	5	the	the	DET
ejpam-515	10	6	analysis	analysis	NOUN
ejpam-515	10	7	of	of	ADP
ejpam-515	10	8	contingency	contingency	NOUN
ejpam-515	10	9	tables	table	NOUN
ejpam-515	10	10	of	of	ADP
ejpam-515	10	11	categorical	categorical	ADJ
ejpam-515	10	12	data	datum	NOUN
ejpam-515	10	13	.	.	PUNCT
ejpam-515	11	1	in	in	ADP
ejpam-515	11	2	addition	addition	NOUN
ejpam-515	11	3	to	to	ADP
ejpam-515	11	4	parameter	parameter	PROPN
ejpam-515	11	5	estimation	estimation	NOUN
ejpam-515	11	6	,	,	PUNCT
ejpam-515	11	7	another	another	DET
ejpam-515	11	8	important	important	ADJ
ejpam-515	11	9	inference	inference	NOUN
ejpam-515	11	10	task	task	NOUN
ejpam-515	11	11	in	in	ADP
ejpam-515	11	12	poisson	poisson	PROPN
ejpam-515	11	13	regression	regression	NOUN
ejpam-515	11	14	is	be	AUX
ejpam-515	11	15	searching	search	VERB
ejpam-515	11	16	for	for	ADP
ejpam-515	11	17	a	a	DET
ejpam-515	11	18	subset	subset	NOUN
ejpam-515	11	19	of	of	ADP
ejpam-515	11	20	available	available	ADJ
ejpam-515	11	21	explanatory	explanatory	ADJ
ejpam-515	11	22	variables	variable	NOUN
ejpam-515	11	23	that	that	PRON
ejpam-515	11	24	can	can	AUX
ejpam-515	11	25	best	well	ADV
ejpam-515	11	26	explain	explain	VERB
ejpam-515	11	27	or	or	CCONJ
ejpam-515	11	28	predict	predict	VERB
ejpam-515	11	29	the	the	DET
ejpam-515	11	30	response	response	NOUN
ejpam-515	11	31	.	.	PUNCT
ejpam-515	12	1	this	this	PRON
ejpam-515	12	2	amounts	amount	VERB
ejpam-515	12	3	to	to	ADP
ejpam-515	12	4	the	the	DET
ejpam-515	12	5	poisson	poisson	PROPN
ejpam-515	12	6	regression	regression	NOUN
ejpam-515	12	7	model	model	NOUN
ejpam-515	12	8	or	or	CCONJ
ejpam-515	12	9	variable	variable	ADJ
ejpam-515	12	10	selection	selection	NOUN
ejpam-515	12	11	.	.	PUNCT
ejpam-515	13	1	many	many	ADJ
ejpam-515	13	2	papers	paper	NOUN
ejpam-515	13	3	can	can	AUX
ejpam-515	13	4	be	be	AUX
ejpam-515	13	5	found	find	VERB
ejpam-515	13	6	in	in	ADP
ejpam-515	13	7	recent	recent	ADJ
ejpam-515	13	8	literature	literature	NOUN
ejpam-515	13	9	in	in	ADP
ejpam-515	13	10	the	the	DET
ejpam-515	13	11	area	area	NOUN
ejpam-515	13	12	of	of	ADP
ejpam-515	13	13	model	model	NOUN
ejpam-515	13	14	selection	selection	NOUN
ejpam-515	13	15	,	,	PUNCT
ejpam-515	13	16	which	which	PRON
ejpam-515	13	17	deal	deal	VERB
ejpam-515	13	18	with	with	ADP
ejpam-515	13	19	different	different	ADJ
ejpam-515	13	20	models	model	NOUN
ejpam-515	13	21	in	in	ADP
ejpam-515	13	22	different	different	ADJ
ejpam-515	13	23	ways	way	NOUN
ejpam-515	13	24	.	.	PUNCT
ejpam-515	14	1	we	we	PRON
ejpam-515	14	2	refer	refer	VERB
ejpam-515	14	3	to	to	ADP
ejpam-515	14	4	[	[	X
ejpam-515	14	5	4	4	NUM
ejpam-515	14	6	]	]	PUNCT
ejpam-515	14	7	and	and	CCONJ
ejpam-515	14	8	[	[	X
ejpam-515	14	9	14	14	NUM
ejpam-515	14	10	]	]	PUNCT
ejpam-515	14	11	and	and	CCONJ
ejpam-515	14	12	references	reference	NOUN
ejpam-515	14	13	therein	therein	ADV
ejpam-515	14	14	for	for	ADP
ejpam-515	14	15	the	the	DET
ejpam-515	14	16	detailed	detailed	ADJ
ejpam-515	14	17	survey	survey	NOUN
ejpam-515	14	18	.	.	PUNCT
ejpam-515	15	1	it	it	PRON
ejpam-515	15	2	appears	appear	VERB
ejpam-515	15	3	that	that	SCONJ
ejpam-515	15	4	,	,	PUNCT
ejpam-515	15	5	while	while	SCONJ
ejpam-515	15	6	it	it	PRON
ejpam-515	15	7	might	might	AUX
ejpam-515	15	8	be	be	AUX
ejpam-515	15	9	implied	imply	VERB
ejpam-515	15	10	according	accord	VERB
ejpam-515	15	11	to	to	ADP
ejpam-515	15	12	some	some	DET
ejpam-515	15	13	general	general	ADJ
ejpam-515	15	14	model	model	NOUN
ejpam-515	15	15	selection	selection	NOUN
ejpam-515	15	16	principle	principle	NOUN
ejpam-515	15	17	,	,	PUNCT
ejpam-515	15	18	the	the	DET
ejpam-515	15	19	poisson	poisson	PROPN
ejpam-515	15	20	regression	regression	NOUN
ejpam-515	15	21	model	model	NOUN
ejpam-515	15	22	selection	selection	NOUN
ejpam-515	15	23	method	method	NOUN
ejpam-515	15	24	by	by	ADP
ejpam-515	15	25	itself	itself	PRON
ejpam-515	15	26	has	have	AUX
ejpam-515	15	27	hardly	hardly	ADV
ejpam-515	15	28	been	be	AUX
ejpam-515	15	29	investigated	investigate	VERB
ejpam-515	15	30	in	in	ADP
ejpam-515	15	31	email	email	NOUN
ejpam-515	15	32	address	address	NOUN
ejpam-515	15	33	:	:	PUNCT
ejpam-515	16	1	g.qian�ms.unimelb.edu.au	g.qian�ms.unimelb.edu.au	PROPN
ejpam-515	16	2	http://www.ejpam.com	http://www.ejpam.com	X
ejpam-515	17	1	417	417	NUM
ejpam-515	17	2	c	c	X
ejpam-515	17	3	©	©	PROPN
ejpam-515	17	4	2010	2010	NUM
ejpam-515	17	5	ejpam	ejpam	NOUN
ejpam-515	17	6	all	all	DET
ejpam-515	17	7	rights	right	NOUN
ejpam-515	17	8	reserved	reserve	VERB
ejpam-515	17	9	.	.	PUNCT
ejpam-515	18	1	g.	g.	PROPN
ejpam-515	18	2	qian	qian	PROPN
ejpam-515	18	3	/	/	SYM
ejpam-515	18	4	eur	eur	PROPN
ejpam-515	18	5	.	.	PUNCT
ejpam-515	19	1	j.	j.	PROPN
ejpam-515	19	2	pure	pure	PROPN
ejpam-515	19	3	appl	appl	PROPN
ejpam-515	19	4	.	.	PROPN
ejpam-515	19	5	math	math	PROPN
ejpam-515	19	6	,	,	PUNCT
ejpam-515	19	7	3	3	NUM
ejpam-515	19	8	(	(	PUNCT
ejpam-515	19	9	2010	2010	NUM
ejpam-515	19	10	)	)	PUNCT
ejpam-515	19	11	,	,	PUNCT
ejpam-515	19	12	417	417	NUM
ejpam-515	19	13	-	-	SYM
ejpam-515	19	14	434	434	NUM
ejpam-515	19	15	418	418	NUM
ejpam-515	19	16	a	a	DET
ejpam-515	19	17	formal	formal	ADJ
ejpam-515	19	18	and	and	CCONJ
ejpam-515	19	19	rigorous	rigorous	ADJ
ejpam-515	19	20	way	way	NOUN
ejpam-515	19	21	.	.	PUNCT
ejpam-515	20	1	the	the	DET
ejpam-515	20	2	lack	lack	NOUN
ejpam-515	20	3	of	of	ADP
ejpam-515	20	4	a	a	DET
ejpam-515	20	5	formal	formal	ADJ
ejpam-515	20	6	theory	theory	NOUN
ejpam-515	20	7	for	for	ADP
ejpam-515	20	8	poisson	poisson	PROPN
ejpam-515	20	9	regression	regression	NOUN
ejpam-515	20	10	model	model	NOUN
ejpam-515	20	11	selection	selection	NOUN
ejpam-515	20	12	creates	create	VERB
ejpam-515	20	13	uncertainty	uncertainty	NOUN
ejpam-515	20	14	and	and	CCONJ
ejpam-515	20	15	inconvenience	inconvenience	NOUN
ejpam-515	20	16	for	for	ADP
ejpam-515	20	17	people	people	NOUN
ejpam-515	20	18	applying	apply	VERB
ejpam-515	20	19	the	the	DET
ejpam-515	20	20	method	method	NOUN
ejpam-515	20	21	for	for	ADP
ejpam-515	20	22	practice	practice	NOUN
ejpam-515	20	23	.	.	PUNCT
ejpam-515	21	1	this	this	PRON
ejpam-515	21	2	motivates	motivate	VERB
ejpam-515	21	3	the	the	DET
ejpam-515	21	4	writing	writing	NOUN
ejpam-515	21	5	of	of	ADP
ejpam-515	21	6	this	this	DET
ejpam-515	21	7	paper	paper	NOUN
ejpam-515	21	8	which	which	PRON
ejpam-515	21	9	focuses	focus	VERB
ejpam-515	21	10	on	on	ADP
ejpam-515	21	11	the	the	DET
ejpam-515	21	12	asymptotic	asymptotic	ADJ
ejpam-515	21	13	performance	performance	NOUN
ejpam-515	21	14	of	of	ADP
ejpam-515	21	15	a	a	DET
ejpam-515	21	16	class	class	NOUN
ejpam-515	21	17	of	of	ADP
ejpam-515	21	18	model	model	NOUN
ejpam-515	21	19	selection	selection	NOUN
ejpam-515	21	20	criteria	criterion	NOUN
ejpam-515	21	21	including	include	VERB
ejpam-515	21	22	aic	aic	PROPN
ejpam-515	21	23	,	,	PUNCT
ejpam-515	21	24	bic	bic	PROPN
ejpam-515	21	25	,	,	PUNCT
ejpam-515	21	26	mallows	mallow	NOUN
ejpam-515	21	27	cp	cp	INTJ
ejpam-515	21	28	and	and	CCONJ
ejpam-515	21	29	stochastic	stochastic	ADJ
ejpam-515	21	30	complexity	complexity	NOUN
ejpam-515	21	31	or	or	CCONJ
ejpam-515	21	32	minimum	minimum	ADJ
ejpam-515	21	33	description	description	NOUN
ejpam-515	21	34	length	length	NOUN
ejpam-515	21	35	for	for	ADP
ejpam-515	21	36	poisson	poisson	NOUN
ejpam-515	21	37	regression	regression	NOUN
ejpam-515	21	38	models	model	NOUN
ejpam-515	21	39	.	.	PUNCT
ejpam-515	22	1	a	a	DET
ejpam-515	22	2	byproduct	byproduct	NOUN
ejpam-515	22	3	of	of	ADP
ejpam-515	22	4	this	this	DET
ejpam-515	22	5	asymptotic	asymptotic	ADJ
ejpam-515	22	6	study	study	NOUN
ejpam-515	22	7	is	be	AUX
ejpam-515	22	8	the	the	DET
ejpam-515	22	9	establishment	establishment	NOUN
ejpam-515	22	10	of	of	ADP
ejpam-515	22	11	the	the	DET
ejpam-515	22	12	law	law	NOUN
ejpam-515	22	13	of	of	ADP
ejpam-515	22	14	iterated	iterated	ADJ
ejpam-515	22	15	logarithm	logarithm	NOUN
ejpam-515	22	16	for	for	ADP
ejpam-515	22	17	the	the	DET
ejpam-515	22	18	maximum	maximum	ADJ
ejpam-515	22	19	likelihood	likelihood	NOUN
ejpam-515	22	20	estimators	estimator	NOUN
ejpam-515	22	21	(	(	PUNCT
ejpam-515	22	22	mle	mle	PROPN
ejpam-515	22	23	)	)	PUNCT
ejpam-515	22	24	in	in	ADP
ejpam-515	22	25	the	the	DET
ejpam-515	22	26	poisson	poisson	NOUN
ejpam-515	22	27	regression	regression	NOUN
ejpam-515	22	28	models	model	NOUN
ejpam-515	22	29	.	.	PUNCT
ejpam-515	23	1	the	the	DET
ejpam-515	23	2	convergence	convergence	NOUN
ejpam-515	23	3	rate	rate	NOUN
ejpam-515	23	4	of	of	ADP
ejpam-515	23	5	the	the	DET
ejpam-515	23	6	mle	mle	NOUN
ejpam-515	23	7	provided	provide	VERB
ejpam-515	23	8	by	by	ADP
ejpam-515	23	9	the	the	DET
ejpam-515	23	10	law	law	NOUN
ejpam-515	23	11	of	of	ADP
ejpam-515	23	12	iterated	iterated	ADJ
ejpam-515	23	13	logarithm	logarithm	NOUN
ejpam-515	23	14	is	be	AUX
ejpam-515	23	15	very	very	ADV
ejpam-515	23	16	useful	useful	ADJ
ejpam-515	23	17	in	in	ADP
ejpam-515	23	18	deriving	derive	VERB
ejpam-515	23	19	precise	precise	ADJ
ejpam-515	23	20	approximations	approximation	NOUN
ejpam-515	23	21	for	for	ADP
ejpam-515	23	22	likelihood	likelihood	NOUN
ejpam-515	23	23	function	function	NOUN
ejpam-515	23	24	.	.	PUNCT
ejpam-515	24	1	in	in	ADP
ejpam-515	24	2	the	the	DET
ejpam-515	24	3	paper	paper	NOUN
ejpam-515	24	4	we	we	PRON
ejpam-515	24	5	first	first	ADV
ejpam-515	24	6	set	set	VERB
ejpam-515	24	7	up	up	ADP
ejpam-515	24	8	a	a	DET
ejpam-515	24	9	model	model	NOUN
ejpam-515	24	10	selection	selection	NOUN
ejpam-515	24	11	framework	framework	NOUN
ejpam-515	24	12	for	for	ADP
ejpam-515	24	13	poisson	poisson	NOUN
ejpam-515	24	14	regression	regression	NOUN
ejpam-515	24	15	models	model	NOUN
ejpam-515	24	16	and	and	CCONJ
ejpam-515	24	17	review	review	VERB
ejpam-515	24	18	several	several	ADJ
ejpam-515	24	19	general	general	ADJ
ejpam-515	24	20	model	model	NOUN
ejpam-515	24	21	selection	selection	NOUN
ejpam-515	24	22	criteria	criterion	NOUN
ejpam-515	24	23	such	such	ADJ
ejpam-515	24	24	as	as	ADP
ejpam-515	24	25	aic	aic	PROPN
ejpam-515	24	26	[	[	X
ejpam-515	24	27	1	1	NUM
ejpam-515	24	28	]	]	PUNCT
ejpam-515	24	29	,	,	PUNCT
ejpam-515	24	30	bic	bic	PROPN
ejpam-515	25	1	[	[	X
ejpam-515	25	2	18	18	NUM
ejpam-515	25	3	]	]	PUNCT
ejpam-515	25	4	and	and	CCONJ
ejpam-515	25	5	stochastic	stochastic	ADJ
ejpam-515	25	6	complexity	complexity	NOUN
ejpam-515	25	7	or	or	CCONJ
ejpam-515	25	8	minimum	minimum	ADJ
ejpam-515	25	9	description	description	NOUN
ejpam-515	25	10	length	length	NOUN
ejpam-515	26	1	[	[	X
ejpam-515	26	2	16	16	NUM
ejpam-515	26	3	]	]	PUNCT
ejpam-515	26	4	in	in	ADP
ejpam-515	26	5	section	section	NOUN
ejpam-515	26	6	2	2	NUM
ejpam-515	26	7	.	.	PUNCT
ejpam-515	27	1	in	in	ADP
ejpam-515	27	2	section	section	NOUN
ejpam-515	27	3	3	3	NUM
ejpam-515	27	4	we	we	PRON
ejpam-515	27	5	present	present	VERB
ejpam-515	27	6	the	the	DET
ejpam-515	27	7	main	main	ADJ
ejpam-515	27	8	results	result	NOUN
ejpam-515	27	9	and	and	CCONJ
ejpam-515	27	10	the	the	DET
ejpam-515	27	11	conditions	condition	NOUN
ejpam-515	27	12	for	for	ADP
ejpam-515	27	13	ensuring	ensure	VERB
ejpam-515	27	14	these	these	DET
ejpam-515	27	15	results	result	NOUN
ejpam-515	27	16	.	.	PUNCT
ejpam-515	28	1	we	we	PRON
ejpam-515	28	2	have	have	AUX
ejpam-515	28	3	shown	show	VERB
ejpam-515	28	4	that	that	SCONJ
ejpam-515	28	5	when	when	SCONJ
ejpam-515	28	6	the	the	DET
ejpam-515	28	7	employed	employ	VERB
ejpam-515	28	8	model	model	NOUN
ejpam-515	28	9	is	be	AUX
ejpam-515	28	10	a	a	DET
ejpam-515	28	11	correct	correct	ADJ
ejpam-515	28	12	one	one	NOUN
ejpam-515	28	13	,	,	PUNCT
ejpam-515	28	14	the	the	DET
ejpam-515	28	15	mle	mle	NOUN
ejpam-515	28	16	β̂	β̂	ADP
ejpam-515	28	17	converges	converge	NOUN
ejpam-515	28	18	almost	almost	ADV
ejpam-515	28	19	surely	surely	ADV
ejpam-515	28	20	to	to	ADP
ejpam-515	28	21	the	the	DET
ejpam-515	28	22	true	true	ADJ
ejpam-515	28	23	parameter	parameter	NOUN
ejpam-515	28	24	value	value	NOUN
ejpam-515	28	25	β0	β0	NOUN
ejpam-515	28	26	with	with	ADP
ejpam-515	28	27	a	a	DET
ejpam-515	28	28	rate	rate	NOUN
ejpam-515	28	29	not	not	PART
ejpam-515	28	30	slower	slow	ADJ
ejpam-515	28	31	than	than	ADP
ejpam-515	28	32	o	o	NOUN
ejpam-515	28	33	(	(	PUNCT
ejpam-515	28	34	p	p	PROPN
ejpam-515	28	35	n−1	n−1	PROPN
ejpam-515	28	36	log	log	NOUN
ejpam-515	28	37	log	log	NOUN
ejpam-515	28	38	n	n	CCONJ
ejpam-515	28	39	)	)	PUNCT
ejpam-515	28	40	.	.	PUNCT
ejpam-515	29	1	we	we	PRON
ejpam-515	29	2	have	have	AUX
ejpam-515	29	3	also	also	ADV
ejpam-515	29	4	shown	show	VERB
ejpam-515	29	5	that	that	SCONJ
ejpam-515	29	6	,	,	PUNCT
ejpam-515	29	7	for	for	ADP
ejpam-515	29	8	a	a	DET
ejpam-515	29	9	model	model	NOUN
ejpam-515	29	10	selection	selection	NOUN
ejpam-515	29	11	criterion	criterion	NOUN
ejpam-515	29	12	consisting	consist	VERB
ejpam-515	29	13	of	of	ADP
ejpam-515	29	14	the	the	DET
ejpam-515	29	15	minus	minus	NOUN
ejpam-515	29	16	log	log	NOUN
ejpam-515	29	17	-	-	PUNCT
ejpam-515	29	18	likelihood	likelihood	NOUN
ejpam-515	29	19	and	and	CCONJ
ejpam-515	29	20	a	a	DET
ejpam-515	29	21	penalty	penalty	NOUN
ejpam-515	29	22	term	term	NOUN
ejpam-515	29	23	,	,	PUNCT
ejpam-515	29	24	it	it	PRON
ejpam-515	29	25	will	will	AUX
ejpam-515	29	26	select	select	VERB
ejpam-515	29	27	the	the	DET
ejpam-515	29	28	simplest	simple	ADJ
ejpam-515	29	29	correct	correct	ADJ
ejpam-515	29	30	model	model	NOUN
ejpam-515	29	31	almost	almost	ADV
ejpam-515	29	32	surely	surely	ADV
ejpam-515	29	33	if	if	SCONJ
ejpam-515	29	34	the	the	DET
ejpam-515	29	35	penalty	penalty	NOUN
ejpam-515	29	36	term	term	NOUN
ejpam-515	29	37	is	be	AUX
ejpam-515	29	38	an	an	DET
ejpam-515	29	39	increasing	increase	VERB
ejpam-515	29	40	function	function	NOUN
ejpam-515	29	41	of	of	ADP
ejpam-515	29	42	the	the	DET
ejpam-515	29	43	model	model	NOUN
ejpam-515	29	44	dimension	dimension	NOUN
ejpam-515	29	45	and	and	CCONJ
ejpam-515	29	46	is	be	AUX
ejpam-515	29	47	of	of	ADP
ejpam-515	29	48	an	an	DET
ejpam-515	29	49	order	order	NOUN
ejpam-515	29	50	higher	high	ADJ
ejpam-515	29	51	than	than	ADP
ejpam-515	29	52	o(log	o(log	PROPN
ejpam-515	29	53	log	log	PROPN
ejpam-515	29	54	n	n	CCONJ
ejpam-515	29	55	)	)	PUNCT
ejpam-515	29	56	and	and	CCONJ
ejpam-515	29	57	lower	low	ADJ
ejpam-515	29	58	than	than	ADP
ejpam-515	29	59	o(n	o(n	NUM
ejpam-515	29	60	)	)	PUNCT
ejpam-515	29	61	.	.	PUNCT
ejpam-515	30	1	the	the	DET
ejpam-515	30	2	detailed	detailed	ADJ
ejpam-515	30	3	proof	proof	NOUN
ejpam-515	30	4	of	of	ADP
ejpam-515	30	5	these	these	DET
ejpam-515	30	6	results	result	NOUN
ejpam-515	30	7	are	be	AUX
ejpam-515	30	8	given	give	VERB
ejpam-515	30	9	in	in	ADP
ejpam-515	30	10	section	section	NOUN
ejpam-515	30	11	4	4	NUM
ejpam-515	30	12	and	and	CCONJ
ejpam-515	30	13	the	the	DET
ejpam-515	30	14	appendix	appendix	NOUN
ejpam-515	30	15	.	.	PUNCT
ejpam-515	31	1	the	the	DET
ejpam-515	31	2	paper	paper	NOUN
ejpam-515	31	3	is	be	AUX
ejpam-515	31	4	concluded	conclude	VERB
ejpam-515	31	5	with	with	ADP
ejpam-515	31	6	a	a	DET
ejpam-515	31	7	discussion	discussion	NOUN
ejpam-515	31	8	given	give	VERB
ejpam-515	31	9	in	in	ADP
ejpam-515	31	10	section	section	NOUN
ejpam-515	31	11	5	5	NUM
ejpam-515	31	12	.	.	SYM
ejpam-515	31	13	2	2	NUM
ejpam-515	31	14	.	.	PUNCT
ejpam-515	31	15	model	model	NOUN
ejpam-515	31	16	selection	selection	NOUN
ejpam-515	31	17	in	in	ADP
ejpam-515	31	18	poisson	poisson	PROPN
ejpam-515	31	19	regression	regression	NOUN
ejpam-515	31	20	model	model	NOUN
ejpam-515	31	21	the	the	DET
ejpam-515	31	22	problem	problem	NOUN
ejpam-515	31	23	to	to	ADP
ejpam-515	31	24	our	our	PRON
ejpam-515	31	25	interest	interest	NOUN
ejpam-515	31	26	is	be	AUX
ejpam-515	31	27	whether	whether	SCONJ
ejpam-515	31	28	any	any	DET
ejpam-515	31	29	component	component	NOUN
ejpam-515	31	30	of	of	ADP
ejpam-515	31	31	a	a	DET
ejpam-515	31	32	given	give	VERB
ejpam-515	31	33	explanatory	explanatory	ADJ
ejpam-515	31	34	vector	vector	NOUN
ejpam-515	31	35	x	x	PUNCT
ejpam-515	31	36	=	=	SYM
ejpam-515	31	37	(	(	PUNCT
ejpam-515	31	38	x1	x1	PROPN
ejpam-515	31	39	,	,	PUNCT
ejpam-515	31	40	·	·	PUNCT
ejpam-515	31	41	·	·	PUNCT
ejpam-515	31	42	·	·	PUNCT
ejpam-515	31	43	,	,	PUNCT
ejpam-515	31	44	xp	xp	X
ejpam-515	31	45	)	)	PUNCT
ejpam-515	31	46	t	t	PROPN
ejpam-515	31	47	has	have	VERB
ejpam-515	31	48	any	any	DET
ejpam-515	31	49	effect	effect	NOUN
ejpam-515	31	50	on	on	ADP
ejpam-515	31	51	a	a	DET
ejpam-515	31	52	response	response	NOUN
ejpam-515	31	53	variable	variable	ADJ
ejpam-515	31	54	y	y	PROPN
ejpam-515	31	55	.	.	PUNCT
ejpam-515	32	1	when	when	SCONJ
ejpam-515	32	2	y	y	PROPN
ejpam-515	32	3	is	be	AUX
ejpam-515	32	4	a	a	DET
ejpam-515	32	5	count	count	NOUN
ejpam-515	32	6	variable	variable	NOUN
ejpam-515	32	7	,	,	PUNCT
ejpam-515	32	8	it	it	PRON
ejpam-515	32	9	is	be	AUX
ejpam-515	32	10	often	often	ADV
ejpam-515	32	11	sensible	sensible	ADJ
ejpam-515	32	12	to	to	PART
ejpam-515	32	13	assume	assume	VERB
ejpam-515	32	14	a	a	DET
ejpam-515	32	15	poisson	poisson	NOUN
ejpam-515	32	16	distribution	distribution	NOUN
ejpam-515	32	17	for	for	ADP
ejpam-515	32	18	y	y	PROPN
ejpam-515	32	19	which	which	PRON
ejpam-515	32	20	has	have	VERB
ejpam-515	32	21	a	a	DET
ejpam-515	32	22	probability	probability	NOUN
ejpam-515	32	23	function	function	NOUN
ejpam-515	32	24	p(y	p(y	PROPN
ejpam-515	32	25	=	=	SYM
ejpam-515	32	26	y	y	NOUN
ejpam-515	32	27	)	)	PUNCT
ejpam-515	32	28	=	=	SYM
ejpam-515	32	29	(	(	PUNCT
ejpam-515	33	1	y!)−1µy	y!)−1µy	NOUN
ejpam-515	33	2	e−µ	e−µ	NOUN
ejpam-515	33	3	(	(	PUNCT
ejpam-515	33	4	y	y	NOUN
ejpam-515	33	5	=	=	SYM
ejpam-515	33	6	0,1,2	0,1,2	NOUN
ejpam-515	33	7	,	,	PUNCT
ejpam-515	33	8	·	·	PUNCT
ejpam-515	33	9	·	·	PUNCT
ejpam-515	33	10	·	·	PUNCT
ejpam-515	33	11	)	)	PUNCT
ejpam-515	33	12	.	.	PUNCT
ejpam-515	34	1	then	then	ADV
ejpam-515	34	2	the	the	DET
ejpam-515	34	3	problem	problem	NOUN
ejpam-515	34	4	can	can	AUX
ejpam-515	34	5	be	be	AUX
ejpam-515	34	6	studied	study	VERB
ejpam-515	34	7	in	in	ADP
ejpam-515	34	8	the	the	DET
ejpam-515	34	9	framework	framework	NOUN
ejpam-515	34	10	of	of	ADP
ejpam-515	34	11	a	a	DET
ejpam-515	34	12	log	log	NOUN
ejpam-515	34	13	-	-	PUNCT
ejpam-515	34	14	linear	linear	NOUN
ejpam-515	34	15	regression	regression	NOUN
ejpam-515	34	16	model	model	NOUN
ejpam-515	34	17	which	which	PRON
ejpam-515	34	18	assumes	assume	VERB
ejpam-515	34	19	a	a	DET
ejpam-515	34	20	linear	linear	ADJ
ejpam-515	34	21	predictor	predictor	NOUN
ejpam-515	34	22	η	η	PROPN
ejpam-515	34	23	=	=	PROPN
ejpam-515	34	24	xtβ	xtβ	PROPN
ejpam-515	34	25	for	for	ADP
ejpam-515	34	26	logarithm	logarithm	NOUN
ejpam-515	34	27	of	of	ADP
ejpam-515	34	28	the	the	DET
ejpam-515	34	29	mean	mean	NOUN
ejpam-515	34	30	of	of	ADP
ejpam-515	34	31	y	y	PROPN
ejpam-515	34	32	,	,	PUNCT
ejpam-515	34	33	i.e.	i.e.	X
ejpam-515	34	34	,	,	PUNCT
ejpam-515	34	35	logµ	logµ	NOUN
ejpam-515	34	36	=	=	SYM
ejpam-515	34	37	η	η	PROPN
ejpam-515	34	38	=	=	PROPN
ejpam-515	34	39	xtβ	xtβ	PROPN
ejpam-515	34	40	,	,	PUNCT
ejpam-515	34	41	where	where	SCONJ
ejpam-515	34	42	β	β	X
ejpam-515	34	43	=	=	SYM
ejpam-515	34	44	(	(	PUNCT
ejpam-515	34	45	β1	β1	PROPN
ejpam-515	34	46	,	,	PUNCT
ejpam-515	34	47	·	·	PUNCT
ejpam-515	34	48	·	·	PUNCT
ejpam-515	34	49	·	·	PUNCT
ejpam-515	34	50	,	,	PUNCT
ejpam-515	34	51	βp	βp	X
ejpam-515	34	52	)	)	PUNCT
ejpam-515	34	53	t	t	PROPN
ejpam-515	34	54	is	be	AUX
ejpam-515	34	55	the	the	DET
ejpam-515	34	56	unknown	unknown	ADJ
ejpam-515	34	57	parameter	parameter	NOUN
ejpam-515	34	58	vector	vector	NOUN
ejpam-515	34	59	of	of	ADP
ejpam-515	34	60	interest	interest	NOUN
ejpam-515	34	61	.	.	PUNCT
ejpam-515	35	1	now	now	ADV
ejpam-515	35	2	let	let	VERB
ejpam-515	35	3	yn	yn	PRON
ejpam-515	35	4	=	=	SYM
ejpam-515	35	5	(	(	PUNCT
ejpam-515	35	6	y1	y1	PROPN
ejpam-515	35	7	,	,	PUNCT
ejpam-515	35	8	·	·	PUNCT
ejpam-515	35	9	·	·	PUNCT
ejpam-515	35	10	·	·	PUNCT
ejpam-515	35	11	,	,	PUNCT
ejpam-515	35	12	yn	yn	PROPN
ejpam-515	35	13	)	)	PUNCT
ejpam-515	35	14	t	t	PROPN
ejpam-515	35	15	be	be	AUX
ejpam-515	35	16	the	the	DET
ejpam-515	35	17	n	n	CCONJ
ejpam-515	35	18	independent	independent	ADJ
ejpam-515	35	19	observations	observation	NOUN
ejpam-515	35	20	from	from	ADP
ejpam-515	35	21	y	y	PROPN
ejpam-515	35	22	,	,	PUNCT
ejpam-515	35	23	with	with	ADP
ejpam-515	35	24	the	the	DET
ejpam-515	35	25	corresponding	corresponding	ADJ
ejpam-515	35	26	explanatory	explanatory	ADJ
ejpam-515	35	27	vectors	vector	NOUN
ejpam-515	35	28	being	be	AUX
ejpam-515	35	29	x1	x1	ADJ
ejpam-515	35	30	,	,	PUNCT
ejpam-515	35	31	·	·	PUNCT
ejpam-515	35	32	·	·	PUNCT
ejpam-515	35	33	·	·	PUNCT
ejpam-515	35	34	,	,	PUNCT
ejpam-515	35	35	xn	xn	PROPN
ejpam-515	35	36	.	.	PROPN
ejpam-515	35	37	denote	denote	VERB
ejpam-515	35	38	xn	xn	PUNCT
ejpam-515	36	1	=	=	PUNCT
ejpam-515	36	2	(	(	PUNCT
ejpam-515	36	3	x1	x1	PROPN
ejpam-515	36	4	,	,	PUNCT
ejpam-515	36	5	·	·	PUNCT
ejpam-515	36	6	·	·	PUNCT
ejpam-515	36	7	·	·	PUNCT
ejpam-515	36	8	,	,	PUNCT
ejpam-515	36	9	xn	xn	X
ejpam-515	36	10	)	)	PUNCT
ejpam-515	36	11	t	t	NOUN
ejpam-515	36	12	as	as	ADP
ejpam-515	36	13	the	the	DET
ejpam-515	36	14	design	design	NOUN
ejpam-515	36	15	matrix	matrix	NOUN
ejpam-515	36	16	.	.	PUNCT
ejpam-515	37	1	then	then	ADV
ejpam-515	37	2	under	under	ADP
ejpam-515	37	3	the	the	DET
ejpam-515	37	4	log	log	NOUN
ejpam-515	37	5	-	-	PUNCT
ejpam-515	37	6	linear	linear	NOUN
ejpam-515	37	7	regression	regression	NOUN
ejpam-515	37	8	model	model	NOUN
ejpam-515	37	9	considered	consider	VERB
ejpam-515	37	10	,	,	PUNCT
ejpam-515	37	11	the	the	DET
ejpam-515	37	12	distribution	distribution	NOUN
ejpam-515	37	13	for	for	ADP
ejpam-515	37	14	yi	yi	PROPN
ejpam-515	37	15	is	be	AUX
ejpam-515	37	16	poisson(µi	poisson(µi	ADJ
ejpam-515	37	17	)	)	PUNCT
ejpam-515	37	18	with	with	ADP
ejpam-515	37	19	µi	µi	PROPN
ejpam-515	37	20	=	=	PUNCT
ejpam-515	37	21	eηi	eηi	PROPN
ejpam-515	38	1	=	=	SYM
ejpam-515	38	2	ext	ext	NOUN
ejpam-515	38	3	i	i	PRON
ejpam-515	38	4	β	β	X
ejpam-515	38	5	;	;	PUNCT
ejpam-515	38	6	and	and	CCONJ
ejpam-515	38	7	the	the	DET
ejpam-515	38	8	log	log	NOUN
ejpam-515	38	9	-	-	PUNCT
ejpam-515	38	10	likelihood	likelihood	NOUN
ejpam-515	38	11	function	function	NOUN
ejpam-515	38	12	for	for	ADP
ejpam-515	38	13	the	the	DET
ejpam-515	38	14	parameter	parameter	NOUN
ejpam-515	38	15	β	β	PROPN
ejpam-515	38	16	is	be	AUX
ejpam-515	38	17	ℓ(β	ℓ(β	PROPN
ejpam-515	38	18	|yn	|yn	NUM
ejpam-515	38	19	,	,	PUNCT
ejpam-515	38	20	xn	xn	PROPN
ejpam-515	38	21	)	)	PUNCT
ejpam-515	38	22	=	=	SYM
ejpam-515	39	1	n	n	PROPN
ejpam-515	39	2	∑	∑	ADV
ejpam-515	39	3	i=1	i=1	PROPN
ejpam-515	39	4	{	{	PUNCT
ejpam-515	39	5	−	−	PUNCT
ejpam-515	39	6	log	log	NOUN
ejpam-515	39	7	yi!+	yi!+	PROPN
ejpam-515	39	8	yi	yi	NOUN
ejpam-515	39	9	logµi	logµi	VERB
ejpam-515	39	10	−µi}=	−µi}=	PROPN
ejpam-515	39	11	−	−	NOUN
ejpam-515	40	1	n	n	ADP
ejpam-515	40	2	∑	∑	ADP
ejpam-515	40	3	i=1	i=1	PROPN
ejpam-515	40	4	log	log	VERB
ejpam-515	40	5	yi!+	yi!+	PROPN
ejpam-515	40	6	n	n	CCONJ
ejpam-515	40	7	∑	∑	PROPN
ejpam-515	40	8	i=1	i=1	PROPN
ejpam-515	40	9	{	{	PUNCT
ejpam-515	40	10	yix	yix	PROPN
ejpam-515	40	11	t	t	PROPN
ejpam-515	40	12	iβ	iβ	ADP
ejpam-515	40	13	−	−	PROPN
ejpam-515	40	14	ext	ext	NOUN
ejpam-515	40	15	i	i	PRON
ejpam-515	40	16	β	β	X
ejpam-515	40	17	}	}	PUNCT
ejpam-515	40	18	.	.	PUNCT
ejpam-515	41	1	(	(	PUNCT
ejpam-515	41	2	1	1	X
ejpam-515	41	3	)	)	PUNCT
ejpam-515	41	4	the	the	DET
ejpam-515	41	5	maximum	maximum	ADJ
ejpam-515	41	6	likelihood	likelihood	NOUN
ejpam-515	41	7	estimator(mle	estimator(mle	NOUN
ejpam-515	41	8	)	)	PUNCT
ejpam-515	41	9	β̂	β̂	ADP
ejpam-515	41	10	is	be	AUX
ejpam-515	41	11	defined	define	VERB
ejpam-515	41	12	to	to	PART
ejpam-515	41	13	be	be	AUX
ejpam-515	41	14	β̂	β̂	ADP
ejpam-515	41	15	=	=	PUNCT
ejpam-515	41	16	arg	arg	NOUN
ejpam-515	41	17	max	max	PROPN
ejpam-515	41	18	β	β	X
ejpam-515	41	19	ℓ(β	ℓ(β	PROPN
ejpam-515	41	20	|yn	|yn	NUM
ejpam-515	41	21	,	,	PUNCT
ejpam-515	41	22	xn	xn	PROPN
ejpam-515	41	23	)	)	PUNCT
ejpam-515	42	1	=	=	SYM
ejpam-515	42	2	arg	arg	NOUN
ejpam-515	42	3	min	min	PROPN
ejpam-515	42	4	β	β	PROPN
ejpam-515	42	5	n	n	PROPN
ejpam-515	42	6	∑	∑	PROPN
ejpam-515	42	7	i=1	i=1	PROPN
ejpam-515	42	8	{	{	PUNCT
ejpam-515	42	9	ext	ext	VERB
ejpam-515	42	10	i	i	NOUN
ejpam-515	42	11	β	β	PROPN
ejpam-515	42	12	−	−	PROPN
ejpam-515	43	1	yix	yix	PROPN
ejpam-515	43	2	t	t	PROPN
ejpam-515	43	3	iβ	iβ	PROPN
ejpam-515	43	4	}	}	PUNCT
ejpam-515	43	5	g.	g.	PROPN
ejpam-515	43	6	qian	qian	PROPN
ejpam-515	43	7	/	/	SYM
ejpam-515	43	8	eur	eur	PROPN
ejpam-515	43	9	.	.	PUNCT
ejpam-515	44	1	j.	j.	PROPN
ejpam-515	44	2	pure	pure	PROPN
ejpam-515	44	3	appl	appl	PROPN
ejpam-515	44	4	.	.	PROPN
ejpam-515	44	5	math	math	PROPN
ejpam-515	44	6	,	,	PUNCT
ejpam-515	44	7	3	3	NUM
ejpam-515	44	8	(	(	PUNCT
ejpam-515	44	9	2010	2010	NUM
ejpam-515	44	10	)	)	PUNCT
ejpam-515	44	11	,	,	PUNCT
ejpam-515	44	12	417	417	NUM
ejpam-515	44	13	-	-	SYM
ejpam-515	44	14	434	434	NUM
ejpam-515	44	15	419	419	NUM
ejpam-515	44	16	which	which	PRON
ejpam-515	44	17	can	can	AUX
ejpam-515	44	18	be	be	AUX
ejpam-515	44	19	solved	solve	VERB
ejpam-515	44	20	from	from	ADP
ejpam-515	44	21	∂	∂	NUM
ejpam-515	44	22	ℓ	ℓ	NOUN
ejpam-515	44	23	∂	∂	NOUN
ejpam-515	44	24	β	β	X
ejpam-515	44	25	=	=	SYM
ejpam-515	44	26	n	n	PROPN
ejpam-515	44	27	∑	∑	PROPN
ejpam-515	44	28	i=1	i=1	PROPN
ejpam-515	44	29	(	(	PUNCT
ejpam-515	44	30	yi	yi	NOUN
ejpam-515	44	31	−µi)xi	−µi)xi	NOUN
ejpam-515	44	32	=	=	PUNCT
ejpam-515	44	33	0	0	X
ejpam-515	44	34	.	.	PUNCT
ejpam-515	45	1	in	in	ADP
ejpam-515	45	2	this	this	DET
ejpam-515	45	3	paper	paper	NOUN
ejpam-515	45	4	we	we	PRON
ejpam-515	45	5	will	will	AUX
ejpam-515	45	6	show	show	VERB
ejpam-515	45	7	that	that	SCONJ
ejpam-515	45	8	the	the	DET
ejpam-515	45	9	estimation	estimation	NOUN
ejpam-515	45	10	error	error	NOUN
ejpam-515	45	11	||β̂	||β̂	X
ejpam-515	45	12	−	−	NOUN
ejpam-515	46	1	β0||	β0||	PROPN
ejpam-515	46	2	=	=	SYM
ejpam-515	47	1	o	o	PROPN
ejpam-515	47	2	(	(	PUNCT
ejpam-515	47	3	p	p	PROPN
ejpam-515	47	4	n−1	n−1	PROPN
ejpam-515	47	5	log	log	NOUN
ejpam-515	47	6	log	log	NOUN
ejpam-515	47	7	n	n	CCONJ
ejpam-515	47	8	)	)	PUNCT
ejpam-515	47	9	almost	almost	ADV
ejpam-515	47	10	surely	surely	ADV
ejpam-515	47	11	under	under	ADP
ejpam-515	47	12	some	some	DET
ejpam-515	47	13	general	general	ADJ
ejpam-515	47	14	conditions	condition	NOUN
ejpam-515	47	15	.	.	PUNCT
ejpam-515	48	1	here	here	ADV
ejpam-515	48	2	β0	β0	PROPN
ejpam-515	48	3	is	be	AUX
ejpam-515	48	4	the	the	DET
ejpam-515	48	5	true	true	ADJ
ejpam-515	48	6	value	value	NOUN
ejpam-515	48	7	of	of	ADP
ejpam-515	48	8	β	β	PROPN
ejpam-515	48	9	and	and	CCONJ
ejpam-515	48	10	||	||	NUM
ejpam-515	48	11	·	·	PUNCT
ejpam-515	49	1	||	||	NUM
ejpam-515	49	2	is	be	AUX
ejpam-515	49	3	the	the	DET
ejpam-515	49	4	euclidean	euclidean	ADJ
ejpam-515	49	5	norm	norm	NOUN
ejpam-515	49	6	.	.	PUNCT
ejpam-515	50	1	the	the	DET
ejpam-515	50	2	fisher	fisher	PROPN
ejpam-515	50	3	information	information	NOUN
ejpam-515	50	4	for	for	ADP
ejpam-515	50	5	the	the	DET
ejpam-515	50	6	parameter	parameter	NOUN
ejpam-515	50	7	β	β	PROPN
ejpam-515	50	8	can	can	AUX
ejpam-515	50	9	be	be	AUX
ejpam-515	50	10	found	find	VERB
ejpam-515	50	11	to	to	PART
ejpam-515	50	12	be	be	AUX
ejpam-515	50	13	in(β	in(β	PUNCT
ejpam-515	50	14	)	)	PUNCT
ejpam-515	51	1	=	=	PUNCT
ejpam-515	51	2	−e	−e	NOUN
ejpam-515	51	3	∂	∂	NUM
ejpam-515	51	4	2ℓ	2ℓ	NUM
ejpam-515	51	5	∂	∂	NUM
ejpam-515	51	6	β∂	β∂	PROPN
ejpam-515	51	7	β	β	X
ejpam-515	51	8	t	t	NOUN
ejpam-515	51	9	=	=	SYM
ejpam-515	51	10	−	−	PROPN
ejpam-515	51	11	∂	∂	NUM
ejpam-515	51	12	2ℓ	2ℓ	PROPN
ejpam-515	51	13	∂	∂	NUM
ejpam-515	51	14	β∂	β∂	PROPN
ejpam-515	51	15	β	β	X
ejpam-515	51	16	t	t	NOUN
ejpam-515	51	17	=	=	PUNCT
ejpam-515	51	18	x	x	SYM
ejpam-515	51	19	t	t	PROPN
ejpam-515	51	20	nu	nu	PROPN
ejpam-515	51	21	nxn	nxn	PROPN
ejpam-515	51	22	,	,	PUNCT
ejpam-515	51	23	(	(	PUNCT
ejpam-515	51	24	2	2	X
ejpam-515	51	25	)	)	PUNCT
ejpam-515	51	26	where	where	SCONJ
ejpam-515	51	27	un	un	PROPN
ejpam-515	51	28	=	=	PROPN
ejpam-515	51	29	diag{µ1	diag{µ1	PROPN
ejpam-515	51	30	,	,	PUNCT
ejpam-515	51	31	·	·	PUNCT
ejpam-515	51	32	·	·	PUNCT
ejpam-515	51	33	·	·	PUNCT
ejpam-515	51	34	,	,	PUNCT
ejpam-515	51	35	µn	µn	PROPN
ejpam-515	51	36	}	}	PUNCT
ejpam-515	51	37	.	.	PUNCT
ejpam-515	52	1	with	with	ADP
ejpam-515	52	2	the	the	DET
ejpam-515	52	3	poisson	poisson	PROPN
ejpam-515	52	4	regression	regression	NOUN
ejpam-515	52	5	model	model	NOUN
ejpam-515	52	6	logµ	logµ	PROPN
ejpam-515	52	7	=	=	SYM
ejpam-515	52	8	xtβ	xtβ	PROPN
ejpam-515	52	9	,	,	PUNCT
ejpam-515	52	10	the	the	DET
ejpam-515	52	11	effect	effect	NOUN
ejpam-515	52	12	of	of	ADP
ejpam-515	52	13	each	each	DET
ejpam-515	52	14	x	x	NOUN
ejpam-515	52	15	variable	variable	NOUN
ejpam-515	52	16	on	on	ADP
ejpam-515	52	17	y	y	PROPN
ejpam-515	52	18	can	can	AUX
ejpam-515	52	19	be	be	AUX
ejpam-515	52	20	measured	measure	VERB
ejpam-515	52	21	by	by	ADP
ejpam-515	52	22	the	the	DET
ejpam-515	52	23	value	value	NOUN
ejpam-515	52	24	of	of	ADP
ejpam-515	52	25	the	the	DET
ejpam-515	52	26	corresponding	corresponding	ADJ
ejpam-515	52	27	β	β	PROPN
ejpam-515	52	28	component	component	NOUN
ejpam-515	52	29	.	.	PUNCT
ejpam-515	53	1	thus	thus	ADV
ejpam-515	53	2	if	if	SCONJ
ejpam-515	53	3	any	any	PRON
ejpam-515	53	4	of	of	ADP
ejpam-515	53	5	the	the	DET
ejpam-515	53	6	β	β	PROPN
ejpam-515	53	7	components	component	NOUN
ejpam-515	53	8	equals	equal	VERB
ejpam-515	53	9	0	0	PUNCT
ejpam-515	53	10	or	or	CCONJ
ejpam-515	53	11	is	be	AUX
ejpam-515	53	12	close	close	ADJ
ejpam-515	53	13	to	to	ADP
ejpam-515	53	14	0	0	NUM
ejpam-515	53	15	,	,	PUNCT
ejpam-515	53	16	there	there	PRON
ejpam-515	53	17	would	would	AUX
ejpam-515	53	18	be	be	AUX
ejpam-515	53	19	no	no	DET
ejpam-515	53	20	necessity	necessity	NOUN
ejpam-515	53	21	to	to	PART
ejpam-515	53	22	include	include	VERB
ejpam-515	53	23	the	the	DET
ejpam-515	53	24	corresponding	correspond	VERB
ejpam-515	53	25	x	x	SYM
ejpam-515	53	26	components	component	NOUN
ejpam-515	53	27	into	into	ADP
ejpam-515	53	28	the	the	DET
ejpam-515	53	29	model	model	NOUN
ejpam-515	53	30	.	.	PUNCT
ejpam-515	54	1	since	since	SCONJ
ejpam-515	54	2	the	the	DET
ejpam-515	54	3	values	value	NOUN
ejpam-515	54	4	of	of	ADP
ejpam-515	54	5	β	β	X
ejpam-515	54	6	can	can	AUX
ejpam-515	54	7	only	only	ADV
ejpam-515	54	8	be	be	AUX
ejpam-515	54	9	estimated	estimate	VERB
ejpam-515	54	10	,	,	PUNCT
ejpam-515	54	11	one	one	PRON
ejpam-515	54	12	needs	need	VERB
ejpam-515	54	13	a	a	DET
ejpam-515	54	14	statistical	statistical	ADJ
ejpam-515	54	15	model	model	NOUN
ejpam-515	54	16	selection	selection	NOUN
ejpam-515	54	17	criterion	criterion	NOUN
ejpam-515	54	18	to	to	PART
ejpam-515	54	19	determine	determine	VERB
ejpam-515	54	20	which	which	PRON
ejpam-515	54	21	x	x	SYM
ejpam-515	54	22	components	component	NOUN
ejpam-515	54	23	have	have	VERB
ejpam-515	54	24	significant	significant	ADJ
ejpam-515	54	25	effects	effect	NOUN
ejpam-515	54	26	on	on	ADP
ejpam-515	54	27	y	y	PRON
ejpam-515	54	28	thus	thus	ADV
ejpam-515	54	29	should	should	AUX
ejpam-515	54	30	be	be	AUX
ejpam-515	54	31	included	include	VERB
ejpam-515	54	32	in	in	ADP
ejpam-515	54	33	the	the	DET
ejpam-515	54	34	model	model	NOUN
ejpam-515	54	35	.	.	PUNCT
ejpam-515	55	1	the	the	DET
ejpam-515	55	2	maximum	maximum	ADJ
ejpam-515	55	3	likelihood	likelihood	NOUN
ejpam-515	55	4	principle	principle	NOUN
ejpam-515	55	5	can	can	AUX
ejpam-515	55	6	not	not	PART
ejpam-515	55	7	serve	serve	VERB
ejpam-515	55	8	as	as	ADP
ejpam-515	55	9	a	a	DET
ejpam-515	55	10	model	model	NOUN
ejpam-515	55	11	selection	selection	NOUN
ejpam-515	55	12	criterion	criterion	NOUN
ejpam-515	55	13	because	because	SCONJ
ejpam-515	55	14	the	the	DET
ejpam-515	55	15	maximum	maximum	ADJ
ejpam-515	55	16	likelihood	likelihood	NOUN
ejpam-515	55	17	for	for	ADP
ejpam-515	55	18	the	the	DET
ejpam-515	55	19	full	full	ADJ
ejpam-515	55	20	model	model	NOUN
ejpam-515	55	21	including	include	VERB
ejpam-515	55	22	all	all	DET
ejpam-515	55	23	the	the	DET
ejpam-515	55	24	available	available	ADJ
ejpam-515	55	25	x	x	PRON
ejpam-515	55	26	components	component	NOUN
ejpam-515	55	27	is	be	AUX
ejpam-515	55	28	always	always	ADV
ejpam-515	55	29	greater	great	ADJ
ejpam-515	55	30	than	than	ADP
ejpam-515	55	31	the	the	DET
ejpam-515	55	32	maximum	maximum	ADJ
ejpam-515	55	33	likelihood	likelihood	NOUN
ejpam-515	55	34	for	for	ADP
ejpam-515	55	35	a	a	DET
ejpam-515	55	36	sub	sub	NOUN
ejpam-515	55	37	-	-	NOUN
ejpam-515	55	38	model	model	NOUN
ejpam-515	55	39	using	use	VERB
ejpam-515	55	40	a	a	DET
ejpam-515	55	41	subset	subset	NOUN
ejpam-515	55	42	of	of	ADP
ejpam-515	55	43	the	the	DET
ejpam-515	55	44	x	x	NOUN
ejpam-515	55	45	components	component	NOUN
ejpam-515	55	46	.	.	PUNCT
ejpam-515	56	1	but	but	CCONJ
ejpam-515	56	2	a	a	DET
ejpam-515	56	3	model	model	NOUN
ejpam-515	56	4	selection	selection	NOUN
ejpam-515	56	5	criterion	criterion	NOUN
ejpam-515	56	6	can	can	AUX
ejpam-515	56	7	be	be	AUX
ejpam-515	56	8	based	base	VERB
ejpam-515	56	9	on	on	ADP
ejpam-515	56	10	a	a	DET
ejpam-515	56	11	penalised	penalise	VERB
ejpam-515	56	12	log	log	NOUN
ejpam-515	56	13	-	-	PUNCT
ejpam-515	56	14	likelihood	likelihood	NOUN
ejpam-515	56	15	.	.	PUNCT
ejpam-515	57	1	let	let	VERB
ejpam-515	57	2	α	α	PRON
ejpam-515	57	3	be	be	AUX
ejpam-515	57	4	a	a	DET
ejpam-515	57	5	pα	pα	NOUN
ejpam-515	57	6	-	-	PUNCT
ejpam-515	57	7	component	component	NOUN
ejpam-515	57	8	sub	sub	NOUN
ejpam-515	57	9	-	-	NOUN
ejpam-515	57	10	vector	vector	NOUN
ejpam-515	57	11	of	of	ADP
ejpam-515	57	12	(	(	PUNCT
ejpam-515	57	13	1,2	1,2	NUM
ejpam-515	57	14	,	,	PUNCT
ejpam-515	57	15	·	·	PUNCT
ejpam-515	57	16	·	·	PUNCT
ejpam-515	57	17	·	·	PUNCT
ejpam-515	57	18	,	,	PUNCT
ejpam-515	57	19	p	p	NOUN
ejpam-515	57	20	)	)	PUNCT
ejpam-515	57	21	.	.	PUNCT
ejpam-515	58	1	let	let	VERB
ejpam-515	58	2	xα	xα	INTJ
ejpam-515	58	3	and	and	CCONJ
ejpam-515	58	4	β(α	β(α	ADJ
ejpam-515	58	5	)	)	PUNCT
ejpam-515	58	6	be	be	VERB
ejpam-515	58	7	the	the	DET
ejpam-515	58	8	sub	sub	NOUN
ejpam-515	58	9	-	-	NOUN
ejpam-515	58	10	vectors	vector	NOUN
ejpam-515	58	11	of	of	ADP
ejpam-515	58	12	x	x	X
ejpam-515	58	13	and	and	CCONJ
ejpam-515	58	14	β	β	X
ejpam-515	58	15	indexed	index	VERB
ejpam-515	58	16	by	by	ADP
ejpam-515	58	17	α	α	NOUN
ejpam-515	58	18	respectively	respectively	ADV
ejpam-515	58	19	.	.	PUNCT
ejpam-515	59	1	further	far	ADV
ejpam-515	59	2	let	let	VERB
ejpam-515	59	3	logµα	logµα	NOUN
ejpam-515	59	4	=	=	PUNCT
ejpam-515	59	5	ηα	ηα	PROPN
ejpam-515	59	6	=	=	SYM
ejpam-515	59	7	xt	xt	PROPN
ejpam-515	59	8	αβα	αβα	PROPN
ejpam-515	59	9	be	be	AUX
ejpam-515	59	10	a	a	DET
ejpam-515	59	11	poisson	poisson	NOUN
ejpam-515	59	12	regression	regression	NOUN
ejpam-515	59	13	model	model	NOUN
ejpam-515	59	14	containing	contain	VERB
ejpam-515	59	15	a	a	DET
ejpam-515	59	16	subset	subset	NOUN
ejpam-515	59	17	of	of	ADP
ejpam-515	59	18	explanatory	explanatory	ADJ
ejpam-515	59	19	variables	variable	NOUN
ejpam-515	59	20	given	give	VERB
ejpam-515	59	21	by	by	ADP
ejpam-515	59	22	xα	xα	PROPN
ejpam-515	59	23	.	.	PUNCT
ejpam-515	60	1	the	the	DET
ejpam-515	60	2	penalised	penalise	VERB
ejpam-515	60	3	log	log	NOUN
ejpam-515	60	4	-	-	PUNCT
ejpam-515	60	5	likelihood	likelihood	NOUN
ejpam-515	60	6	based	base	VERB
ejpam-515	60	7	model	model	NOUN
ejpam-515	60	8	selection	selection	NOUN
ejpam-515	60	9	criterion	criterion	NOUN
ejpam-515	60	10	can	can	AUX
ejpam-515	60	11	be	be	AUX
ejpam-515	60	12	expressed	express	VERB
ejpam-515	60	13	as	as	ADP
ejpam-515	60	14	s(ηα	s(ηα	PROPN
ejpam-515	60	15	)	)	PUNCT
ejpam-515	60	16	=	=	SYM
ejpam-515	60	17	−ℓ(β̂(α)|yn	−ℓ(β̂(α)|yn	ADJ
ejpam-515	60	18	,	,	PUNCT
ejpam-515	60	19	xnα	xnα	PROPN
ejpam-515	60	20	)	)	PUNCT
ejpam-515	61	1	+	+	CCONJ
ejpam-515	61	2	c(n	c(n	ADJ
ejpam-515	61	3	,	,	PUNCT
ejpam-515	61	4	β̂(α	β̂(α	ADJ
ejpam-515	61	5	)	)	PUNCT
ejpam-515	61	6	)	)	PUNCT
ejpam-515	61	7	,	,	PUNCT
ejpam-515	61	8	(	(	PUNCT
ejpam-515	61	9	3	3	X
ejpam-515	61	10	)	)	PUNCT
ejpam-515	61	11	where	where	SCONJ
ejpam-515	61	12	the	the	DET
ejpam-515	61	13	first	first	ADJ
ejpam-515	61	14	term	term	NOUN
ejpam-515	61	15	is	be	AUX
ejpam-515	61	16	the	the	DET
ejpam-515	61	17	minus	minus	CCONJ
ejpam-515	61	18	maximum	maximum	ADJ
ejpam-515	61	19	log	log	NOUN
ejpam-515	61	20	-	-	PUNCT
ejpam-515	61	21	likelihood	likelihood	NOUN
ejpam-515	61	22	measuring	measure	VERB
ejpam-515	61	23	the	the	DET
ejpam-515	61	24	goodness	goodness	NOUN
ejpam-515	61	25	of	of	ADP
ejpam-515	61	26	fit	fit	NOUN
ejpam-515	61	27	of	of	ADP
ejpam-515	61	28	model	model	NOUN
ejpam-515	61	29	ηα	ηα	PROPN
ejpam-515	61	30	,	,	PUNCT
ejpam-515	61	31	and	and	CCONJ
ejpam-515	61	32	the	the	DET
ejpam-515	61	33	second	second	ADJ
ejpam-515	61	34	term	term	NOUN
ejpam-515	61	35	measures	measure	VERB
ejpam-515	61	36	the	the	DET
ejpam-515	61	37	complexity	complexity	NOUN
ejpam-515	61	38	of	of	ADP
ejpam-515	61	39	the	the	DET
ejpam-515	61	40	model	model	NOUN
ejpam-515	61	41	.	.	PUNCT
ejpam-515	62	1	the	the	DET
ejpam-515	62	2	matrix	matrix	NOUN
ejpam-515	62	3	xnα	xnα	NOUN
ejpam-515	62	4	comprises	comprise	VERB
ejpam-515	62	5	those	those	DET
ejpam-515	62	6	columns	column	NOUN
ejpam-515	62	7	of	of	ADP
ejpam-515	62	8	xn	xn	PROPN
ejpam-515	62	9	indexed	index	VERB
ejpam-515	62	10	by	by	ADP
ejpam-515	62	11	α	α	NOUN
ejpam-515	62	12	;	;	PUNCT
ejpam-515	62	13	and	and	CCONJ
ejpam-515	62	14	β̂(α	β̂(α	PRON
ejpam-515	62	15	)	)	PUNCT
ejpam-515	62	16	is	be	AUX
ejpam-515	62	17	the	the	DET
ejpam-515	62	18	mle	mle	NOUN
ejpam-515	62	19	of	of	ADP
ejpam-515	62	20	β(α	β(α	PROPN
ejpam-515	62	21	)	)	PUNCT
ejpam-515	62	22	.	.	PUNCT
ejpam-515	63	1	under	under	ADP
ejpam-515	63	2	the	the	DET
ejpam-515	63	3	criterion	criterion	NOUN
ejpam-515	63	4	(	(	PUNCT
ejpam-515	63	5	3	3	NUM
ejpam-515	63	6	)	)	PUNCT
ejpam-515	63	7	,	,	PUNCT
ejpam-515	63	8	those	those	DET
ejpam-515	63	9	sub	sub	NOUN
ejpam-515	63	10	-	-	NOUN
ejpam-515	63	11	models	model	NOUN
ejpam-515	63	12	having	have	VERB
ejpam-515	63	13	both	both	CCONJ
ejpam-515	63	14	better	well	ADJ
ejpam-515	63	15	goodness	goodness	NOUN
ejpam-515	63	16	of	of	ADP
ejpam-515	63	17	fit	fit	ADJ
ejpam-515	63	18	and	and	CCONJ
ejpam-515	63	19	smaller	small	ADJ
ejpam-515	63	20	complexity	complexity	NOUN
ejpam-515	63	21	will	will	AUX
ejpam-515	63	22	be	be	AUX
ejpam-515	63	23	preferred	prefer	VERB
ejpam-515	63	24	than	than	ADP
ejpam-515	63	25	the	the	DET
ejpam-515	63	26	others	other	NOUN
ejpam-515	63	27	;	;	PUNCT
ejpam-515	63	28	and	and	CCONJ
ejpam-515	63	29	the	the	DET
ejpam-515	63	30	best	good	ADJ
ejpam-515	63	31	model	model	NOUN
ejpam-515	63	32	will	will	AUX
ejpam-515	63	33	be	be	AUX
ejpam-515	63	34	the	the	DET
ejpam-515	63	35	one	one	NOUN
ejpam-515	63	36	achieving	achieve	VERB
ejpam-515	63	37	the	the	DET
ejpam-515	63	38	smallest	small	ADJ
ejpam-515	63	39	s(ηα	s(ηα	NOUN
ejpam-515	63	40	)	)	PUNCT
ejpam-515	63	41	value	value	NOUN
ejpam-515	63	42	.	.	PUNCT
ejpam-515	64	1	many	many	ADJ
ejpam-515	64	2	commonly	commonly	ADV
ejpam-515	64	3	used	use	VERB
ejpam-515	64	4	model	model	NOUN
ejpam-515	64	5	selection	selection	NOUN
ejpam-515	64	6	criteria	criterion	NOUN
ejpam-515	64	7	,	,	PUNCT
ejpam-515	64	8	such	such	ADJ
ejpam-515	64	9	as	as	ADP
ejpam-515	64	10	aic	aic	PROPN
ejpam-515	64	11	[	[	X
ejpam-515	64	12	1	1	NUM
ejpam-515	64	13	]	]	PUNCT
ejpam-515	64	14	,	,	PUNCT
ejpam-515	64	15	bic	bic	PROPN
ejpam-515	65	1	[	[	X
ejpam-515	65	2	18	18	NUM
ejpam-515	65	3	]	]	PUNCT
ejpam-515	65	4	,	,	PUNCT
ejpam-515	65	5	cp	cp	PROPN
ejpam-515	66	1	[	[	X
ejpam-515	66	2	7	7	NUM
ejpam-515	66	3	]	]	PUNCT
ejpam-515	66	4	and	and	CCONJ
ejpam-515	66	5	stochastic	stochastic	ADJ
ejpam-515	66	6	complexity	complexity	NOUN
ejpam-515	66	7	criterion	criterion	NOUN
ejpam-515	66	8	(	(	PUNCT
ejpam-515	66	9	scc)[16	scc)[16	PROPN
ejpam-515	66	10	,	,	PUNCT
ejpam-515	66	11	17	17	NUM
ejpam-515	66	12	,	,	PUNCT
ejpam-515	66	13	11	11	NUM
ejpam-515	66	14	]	]	PUNCT
ejpam-515	66	15	,	,	PUNCT
ejpam-515	66	16	are	be	AUX
ejpam-515	66	17	of	of	ADP
ejpam-515	66	18	the	the	DET
ejpam-515	66	19	form	form	NOUN
ejpam-515	66	20	given	give	VERB
ejpam-515	66	21	by	by	ADP
ejpam-515	66	22	(	(	PUNCT
ejpam-515	66	23	3	3	NUM
ejpam-515	66	24	)	)	PUNCT
ejpam-515	66	25	.	.	PUNCT
ejpam-515	67	1	for	for	ADP
ejpam-515	67	2	example	example	NOUN
ejpam-515	67	3	,	,	PUNCT
ejpam-515	67	4	for	for	ADP
ejpam-515	67	5	aic	aic	PROPN
ejpam-515	67	6	and	and	CCONJ
ejpam-515	67	7	cp	cp	PROPN
ejpam-515	67	8	c(n	c(n	PROPN
ejpam-515	67	9	,	,	PUNCT
ejpam-515	67	10	β̂(α	β̂(α	PRON
ejpam-515	67	11	)	)	PUNCT
ejpam-515	67	12	)	)	PUNCT
ejpam-515	68	1	=	=	PUNCT
ejpam-515	68	2	pα	pα	NOUN
ejpam-515	68	3	;	;	PUNCT
ejpam-515	68	4	for	for	ADP
ejpam-515	68	5	bic	bic	PROPN
ejpam-515	68	6	c(n	c(n	PROPN
ejpam-515	68	7	,	,	PUNCT
ejpam-515	68	8	β̂(α	β̂(α	PRON
ejpam-515	68	9	)	)	PUNCT
ejpam-515	68	10	)	)	PUNCT
ejpam-515	69	1	=	=	SYM
ejpam-515	69	2	1	1	NUM
ejpam-515	69	3	2	2	NUM
ejpam-515	69	4	pα	pα	NOUN
ejpam-515	69	5	log	log	NOUN
ejpam-515	69	6	n	n	CCONJ
ejpam-515	69	7	;	;	PUNCT
ejpam-515	69	8	and	and	CCONJ
ejpam-515	69	9	for	for	ADP
ejpam-515	69	10	scc	scc	PROPN
ejpam-515	69	11	c(n	c(n	PROPN
ejpam-515	69	12	,	,	PUNCT
ejpam-515	69	13	β̂(α	β̂(α	PRON
ejpam-515	69	14	)	)	PUNCT
ejpam-515	69	15	)	)	PUNCT
ejpam-515	70	1	=	=	SYM
ejpam-515	70	2	1	1	NUM
ejpam-515	70	3	2	2	NUM
ejpam-515	70	4	log	log	NOUN
ejpam-515	70	5	|in(β̂(α))|+	|in(β̂(α))|+	NOUN
ejpam-515	70	6	∑pα	∑pα	PUNCT
ejpam-515	70	7	i=2	i=2	PROPN
ejpam-515	70	8	log(|β̂(α)i|+	log(|β̂(α)i|+	VERB
ejpam-515	70	9	ǫn−1/4	ǫn−1/4	NUM
ejpam-515	70	10	)	)	PUNCT
ejpam-515	70	11	where	where	SCONJ
ejpam-515	70	12	β̂(α)i	β̂(α)i	PROPN
ejpam-515	70	13	is	be	AUX
ejpam-515	70	14	the	the	DET
ejpam-515	70	15	i	i	PROPN
ejpam-515	70	16	-	-	PUNCT
ejpam-515	70	17	th	th	X
ejpam-515	70	18	component	component	NOUN
ejpam-515	70	19	of	of	ADP
ejpam-515	70	20	β̂(α	β̂(α	PRON
ejpam-515	70	21	)	)	PUNCT
ejpam-515	70	22	,	,	PUNCT
ejpam-515	70	23	and	and	CCONJ
ejpam-515	70	24	ǫ	ǫ	PRON
ejpam-515	70	25	is	be	AUX
ejpam-515	70	26	a	a	DET
ejpam-515	70	27	specified	specified	ADJ
ejpam-515	70	28	quantity	quantity	NOUN
ejpam-515	70	29	to	to	PART
ejpam-515	70	30	ensure	ensure	VERB
ejpam-515	70	31	the	the	DET
ejpam-515	70	32	invariance	invariance	NOUN
ejpam-515	70	33	of	of	ADP
ejpam-515	70	34	the	the	DET
ejpam-515	70	35	scc	scc	NOUN
ejpam-515	70	36	[	[	X
ejpam-515	70	37	see	see	VERB
ejpam-515	70	38	11	11	NUM
ejpam-515	70	39	,	,	PUNCT
ejpam-515	70	40	for	for	ADP
ejpam-515	70	41	details	detail	NOUN
ejpam-515	70	42	]	]	PUNCT
ejpam-515	70	43	.	.	PUNCT
ejpam-515	71	1	assuming	assume	VERB
ejpam-515	71	2	that	that	SCONJ
ejpam-515	71	3	the	the	DET
ejpam-515	71	4	model	model	NOUN
ejpam-515	71	5	logµ	logµ	NOUN
ejpam-515	71	6	=	=	PRON
ejpam-515	71	7	xtβ	xtβ	PROPN
ejpam-515	71	8	is	be	AUX
ejpam-515	71	9	the	the	DET
ejpam-515	71	10	full	full	ADJ
ejpam-515	71	11	model	model	NOUN
ejpam-515	71	12	which	which	PRON
ejpam-515	71	13	includes	include	VERB
ejpam-515	71	14	all	all	DET
ejpam-515	71	15	the	the	DET
ejpam-515	71	16	explanatory	explanatory	ADJ
ejpam-515	71	17	variables	variable	NOUN
ejpam-515	71	18	available	available	ADJ
ejpam-515	71	19	and	and	CCONJ
ejpam-515	71	20	the	the	DET
ejpam-515	71	21	first	first	ADJ
ejpam-515	71	22	component	component	NOUN
ejpam-515	71	23	of	of	ADP
ejpam-515	71	24	x	x	NOUN
ejpam-515	71	25	is	be	AUX
ejpam-515	71	26	an	an	DET
ejpam-515	71	27	intercept	intercept	NOUN
ejpam-515	71	28	term	term	NOUN
ejpam-515	71	29	,	,	PUNCT
ejpam-515	71	30	there	there	PRON
ejpam-515	71	31	will	will	AUX
ejpam-515	71	32	be	be	AUX
ejpam-515	71	33	in	in	ADP
ejpam-515	71	34	total	total	ADJ
ejpam-515	71	35	2p−1	2p−1	NUM
ejpam-515	71	36	sub	sub	NOUN
ejpam-515	71	37	-	-	NOUN
ejpam-515	71	38	models	model	NOUN
ejpam-515	71	39	of	of	ADP
ejpam-515	71	40	the	the	DET
ejpam-515	71	41	form	form	NOUN
ejpam-515	71	42	logµα	logµα	NOUN
ejpam-515	71	43	=	=	SYM
ejpam-515	71	44	xt	xt	PROPN
ejpam-515	71	45	αβ(α	αβ(α	X
ejpam-515	71	46	)	)	PUNCT
ejpam-515	71	47	for	for	ADP
ejpam-515	71	48	selection	selection	NOUN
ejpam-515	71	49	,	,	PUNCT
ejpam-515	71	50	provided	provide	VERB
ejpam-515	71	51	that	that	SCONJ
ejpam-515	71	52	only	only	ADV
ejpam-515	71	53	those	those	DET
ejpam-515	71	54	models	model	NOUN
ejpam-515	71	55	having	have	VERB
ejpam-515	71	56	an	an	DET
ejpam-515	71	57	intercept	intercept	NOUN
ejpam-515	71	58	term	term	NOUN
ejpam-515	71	59	are	be	AUX
ejpam-515	71	60	considered	consider	VERB
ejpam-515	71	61	.	.	PUNCT
ejpam-515	72	1	in	in	ADP
ejpam-515	72	2	this	this	DET
ejpam-515	72	3	paper	paper	NOUN
ejpam-515	72	4	we	we	PRON
ejpam-515	72	5	assume	assume	VERB
ejpam-515	72	6	that	that	SCONJ
ejpam-515	72	7	some	some	DET
ejpam-515	72	8	components	component	NOUN
ejpam-515	72	9	of	of	ADP
ejpam-515	72	10	β0	β0	NOUN
ejpam-515	72	11	=	=	PUNCT
ejpam-515	72	12	(	(	PUNCT
ejpam-515	72	13	β01	β01	PROPN
ejpam-515	72	14	,	,	PUNCT
ejpam-515	72	15	·	·	PUNCT
ejpam-515	72	16	·	·	PUNCT
ejpam-515	72	17	·	·	PUNCT
ejpam-515	72	18	,	,	PUNCT
ejpam-515	72	19	β0p	β0p	PROPN
ejpam-515	72	20	)	)	PUNCT
ejpam-515	72	21	t	t	PROPN
ejpam-515	72	22	,	,	PUNCT
ejpam-515	72	23	the	the	DET
ejpam-515	72	24	true	true	ADJ
ejpam-515	72	25	value	value	NOUN
ejpam-515	72	26	of	of	ADP
ejpam-515	72	27	β	β	PROPN
ejpam-515	72	28	,	,	PUNCT
ejpam-515	72	29	are	be	AUX
ejpam-515	72	30	equal	equal	ADJ
ejpam-515	72	31	to	to	ADP
ejpam-515	72	32	0	0	NUM
ejpam-515	72	33	.	.	PUNCT
ejpam-515	73	1	we	we	PRON
ejpam-515	73	2	also	also	ADV
ejpam-515	73	3	use	use	VERB
ejpam-515	73	4	α	α	NOUN
ejpam-515	73	5	to	to	PART
ejpam-515	73	6	represent	represent	VERB
ejpam-515	73	7	a	a	DET
ejpam-515	73	8	poisson	poisson	NOUN
ejpam-515	73	9	regression	regression	NOUN
ejpam-515	73	10	sub	sub	ADJ
ejpam-515	73	11	-	-	ADJ
ejpam-515	73	12	model	model	ADJ
ejpam-515	73	13	logµα	logµα	PROPN
ejpam-515	73	14	=	=	SYM
ejpam-515	73	15	xt	xt	PROPN
ejpam-515	73	16	αβ(α	αβ(α	NOUN
ejpam-515	73	17	)	)	PUNCT
ejpam-515	73	18	,	,	PUNCT
ejpam-515	73	19	which	which	PRON
ejpam-515	73	20	is	be	AUX
ejpam-515	73	21	actually	actually	ADV
ejpam-515	73	22	a	a	DET
ejpam-515	73	23	one	one	NUM
ejpam-515	73	24	-	-	PUNCT
ejpam-515	73	25	to	to	ADP
ejpam-515	73	26	-	-	PUNCT
ejpam-515	73	27	one	one	NUM
ejpam-515	73	28	representation	representation	NOUN
ejpam-515	73	29	.	.	PUNCT
ejpam-515	74	1	then	then	ADV
ejpam-515	74	2	all	all	DET
ejpam-515	74	3	the	the	DET
ejpam-515	74	4	2p	2p	NUM
ejpam-515	74	5	−	−	NOUN
ejpam-515	74	6	1	1	NUM
ejpam-515	74	7	sub	sub	NOUN
ejpam-515	74	8	-	-	NOUN
ejpam-515	74	9	models	model	NOUN
ejpam-515	74	10	can	can	AUX
ejpam-515	74	11	be	be	AUX
ejpam-515	74	12	classified	classify	VERB
ejpam-515	74	13	into	into	ADP
ejpam-515	74	14	the	the	DET
ejpam-515	74	15	following	follow	VERB
ejpam-515	74	16	two	two	NUM
ejpam-515	74	17	groups	group	NOUN
ejpam-515	74	18	:	:	PUNCT
ejpam-515	74	19	g.	g.	PROPN
ejpam-515	74	20	qian	qian	PROPN
ejpam-515	74	21	/	/	SYM
ejpam-515	74	22	eur	eur	PROPN
ejpam-515	74	23	.	.	PUNCT
ejpam-515	75	1	j.	j.	PROPN
ejpam-515	75	2	pure	pure	PROPN
ejpam-515	75	3	appl	appl	PROPN
ejpam-515	75	4	.	.	PROPN
ejpam-515	75	5	math	math	PROPN
ejpam-515	75	6	,	,	PUNCT
ejpam-515	75	7	3	3	NUM
ejpam-515	75	8	(	(	PUNCT
ejpam-515	75	9	2010	2010	NUM
ejpam-515	75	10	)	)	PUNCT
ejpam-515	75	11	,	,	PUNCT
ejpam-515	75	12	417	417	NUM
ejpam-515	75	13	-	-	SYM
ejpam-515	75	14	434	434	NUM
ejpam-515	75	15	420	420	NUM
ejpam-515	75	16	1	1	NUM
ejpam-515	75	17	.	.	PUNCT
ejpam-515	75	18	ac	ac	PROPN
ejpam-515	76	1	=	=	PUNCT
ejpam-515	76	2	{	{	PUNCT
ejpam-515	76	3	α	α	NOUN
ejpam-515	76	4	:	:	PUNCT
ejpam-515	76	5	β0i	β0i	SYM
ejpam-515	76	6	=	=	SYM
ejpam-515	76	7	0	0	NUM
ejpam-515	76	8	for	for	ADP
ejpam-515	76	9	any	any	DET
ejpam-515	76	10	i	i	PRON
ejpam-515	76	11	6∈	6∈	NOUN
ejpam-515	76	12	α	α	NOUN
ejpam-515	76	13	}	}	PUNCT
ejpam-515	76	14	;	;	PUNCT
ejpam-515	76	15	2	2	X
ejpam-515	76	16	.	.	X
ejpam-515	77	1	aw	aw	INTJ
ejpam-515	77	2	=	=	SYM
ejpam-515	77	3	{	{	PUNCT
ejpam-515	77	4	α	α	NOUN
ejpam-515	77	5	:	:	PUNCT
ejpam-515	77	6	β0i	β0i	SYM
ejpam-515	77	7	6=	6=	SYM
ejpam-515	77	8	0	0	NUM
ejpam-515	77	9	for	for	ADP
ejpam-515	77	10	some	some	PRON
ejpam-515	78	1	i	i	PRON
ejpam-515	78	2	6∈	6∈	NOUN
ejpam-515	78	3	α	α	NOUN
ejpam-515	78	4	}	}	PUNCT
ejpam-515	78	5	.	.	PUNCT
ejpam-515	79	1	apparently	apparently	ADV
ejpam-515	79	2	,	,	PUNCT
ejpam-515	79	3	every	every	DET
ejpam-515	79	4	sub	sub	NOUN
ejpam-515	79	5	-	-	NOUN
ejpam-515	79	6	model	model	NOUN
ejpam-515	79	7	in	in	ADP
ejpam-515	79	8	aw	aw	INTJ
ejpam-515	79	9	is	be	AUX
ejpam-515	79	10	a	a	DET
ejpam-515	79	11	wrong	wrong	ADJ
ejpam-515	79	12	model	model	NOUN
ejpam-515	79	13	which	which	PRON
ejpam-515	79	14	misses	miss	VERB
ejpam-515	79	15	at	at	ADV
ejpam-515	79	16	least	least	ADV
ejpam-515	79	17	one	one	NUM
ejpam-515	79	18	x	x	NOUN
ejpam-515	79	19	variable	variable	NOUN
ejpam-515	79	20	having	have	VERB
ejpam-515	79	21	non	non	ADJ
ejpam-515	79	22	-	-	ADJ
ejpam-515	79	23	zero	zero	NUM
ejpam-515	79	24	effect	effect	NOUN
ejpam-515	79	25	on	on	ADP
ejpam-515	79	26	y	y	PROPN
ejpam-515	79	27	,	,	PUNCT
ejpam-515	79	28	while	while	SCONJ
ejpam-515	79	29	every	every	DET
ejpam-515	79	30	sub	sub	ADJ
ejpam-515	79	31	-	-	ADJ
ejpam-515	79	32	model	model	ADJ
ejpam-515	79	33	inac	inac	NOUN
ejpam-515	79	34	is	be	AUX
ejpam-515	79	35	a	a	DET
ejpam-515	79	36	correct	correct	ADJ
ejpam-515	79	37	model	model	NOUN
ejpam-515	79	38	which	which	PRON
ejpam-515	79	39	includes	include	VERB
ejpam-515	79	40	all	all	DET
ejpam-515	79	41	x	x	PROPN
ejpam-515	79	42	variables	variable	NOUN
ejpam-515	79	43	having	have	VERB
ejpam-515	79	44	non	non	ADJ
ejpam-515	79	45	-	-	ADJ
ejpam-515	79	46	zero	zero	NUM
ejpam-515	79	47	effects	effect	NOUN
ejpam-515	79	48	on	on	ADP
ejpam-515	79	49	y	y	PROPN
ejpam-515	79	50	.	.	PUNCT
ejpam-515	80	1	but	but	CCONJ
ejpam-515	80	2	the	the	DET
ejpam-515	80	3	models	model	NOUN
ejpam-515	80	4	inac	inac	ADJ
ejpam-515	80	5	may	may	AUX
ejpam-515	80	6	contain	contain	VERB
ejpam-515	80	7	some	some	DET
ejpam-515	80	8	redundant	redundant	ADJ
ejpam-515	80	9	x	x	NOUN
ejpam-515	80	10	variables	variable	NOUN
ejpam-515	80	11	having	have	VERB
ejpam-515	80	12	no	no	DET
ejpam-515	80	13	effects	effect	NOUN
ejpam-515	80	14	on	on	ADP
ejpam-515	80	15	y	y	PROPN
ejpam-515	80	16	.	.	PUNCT
ejpam-515	81	1	an	an	DET
ejpam-515	81	2	ideal	ideal	ADJ
ejpam-515	81	3	model	model	NOUN
ejpam-515	81	4	selection	selection	NOUN
ejpam-515	81	5	criterion	criterion	NOUN
ejpam-515	81	6	should	should	AUX
ejpam-515	81	7	render	render	VERB
ejpam-515	81	8	the	the	DET
ejpam-515	81	9	selection	selection	NOUN
ejpam-515	81	10	of	of	ADP
ejpam-515	81	11	the	the	DET
ejpam-515	81	12	simplest	simple	ADJ
ejpam-515	81	13	correct	correct	ADJ
ejpam-515	81	14	model	model	NOUN
ejpam-515	81	15	in	in	ADP
ejpam-515	81	16	ac	ac	PROPN
ejpam-515	81	17	containing	contain	VERB
ejpam-515	81	18	no	no	DET
ejpam-515	81	19	redundant	redundant	ADJ
ejpam-515	81	20	explanatory	explanatory	ADJ
ejpam-515	81	21	variables	variable	NOUN
ejpam-515	81	22	.	.	PUNCT
ejpam-515	82	1	in	in	ADP
ejpam-515	82	2	this	this	DET
ejpam-515	82	3	paper	paper	NOUN
ejpam-515	82	4	we	we	PRON
ejpam-515	82	5	will	will	AUX
ejpam-515	82	6	show	show	VERB
ejpam-515	82	7	that	that	SCONJ
ejpam-515	82	8	the	the	DET
ejpam-515	82	9	penalised	penalise	VERB
ejpam-515	82	10	log	log	NOUN
ejpam-515	82	11	-	-	PUNCT
ejpam-515	82	12	likelihood	likelihood	NOUN
ejpam-515	82	13	based	base	VERB
ejpam-515	82	14	model	model	NOUN
ejpam-515	82	15	selection	selection	NOUN
ejpam-515	82	16	criterion	criterion	NOUN
ejpam-515	82	17	,	,	PUNCT
ejpam-515	82	18	under	under	ADP
ejpam-515	82	19	some	some	DET
ejpam-515	82	20	general	general	ADJ
ejpam-515	82	21	conditions	condition	NOUN
ejpam-515	82	22	,	,	PUNCT
ejpam-515	82	23	selects	select	VERB
ejpam-515	82	24	the	the	DET
ejpam-515	82	25	simplest	simple	ADJ
ejpam-515	82	26	correct	correct	ADJ
ejpam-515	82	27	model	model	NOUN
ejpam-515	82	28	with	with	ADP
ejpam-515	82	29	probability	probability	NOUN
ejpam-515	82	30	1	1	NUM
ejpam-515	82	31	as	as	SCONJ
ejpam-515	82	32	the	the	DET
ejpam-515	82	33	sample	sample	NOUN
ejpam-515	82	34	size	size	NOUN
ejpam-515	82	35	n	n	PRON
ejpam-515	82	36	goes	go	VERB
ejpam-515	82	37	to	to	ADP
ejpam-515	82	38	infinity	infinity	NOUN
ejpam-515	82	39	.	.	PUNCT
ejpam-515	83	1	for	for	ADP
ejpam-515	83	2	simplicity	simplicity	NOUN
ejpam-515	83	3	of	of	ADP
ejpam-515	83	4	the	the	DET
ejpam-515	83	5	presentation	presentation	NOUN
ejpam-515	83	6	,	,	PUNCT
ejpam-515	83	7	we	we	PRON
ejpam-515	83	8	assume	assume	VERB
ejpam-515	83	9	the	the	DET
ejpam-515	83	10	simplest	simple	ADJ
ejpam-515	83	11	correct	correct	ADJ
ejpam-515	83	12	model	model	NOUN
ejpam-515	83	13	in	in	ADP
ejpam-515	83	14	ac	ac	PROPN
ejpam-515	83	15	to	to	PART
ejpam-515	83	16	be	be	AUX
ejpam-515	83	17	unique	unique	ADJ
ejpam-515	83	18	,	,	PUNCT
ejpam-515	83	19	which	which	PRON
ejpam-515	83	20	is	be	AUX
ejpam-515	83	21	the	the	DET
ejpam-515	83	22	case	case	NOUN
ejpam-515	83	23	if	if	SCONJ
ejpam-515	83	24	all	all	DET
ejpam-515	83	25	components	component	NOUN
ejpam-515	83	26	of	of	ADP
ejpam-515	83	27	x	x	PRON
ejpam-515	83	28	are	be	AUX
ejpam-515	83	29	linearly	linearly	ADV
ejpam-515	83	30	independent	independent	ADJ
ejpam-515	83	31	of	of	ADP
ejpam-515	83	32	each	each	DET
ejpam-515	83	33	other	other	ADJ
ejpam-515	83	34	.	.	PUNCT
ejpam-515	84	1	3	3	X
ejpam-515	84	2	.	.	NOUN
ejpam-515	84	3	conditions	condition	NOUN
ejpam-515	84	4	and	and	CCONJ
ejpam-515	84	5	main	main	ADJ
ejpam-515	84	6	results	result	NOUN
ejpam-515	84	7	let	let	VERB
ejpam-515	84	8	λ1{s	λ1{s	VERB
ejpam-515	84	9	}	}	PUNCT
ejpam-515	84	10	≤	≤	NOUN
ejpam-515	84	11	·	·	PUNCT
ejpam-515	84	12	·	·	PUNCT
ejpam-515	84	13	·	·	PUNCT
ejpam-515	85	1	≤	≤	NUM
ejpam-515	85	2	λp{s	λp{	VERB
ejpam-515	85	3	}	}	PUNCT
ejpam-515	85	4	be	be	AUX
ejpam-515	85	5	the	the	DET
ejpam-515	85	6	p	p	NOUN
ejpam-515	85	7	eigenvalues	eigenvalue	NOUN
ejpam-515	85	8	of	of	ADP
ejpam-515	85	9	a	a	DET
ejpam-515	85	10	p	p	NOUN
ejpam-515	85	11	×	×	NOUN
ejpam-515	85	12	p	p	X
ejpam-515	85	13	symmetric	symmetric	ADJ
ejpam-515	85	14	matrix	matrix	NOUN
ejpam-515	85	15	s.	s.	PROPN
ejpam-515	85	16	also	also	ADV
ejpam-515	85	17	let	let	VERB
ejpam-515	85	18	b	b	NOUN
ejpam-515	85	19	=	=	SYM
ejpam-515	85	20	1	1	NUM
ejpam-515	85	21	2	2	NUM
ejpam-515	85	22	min1≤i≤pα0	min1≤i≤pα0	PROPN
ejpam-515	85	23	|β0(α0)i|	|β0(α0)i|	NOUN
ejpam-515	85	24	,	,	PUNCT
ejpam-515	85	25	where	where	SCONJ
ejpam-515	85	26	α0	α0	PROPN
ejpam-515	85	27	is	be	AUX
ejpam-515	85	28	the	the	DET
ejpam-515	85	29	correct	correct	ADJ
ejpam-515	85	30	model	model	NOUN
ejpam-515	85	31	in	in	ADP
ejpam-515	85	32	ac	ac	PROPN
ejpam-515	85	33	with	with	ADP
ejpam-515	85	34	the	the	DET
ejpam-515	85	35	smallest	small	ADJ
ejpam-515	85	36	dimension	dimension	NOUN
ejpam-515	85	37	,	,	PUNCT
ejpam-515	85	38	and	and	CCONJ
ejpam-515	85	39	β0(α0)i	β0(α0)i	NUM
ejpam-515	85	40	is	be	AUX
ejpam-515	85	41	the	the	DET
ejpam-515	85	42	i	i	PROPN
ejpam-515	85	43	-	-	PUNCT
ejpam-515	85	44	th	th	VERB
ejpam-515	85	45	component	component	NOUN
ejpam-515	85	46	of	of	ADP
ejpam-515	85	47	β0(α0	β0(α0	NUM
ejpam-515	85	48	)	)	PUNCT
ejpam-515	85	49	.	.	PUNCT
ejpam-515	86	1	we	we	PRON
ejpam-515	86	2	assume	assume	VERB
ejpam-515	86	3	b	b	X
ejpam-515	86	4	>	>	X
ejpam-515	86	5	0	0	PUNCT
ejpam-515	87	1	in	in	ADP
ejpam-515	87	2	this	this	DET
ejpam-515	87	3	paper	paper	NOUN
ejpam-515	87	4	.	.	PUNCT
ejpam-515	88	1	note	note	VERB
ejpam-515	88	2	that	that	SCONJ
ejpam-515	88	3	b	b	NOUN
ejpam-515	88	4	is	be	AUX
ejpam-515	88	5	only	only	ADV
ejpam-515	88	6	used	use	VERB
ejpam-515	88	7	in	in	ADP
ejpam-515	88	8	the	the	DET
ejpam-515	88	9	proof	proof	NOUN
ejpam-515	88	10	of	of	ADP
ejpam-515	88	11	theorem	theorem	NOUN
ejpam-515	88	12	3	3	NUM
ejpam-515	88	13	in	in	ADP
ejpam-515	88	14	this	this	DET
ejpam-515	88	15	paper	paper	NOUN
ejpam-515	88	16	;	;	PUNCT
ejpam-515	88	17	and	and	CCONJ
ejpam-515	88	18	b	b	X
ejpam-515	88	19	=	=	SYM
ejpam-515	88	20	0	0	NUM
ejpam-515	88	21	represents	represent	VERB
ejpam-515	88	22	the	the	DET
ejpam-515	88	23	case	case	NOUN
ejpam-515	88	24	whereaw	whereaw	NOUN
ejpam-515	88	25	is	be	AUX
ejpam-515	88	26	an	an	DET
ejpam-515	88	27	empty	empty	ADJ
ejpam-515	88	28	set	set	NOUN
ejpam-515	88	29	thereby	thereby	ADV
ejpam-515	88	30	theorem	theorem	VERB
ejpam-515	88	31	3	3	NUM
ejpam-515	88	32	is	be	AUX
ejpam-515	88	33	not	not	PART
ejpam-515	88	34	applicable	applicable	ADJ
ejpam-515	88	35	.	.	PUNCT
ejpam-515	89	1	further	far	ADV
ejpam-515	89	2	define	define	VERB
ejpam-515	89	3	δn	δn	NOUN
ejpam-515	89	4	=	=	PUNCT
ejpam-515	89	5	q	q	PROPN
ejpam-515	89	6	max	max	PROPN
ejpam-515	89	7	1≤i≤n	1≤i≤n	NUM
ejpam-515	89	8	µ0ix	µ0ix	SYM
ejpam-515	89	9	t	t	NOUN
ejpam-515	90	1	i	i	PROPN
ejpam-515	90	2	in(β0	in(β0	PROPN
ejpam-515	90	3	)	)	PUNCT
ejpam-515	91	1	−1xi	−1xi	PROPN
ejpam-515	91	2	and	and	CCONJ
ejpam-515	91	3	ξn	ξn	PROPN
ejpam-515	92	1	=	=	PUNCT
ejpam-515	92	2	q	q	PROPN
ejpam-515	92	3	max	max	PROPN
ejpam-515	92	4	1≤i≤n	1≤i≤n	NUM
ejpam-515	92	5	xt	xt	PROPN
ejpam-515	92	6	i	i	PROPN
ejpam-515	92	7	in(β0	in(β0	PROPN
ejpam-515	92	8	)	)	PUNCT
ejpam-515	93	1	−1xi	−1xi	PROPN
ejpam-515	93	2	where	where	SCONJ
ejpam-515	93	3	µ0i	µ0i	PROPN
ejpam-515	93	4	=	=	SYM
ejpam-515	93	5	ext	ext	NOUN
ejpam-515	94	1	i	i	PRON
ejpam-515	94	2	β0	β0	PROPN
ejpam-515	94	3	is	be	AUX
ejpam-515	94	4	the	the	DET
ejpam-515	94	5	true	true	ADJ
ejpam-515	94	6	value	value	NOUN
ejpam-515	94	7	of	of	ADP
ejpam-515	94	8	µi	µi	PROPN
ejpam-515	94	9	.	.	PUNCT
ejpam-515	95	1	the	the	DET
ejpam-515	95	2	following	follow	VERB
ejpam-515	95	3	conditions	condition	NOUN
ejpam-515	95	4	will	will	AUX
ejpam-515	95	5	be	be	AUX
ejpam-515	95	6	required	require	VERB
ejpam-515	95	7	in	in	ADP
ejpam-515	95	8	various	various	ADJ
ejpam-515	95	9	places	place	NOUN
ejpam-515	95	10	in	in	ADP
ejpam-515	95	11	proving	prove	VERB
ejpam-515	95	12	our	our	PRON
ejpam-515	95	13	main	main	ADJ
ejpam-515	95	14	results	result	NOUN
ejpam-515	95	15	:	:	PUNCT
ejpam-515	95	16	(	(	PUNCT
ejpam-515	95	17	c.1	c.1	NOUN
ejpam-515	95	18	)	)	PUNCT
ejpam-515	95	19	.	.	PUNCT
ejpam-515	96	1	limn→∞λ	limn→∞λ	PROPN
ejpam-515	96	2	j{in(β0	j{in(β0	PROPN
ejpam-515	96	3	)	)	PUNCT
ejpam-515	96	4	}	}	PUNCT
ejpam-515	96	5	=	=	SYM
ejpam-515	96	6	∞	∞	PROPN
ejpam-515	96	7	,	,	PUNCT
ejpam-515	96	8	j	j	PROPN
ejpam-515	96	9	=	=	SYM
ejpam-515	96	10	1	1	NUM
ejpam-515	96	11	,	,	PUNCT
ejpam-515	96	12	·	·	PUNCT
ejpam-515	96	13	·	·	PUNCT
ejpam-515	96	14	·	·	PUNCT
ejpam-515	96	15	,	,	PUNCT
ejpam-515	97	1	p.	p.	NOUN
ejpam-515	97	2	also	also	ADV
ejpam-515	97	3	there	there	PRON
ejpam-515	97	4	exists	exist	VERB
ejpam-515	97	5	a	a	DET
ejpam-515	97	6	constant	constant	ADJ
ejpam-515	97	7	b0	b0	NOUN
ejpam-515	97	8	such	such	ADJ
ejpam-515	97	9	that	that	SCONJ
ejpam-515	97	10	0	0	NUM
ejpam-515	97	11	<	<	X
ejpam-515	97	12	λp{in(β0	λp{in(β0	PROPN
ejpam-515	97	13	)	)	PUNCT
ejpam-515	97	14	}	}	PUNCT
ejpam-515	97	15	≤	≤	NUM
ejpam-515	97	16	b0λ1{in(β0	b0λ1{in(β0	NOUN
ejpam-515	97	17	)	)	PUNCT
ejpam-515	97	18	}	}	PUNCT
ejpam-515	97	19	.	.	PUNCT
ejpam-515	98	1	(	(	PUNCT
ejpam-515	98	2	c.2	c.2	NOUN
ejpam-515	98	3	)	)	PUNCT
ejpam-515	98	4	.	.	PUNCT
ejpam-515	99	1	b1n≤	b1n≤	PROPN
ejpam-515	99	2	λp{in(β0	λp{in(β0	NUM
ejpam-515	99	3	)	)	PUNCT
ejpam-515	99	4	}	}	PUNCT
ejpam-515	99	5	≤	≤	NUM
ejpam-515	99	6	b2n	b2n	VERB
ejpam-515	99	7	for	for	ADP
ejpam-515	99	8	some	some	DET
ejpam-515	99	9	positive	positive	ADJ
ejpam-515	99	10	constants	constant	NOUN
ejpam-515	99	11	b1	b1	NOUN
ejpam-515	99	12	and	and	CCONJ
ejpam-515	99	13	b2	b2	NOUN
ejpam-515	99	14	.	.	PUNCT
ejpam-515	100	1	(	(	PUNCT
ejpam-515	100	2	c.3	c.3	NOUN
ejpam-515	100	3	)	)	PUNCT
ejpam-515	100	4	.	.	PUNCT
ejpam-515	101	1	ξn	ξn	PROPN
ejpam-515	101	2	p	p	NOUN
ejpam-515	101	3	log	log	NOUN
ejpam-515	101	4	logλp{in(β0)}=	logλp{in(β0)}=	PROPN
ejpam-515	101	5	o(1	o(1	PROPN
ejpam-515	101	6	)	)	PUNCT
ejpam-515	101	7	.	.	PUNCT
ejpam-515	102	1	(	(	PUNCT
ejpam-515	102	2	c.4	c.4	PROPN
ejpam-515	102	3	)	)	PUNCT
ejpam-515	102	4	.	.	PUNCT
ejpam-515	103	1	δn	δn	PROPN
ejpam-515	103	2	p	p	ADJ
ejpam-515	103	3	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	103	4	)	)	PUNCT
ejpam-515	103	5	}	}	PUNCT
ejpam-515	104	1	p	p	NOUN
ejpam-515	104	2	log	log	NOUN
ejpam-515	104	3	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	104	4	)	)	PUNCT
ejpam-515	104	5	}	}	PUNCT
ejpam-515	104	6	=	=	SYM
ejpam-515	104	7	o(1	o(1	NOUN
ejpam-515	104	8	)	)	PUNCT
ejpam-515	104	9	.	.	PUNCT
ejpam-515	105	1	(	(	PUNCT
ejpam-515	105	2	c.5	c.5	NOUN
ejpam-515	105	3	)	)	PUNCT
ejpam-515	105	4	.	.	PUNCT
ejpam-515	106	1	δn	δn	PROPN
ejpam-515	106	2	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	106	3	)	)	PUNCT
ejpam-515	106	4	}	}	PUNCT
ejpam-515	106	5	p	p	NOUN
ejpam-515	106	6	log	log	NOUN
ejpam-515	106	7	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	106	8	)	)	PUNCT
ejpam-515	106	9	}	}	PUNCT
ejpam-515	106	10	=	=	SYM
ejpam-515	106	11	o(1	o(1	NOUN
ejpam-515	106	12	)	)	PUNCT
ejpam-515	106	13	.	.	PUNCT
ejpam-515	107	1	(	(	PUNCT
ejpam-515	107	2	c.6	c.6	NOUN
ejpam-515	107	3	)	)	PUNCT
ejpam-515	107	4	.	.	PUNCT
ejpam-515	108	1	ξn	ξn	PROPN
ejpam-515	108	2	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	108	3	)	)	PUNCT
ejpam-515	108	4	}	}	PUNCT
ejpam-515	108	5	p	p	NOUN
ejpam-515	108	6	log	log	NOUN
ejpam-515	108	7	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	108	8	)	)	PUNCT
ejpam-515	108	9	}	}	PUNCT
ejpam-515	108	10	=	=	SYM
ejpam-515	108	11	o(1	o(1	NOUN
ejpam-515	108	12	)	)	PUNCT
ejpam-515	108	13	.	.	PUNCT
ejpam-515	109	1	(	(	PUNCT
ejpam-515	109	2	c.7	c.7	NOUN
ejpam-515	109	3	)	)	PUNCT
ejpam-515	109	4	.	.	PUNCT
ejpam-515	110	1	λ1{x	λ1{x	PROPN
ejpam-515	110	2	t	t	PROPN
ejpam-515	110	3	nmnxn	nmnxn	PROPN
ejpam-515	110	4	}	}	PUNCT
ejpam-515	110	5	≥	≥	NUM
ejpam-515	110	6	b3n	b3n	NOUN
ejpam-515	110	7	for	for	ADP
ejpam-515	110	8	some	some	DET
ejpam-515	110	9	constant	constant	ADJ
ejpam-515	110	10	b3	b3	PROPN
ejpam-515	110	11	>	>	X
ejpam-515	110	12	0	0	PROPN
ejpam-515	110	13	,	,	PUNCT
ejpam-515	110	14	where	where	SCONJ
ejpam-515	110	15	mn	mn	PROPN
ejpam-515	110	16	=	=	PROPN
ejpam-515	110	17	diag{µ01e−2b‖x1‖	diag{µ01e−2b‖x1‖	PROPN
ejpam-515	110	18	,	,	PUNCT
ejpam-515	110	19	·	·	PUNCT
ejpam-515	110	20	·	·	PUNCT
ejpam-515	110	21	·	·	PUNCT
ejpam-515	110	22	,	,	PUNCT
ejpam-515	110	23	µ0ne−2b‖xn‖	µ0ne−2b‖xn‖	ADP
ejpam-515	110	24	}	}	PUNCT
ejpam-515	110	25	.	.	PUNCT
ejpam-515	111	1	g.	g.	PROPN
ejpam-515	111	2	qian	qian	PROPN
ejpam-515	111	3	/	/	SYM
ejpam-515	111	4	eur	eur	PROPN
ejpam-515	111	5	.	.	PUNCT
ejpam-515	112	1	j.	j.	PROPN
ejpam-515	112	2	pure	pure	PROPN
ejpam-515	112	3	appl	appl	PROPN
ejpam-515	112	4	.	.	PROPN
ejpam-515	112	5	math	math	PROPN
ejpam-515	112	6	,	,	PUNCT
ejpam-515	112	7	3	3	NUM
ejpam-515	112	8	(	(	PUNCT
ejpam-515	112	9	2010	2010	NUM
ejpam-515	112	10	)	)	PUNCT
ejpam-515	112	11	,	,	PUNCT
ejpam-515	112	12	417	417	NUM
ejpam-515	112	13	-	-	SYM
ejpam-515	112	14	434	434	NUM
ejpam-515	112	15	421	421	NUM
ejpam-515	112	16	note	note	NOUN
ejpam-515	112	17	that	that	SCONJ
ejpam-515	112	18	these	these	DET
ejpam-515	112	19	conditions	condition	NOUN
ejpam-515	112	20	are	be	AUX
ejpam-515	112	21	not	not	PART
ejpam-515	112	22	completely	completely	ADV
ejpam-515	112	23	independent	independent	ADJ
ejpam-515	112	24	of	of	ADP
ejpam-515	112	25	each	each	DET
ejpam-515	112	26	other	other	ADJ
ejpam-515	112	27	.	.	PUNCT
ejpam-515	113	1	firstly	firstly	ADV
ejpam-515	113	2	,	,	PUNCT
ejpam-515	113	3	one	one	PRON
ejpam-515	113	4	can	can	AUX
ejpam-515	113	5	see	see	VERB
ejpam-515	113	6	that	that	SCONJ
ejpam-515	113	7	both	both	PRON
ejpam-515	113	8	(	(	PUNCT
ejpam-515	113	9	c.1	c.1	NOUN
ejpam-515	113	10	)	)	PUNCT
ejpam-515	113	11	and	and	CCONJ
ejpam-515	113	12	(	(	PUNCT
ejpam-515	113	13	c.2	c.2	X
ejpam-515	113	14	)	)	PUNCT
ejpam-515	113	15	are	be	AUX
ejpam-515	113	16	implied	imply	VERB
ejpam-515	113	17	from	from	ADP
ejpam-515	113	18	b1n	b1n	PROPN
ejpam-515	113	19	≤	≤	ADV
ejpam-515	113	20	λ1{in(β0	λ1{in(β0	NOUN
ejpam-515	113	21	)	)	PUNCT
ejpam-515	113	22	}	}	PUNCT
ejpam-515	113	23	≤	≤	NOUN
ejpam-515	113	24	λp{in(β0	λp{in(β0	NUM
ejpam-515	113	25	)	)	PUNCT
ejpam-515	113	26	}	}	PUNCT
ejpam-515	113	27	≤	≤	NUM
ejpam-515	113	28	b2n	b2n	NOUN
ejpam-515	113	29	.	.	PUNCT
ejpam-515	114	1	secondly	secondly	ADV
ejpam-515	114	2	,	,	PUNCT
ejpam-515	114	3	conditions	condition	NOUN
ejpam-515	114	4	(	(	PUNCT
ejpam-515	114	5	c.1	c.1	NOUN
ejpam-515	114	6	)	)	PUNCT
ejpam-515	114	7	,	,	PUNCT
ejpam-515	114	8	(	(	PUNCT
ejpam-515	114	9	c.2	c.2	NOUN
ejpam-515	114	10	)	)	PUNCT
ejpam-515	114	11	and	and	CCONJ
ejpam-515	114	12	(	(	PUNCT
ejpam-515	114	13	c.7	c.7	NOUN
ejpam-515	114	14	)	)	PUNCT
ejpam-515	114	15	together	together	ADV
ejpam-515	114	16	suggest	suggest	VERB
ejpam-515	114	17	that	that	SCONJ
ejpam-515	114	18	all	all	DET
ejpam-515	114	19	the	the	DET
ejpam-515	114	20	eigenvalues	eigenvalue	NOUN
ejpam-515	114	21	of	of	ADP
ejpam-515	114	22	in(β0	in(β0	NOUN
ejpam-515	114	23	)	)	PUNCT
ejpam-515	114	24	and	and	CCONJ
ejpam-515	114	25	xnmnxn	xnmnxn	NOUN
ejpam-515	114	26	are	be	AUX
ejpam-515	114	27	of	of	ADP
ejpam-515	114	28	order	order	NOUN
ejpam-515	114	29	o(n	o(n	NUM
ejpam-515	114	30	)	)	PUNCT
ejpam-515	114	31	.	.	PUNCT
ejpam-515	115	1	thirdly	thirdly	ADV
ejpam-515	115	2	,	,	PUNCT
ejpam-515	115	3	conditions	condition	NOUN
ejpam-515	115	4	(	(	PUNCT
ejpam-515	115	5	c.3	c.3	NOUN
ejpam-515	115	6	)	)	PUNCT
ejpam-515	115	7	to	to	PART
ejpam-515	115	8	(	(	PUNCT
ejpam-515	115	9	c.6	c.6	X
ejpam-515	115	10	)	)	PUNCT
ejpam-515	115	11	together	together	ADV
ejpam-515	115	12	with	with	ADP
ejpam-515	115	13	(	(	PUNCT
ejpam-515	115	14	c.2	c.2	X
ejpam-515	115	15	)	)	PUNCT
ejpam-515	115	16	indicate	indicate	VERB
ejpam-515	115	17	ξn	ξn	PROPN
ejpam-515	115	18	p	p	PROPN
ejpam-515	115	19	log	log	NOUN
ejpam-515	115	20	log	log	NOUN
ejpam-515	115	21	n→	n→	ADV
ejpam-515	115	22	0	0	NUM
ejpam-515	115	23	,	,	PUNCT
ejpam-515	115	24	δn	δn	ADP
ejpam-515	115	25	p	p	NOUN
ejpam-515	115	26	log	log	NOUN
ejpam-515	115	27	n	n	CCONJ
ejpam-515	115	28	log	log	VERB
ejpam-515	115	29	log	log	NOUN
ejpam-515	115	30	n→	n→	ADV
ejpam-515	115	31	0	0	NUM
ejpam-515	115	32	,	,	PUNCT
ejpam-515	115	33	δn	δn	PROPN
ejpam-515	115	34	log	log	VERB
ejpam-515	115	35	n	n	CCONJ
ejpam-515	115	36	p	p	NOUN
ejpam-515	115	37	log	log	NOUN
ejpam-515	115	38	log	log	NOUN
ejpam-515	115	39	n→	n→	ADV
ejpam-515	115	40	0	0	NUM
ejpam-515	116	1	and	and	CCONJ
ejpam-515	116	2	ξn	ξn	PROPN
ejpam-515	116	3	log	log	NOUN
ejpam-515	116	4	n	n	PRON
ejpam-515	116	5	p	p	NOUN
ejpam-515	116	6	log	log	NOUN
ejpam-515	116	7	log	log	NOUN
ejpam-515	116	8	n→	n→	ADV
ejpam-515	116	9	0	0	NUM
ejpam-515	116	10	respectively	respectively	ADV
ejpam-515	116	11	.	.	PUNCT
ejpam-515	117	1	finally	finally	ADV
ejpam-515	117	2	,	,	PUNCT
ejpam-515	117	3	under	under	ADP
ejpam-515	117	4	condition	condition	NOUN
ejpam-515	117	5	(	(	PUNCT
ejpam-515	117	6	c.1	c.1	NOUN
ejpam-515	117	7	)	)	PUNCT
ejpam-515	117	8	,	,	PUNCT
ejpam-515	117	9	(	(	PUNCT
ejpam-515	117	10	c.3	c.3	NOUN
ejpam-515	117	11	)	)	PUNCT
ejpam-515	117	12	is	be	AUX
ejpam-515	117	13	implied	imply	VERB
ejpam-515	117	14	by	by	ADP
ejpam-515	117	15	(	(	PUNCT
ejpam-515	117	16	c.6	c.6	NOUN
ejpam-515	117	17	)	)	PUNCT
ejpam-515	117	18	;	;	PUNCT
ejpam-515	117	19	and	and	CCONJ
ejpam-515	117	20	(	(	PUNCT
ejpam-515	117	21	c.4	c.4	X
ejpam-515	117	22	)	)	PUNCT
ejpam-515	117	23	is	be	AUX
ejpam-515	117	24	implied	imply	VERB
ejpam-515	117	25	by	by	ADP
ejpam-515	117	26	(	(	PUNCT
ejpam-515	117	27	c.5	c.5	NOUN
ejpam-515	117	28	)	)	PUNCT
ejpam-515	117	29	.	.	PUNCT
ejpam-515	118	1	although	although	SCONJ
ejpam-515	118	2	conditions	condition	NOUN
ejpam-515	118	3	(	(	PUNCT
ejpam-515	118	4	c.1	c.1	NOUN
ejpam-515	118	5	)	)	PUNCT
ejpam-515	118	6	to	to	ADP
ejpam-515	118	7	(	(	PUNCT
ejpam-515	118	8	c.7	c.7	NOUN
ejpam-515	118	9	)	)	PUNCT
ejpam-515	118	10	may	may	AUX
ejpam-515	118	11	be	be	AUX
ejpam-515	118	12	simplified	simplify	VERB
ejpam-515	118	13	according	accord	VERB
ejpam-515	118	14	to	to	ADP
ejpam-515	118	15	the	the	DET
ejpam-515	118	16	preceding	precede	VERB
ejpam-515	118	17	discussion	discussion	NOUN
ejpam-515	118	18	,	,	PUNCT
ejpam-515	118	19	we	we	PRON
ejpam-515	118	20	prefer	prefer	VERB
ejpam-515	118	21	not	not	PART
ejpam-515	118	22	to	to	PART
ejpam-515	118	23	do	do	VERB
ejpam-515	118	24	so	so	ADV
ejpam-515	118	25	in	in	ADP
ejpam-515	118	26	order	order	NOUN
ejpam-515	118	27	to	to	PART
ejpam-515	118	28	clarify	clarify	VERB
ejpam-515	118	29	that	that	PRON
ejpam-515	118	30	to	to	ADP
ejpam-515	118	31	what	what	DET
ejpam-515	118	32	extent	extent	NOUN
ejpam-515	118	33	each	each	DET
ejpam-515	118	34	condition	condition	NOUN
ejpam-515	118	35	is	be	AUX
ejpam-515	118	36	required	require	VERB
ejpam-515	118	37	in	in	ADP
ejpam-515	118	38	the	the	DET
ejpam-515	118	39	proof	proof	NOUN
ejpam-515	118	40	.	.	PUNCT
ejpam-515	119	1	the	the	DET
ejpam-515	119	2	conditions	condition	NOUN
ejpam-515	119	3	(	(	PUNCT
ejpam-515	119	4	c.1	c.1	NOUN
ejpam-515	119	5	)	)	PUNCT
ejpam-515	119	6	to	to	ADP
ejpam-515	119	7	(	(	PUNCT
ejpam-515	119	8	c.7	c.7	NOUN
ejpam-515	119	9	)	)	PUNCT
ejpam-515	119	10	are	be	AUX
ejpam-515	119	11	essentially	essentially	ADV
ejpam-515	119	12	about	about	ADP
ejpam-515	119	13	the	the	DET
ejpam-515	119	14	behavior	behavior	NOUN
ejpam-515	119	15	of	of	ADP
ejpam-515	119	16	the	the	DET
ejpam-515	119	17	explanatory	explanatory	ADJ
ejpam-515	119	18	variables	variable	NOUN
ejpam-515	119	19	x.	x.	VERB
ejpam-515	119	20	roughly	roughly	ADV
ejpam-515	119	21	speaking	speak	VERB
ejpam-515	119	22	,	,	PUNCT
ejpam-515	119	23	they	they	PRON
ejpam-515	119	24	mean	mean	VERB
ejpam-515	119	25	most	most	ADJ
ejpam-515	119	26	of	of	ADP
ejpam-515	119	27	the	the	DET
ejpam-515	119	28	observations	observation	NOUN
ejpam-515	119	29	{	{	PUNCT
ejpam-515	119	30	x1	x1	PROPN
ejpam-515	119	31	,	,	PUNCT
ejpam-515	119	32	·	·	PUNCT
ejpam-515	119	33	·	·	PUNCT
ejpam-515	119	34	·	·	PUNCT
ejpam-515	119	35	,	,	PUNCT
ejpam-515	119	36	xn	xn	X
ejpam-515	119	37	}	}	PUNCT
ejpam-515	119	38	should	should	AUX
ejpam-515	119	39	be	be	AUX
ejpam-515	119	40	finite	finite	ADJ
ejpam-515	119	41	and	and	CCONJ
ejpam-515	119	42	stay	stay	VERB
ejpam-515	119	43	away	away	ADV
ejpam-515	119	44	from	from	ADP
ejpam-515	119	45	0	0	NUM
ejpam-515	119	46	;	;	PUNCT
ejpam-515	119	47	and	and	CCONJ
ejpam-515	119	48	if	if	SCONJ
ejpam-515	119	49	a	a	DET
ejpam-515	119	50	subsequence	subsequence	NOUN
ejpam-515	119	51	of	of	ADP
ejpam-515	119	52	{	{	PUNCT
ejpam-515	119	53	x1	x1	PROPN
ejpam-515	119	54	,	,	PUNCT
ejpam-515	119	55	·	·	PUNCT
ejpam-515	119	56	·	·	PUNCT
ejpam-515	119	57	·	·	PUNCT
ejpam-515	119	58	,	,	PUNCT
ejpam-515	119	59	xn	xn	X
ejpam-515	119	60	}	}	PUNCT
ejpam-515	119	61	diverges	diverge	VERB
ejpam-515	119	62	to	to	ADP
ejpam-515	119	63	infinity	infinity	NOUN
ejpam-515	119	64	,	,	PUNCT
ejpam-515	119	65	it	it	PRON
ejpam-515	119	66	should	should	AUX
ejpam-515	119	67	do	do	VERB
ejpam-515	119	68	so	so	ADV
ejpam-515	119	69	with	with	ADP
ejpam-515	119	70	an	an	DET
ejpam-515	119	71	appropriate	appropriate	ADJ
ejpam-515	119	72	rate	rate	NOUN
ejpam-515	119	73	.	.	PUNCT
ejpam-515	120	1	in	in	ADP
ejpam-515	120	2	fact	fact	NOUN
ejpam-515	120	3	,	,	PUNCT
ejpam-515	120	4	if	if	SCONJ
ejpam-515	120	5	we	we	PRON
ejpam-515	120	6	assume	assume	VERB
ejpam-515	120	7	x	x	PRON
ejpam-515	120	8	is	be	AUX
ejpam-515	120	9	a	a	DET
ejpam-515	120	10	random	random	ADJ
ejpam-515	120	11	vector	vector	NOUN
ejpam-515	120	12	and	and	CCONJ
ejpam-515	120	13	x1	x1	NUM
ejpam-515	120	14	,	,	PUNCT
ejpam-515	120	15	·	·	PUNCT
ejpam-515	120	16	·	·	PUNCT
ejpam-515	120	17	·	·	PUNCT
ejpam-515	120	18	,	,	PUNCT
ejpam-515	120	19	xn	xn	PROPN
ejpam-515	120	20	as	as	ADP
ejpam-515	120	21	i.i.d	i.i.d	PROPN
ejpam-515	120	22	.	.	PUNCT
ejpam-515	121	1	observations	observation	NOUN
ejpam-515	121	2	from	from	ADP
ejpam-515	121	3	x	x	PRON
ejpam-515	121	4	,	,	PUNCT
ejpam-515	121	5	then	then	ADV
ejpam-515	121	6	the	the	DET
ejpam-515	121	7	following	following	NOUN
ejpam-515	121	8	are	be	AUX
ejpam-515	121	9	sufficient	sufficient	ADJ
ejpam-515	121	10	for	for	ADP
ejpam-515	121	11	(	(	PUNCT
ejpam-515	121	12	c.1	c.1	NOUN
ejpam-515	121	13	)	)	PUNCT
ejpam-515	121	14	to	to	ADP
ejpam-515	121	15	(	(	PUNCT
ejpam-515	121	16	c.7	c.7	NOUN
ejpam-515	121	17	)	)	PUNCT
ejpam-515	121	18	to	to	PART
ejpam-515	121	19	hold	hold	VERB
ejpam-515	121	20	:	:	PUNCT
ejpam-515	121	21	(	(	PUNCT
ejpam-515	121	22	s.1	s.1	NUM
ejpam-515	121	23	)	)	PUNCT
ejpam-515	121	24	.	.	PUNCT
ejpam-515	122	1	the	the	DET
ejpam-515	122	2	moment	moment	NOUN
ejpam-515	122	3	generating	generate	VERB
ejpam-515	122	4	function	function	NOUN
ejpam-515	122	5	eext	eext	NOUN
ejpam-515	122	6	s	s	NOUN
ejpam-515	122	7	exists	exist	VERB
ejpam-515	122	8	for	for	ADP
ejpam-515	122	9	||s||	||s||	ADJ
ejpam-515	122	10	≤	≤	NOUN
ejpam-515	122	11	||β0||	||β0||	NOUN
ejpam-515	122	12	+	+	CCONJ
ejpam-515	122	13	s0	s0	NOUN
ejpam-515	122	14	for	for	ADP
ejpam-515	122	15	some	some	DET
ejpam-515	122	16	constant	constant	ADJ
ejpam-515	122	17	s0	s0	NOUN
ejpam-515	122	18	>	>	X
ejpam-515	122	19	0	0	X
ejpam-515	122	20	.	.	PUNCT
ejpam-515	123	1	this	this	PRON
ejpam-515	123	2	implies	imply	VERB
ejpam-515	123	3	that	that	SCONJ
ejpam-515	123	4	all	all	PRON
ejpam-515	123	5	of	of	ADP
ejpam-515	123	6	e(extβ0xtx)κ	e(extβ0xtx)κ	PROPN
ejpam-515	123	7	,	,	PUNCT
ejpam-515	123	8	e(extβ0−2b‖x‖xtx)κ	e(extβ0−2b‖x‖xtx)κ	PROPN
ejpam-515	123	9	and	and	CCONJ
ejpam-515	123	10	e(xtx)κ	e(xtx)κ	PROPN
ejpam-515	123	11	are	be	AUX
ejpam-515	123	12	finite	finite	ADJ
ejpam-515	123	13	for	for	ADP
ejpam-515	123	14	some	some	PRON
ejpam-515	123	15	κ	κ	NOUN
ejpam-515	123	16	>	>	X
ejpam-515	123	17	1	1	NUM
ejpam-515	123	18	.	.	PUNCT
ejpam-515	123	19	(	(	PUNCT
ejpam-515	123	20	s.2	s.2	NOUN
ejpam-515	123	21	)	)	PUNCT
ejpam-515	123	22	.	.	PUNCT
ejpam-515	124	1	p(xtv	p(xtv	PROPN
ejpam-515	124	2	6=	6=	ADP
ejpam-515	124	3	0	0	NUM
ejpam-515	124	4	)	)	PUNCT
ejpam-515	124	5	>	>	X
ejpam-515	124	6	0	0	PUNCT
ejpam-515	125	1	for	for	ADP
ejpam-515	125	2	all	all	DET
ejpam-515	125	3	v	v	NOUN
ejpam-515	125	4	6=	6=	PRON
ejpam-515	125	5	0	0	NUM
ejpam-515	125	6	in	in	ADP
ejpam-515	125	7	r	r	NOUN
ejpam-515	125	8	p	p	NOUN
ejpam-515	125	9	,	,	PUNCT
ejpam-515	125	10	which	which	PRON
ejpam-515	125	11	implies	imply	VERB
ejpam-515	125	12	eextβ0xxt	eextβ0xxt	NOUN
ejpam-515	125	13	,	,	PUNCT
ejpam-515	125	14	eextβ0−2b‖x‖xxt	eextβ0−2b‖x‖xxt	NOUN
ejpam-515	125	15	and	and	CCONJ
ejpam-515	125	16	exxt	exxt	NOUN
ejpam-515	125	17	are	be	AUX
ejpam-515	125	18	all	all	PRON
ejpam-515	125	19	positive	positive	ADJ
ejpam-515	125	20	definite	definite	NOUN
ejpam-515	125	21	.	.	PUNCT
ejpam-515	126	1	to	to	PART
ejpam-515	126	2	see	see	VERB
ejpam-515	126	3	the	the	DET
ejpam-515	126	4	sufficiency	sufficiency	NOUN
ejpam-515	126	5	of	of	ADP
ejpam-515	126	6	(	(	PUNCT
ejpam-515	126	7	s.1	s.1	PROPN
ejpam-515	126	8	)	)	PUNCT
ejpam-515	126	9	and	and	CCONJ
ejpam-515	126	10	(	(	PUNCT
ejpam-515	126	11	s.2	s.2	NOUN
ejpam-515	126	12	)	)	PUNCT
ejpam-515	126	13	,	,	PUNCT
ejpam-515	126	14	one	one	PRON
ejpam-515	126	15	can	can	AUX
ejpam-515	126	16	apply	apply	VERB
ejpam-515	126	17	the	the	DET
ejpam-515	126	18	strong	strong	ADJ
ejpam-515	126	19	law	law	NOUN
ejpam-515	126	20	of	of	ADP
ejpam-515	126	21	large	large	ADJ
ejpam-515	126	22	numbers	number	NOUN
ejpam-515	126	23	for	for	ADP
ejpam-515	126	24	the	the	DET
ejpam-515	126	25	i.i.d	i.i.d	NOUN
ejpam-515	126	26	.	.	PUNCT
ejpam-515	127	1	random	random	ADJ
ejpam-515	127	2	variables	variable	NOUN
ejpam-515	127	3	x1	x1	PROPN
ejpam-515	127	4	,	,	PUNCT
ejpam-515	127	5	·	·	PUNCT
ejpam-515	127	6	·	·	PUNCT
ejpam-515	127	7	·	·	PUNCT
ejpam-515	127	8	,	,	PUNCT
ejpam-515	127	9	xn	xn	PROPN
ejpam-515	127	10	,	,	PUNCT
ejpam-515	127	11	·	·	PUNCT
ejpam-515	127	12	·	·	PUNCT
ejpam-515	127	13	·	·	PUNCT
ejpam-515	128	1	under	under	ADP
ejpam-515	128	2	condition	condition	NOUN
ejpam-515	128	3	(	(	PUNCT
ejpam-515	128	4	s.1	s.1	PROPN
ejpam-515	128	5	)	)	PUNCT
ejpam-515	128	6	,	,	PUNCT
ejpam-515	128	7	which	which	PRON
ejpam-515	128	8	gives	give	VERB
ejpam-515	128	9	the	the	DET
ejpam-515	128	10	following	follow	VERB
ejpam-515	128	11	results	result	NOUN
ejpam-515	128	12	:	:	PUNCT
ejpam-515	128	13	1	1	NUM
ejpam-515	128	14	n	n	NOUN
ejpam-515	128	15	x	x	PROPN
ejpam-515	128	16	t	t	PROPN
ejpam-515	128	17	nunxn−	nunxn−	VERB
ejpam-515	128	18	eextβ0xxt	eextβ0xxt	PROPN
ejpam-515	128	19	a.s.→	a.s.→	ADJ
ejpam-515	128	20	0	0	NUM
ejpam-515	128	21	,	,	PUNCT
ejpam-515	128	22	1	1	NUM
ejpam-515	128	23	n	n	NOUN
ejpam-515	128	24	x	x	PROPN
ejpam-515	128	25	t	t	PROPN
ejpam-515	128	26	nmnxn−	nmnxn−	PROPN
ejpam-515	128	27	eextβ0−2b‖x‖xxt	eextβ0−2b‖x‖xxt	VERB
ejpam-515	128	28	a.s.→	a.s.→	ADJ
ejpam-515	128	29	0	0	NUM
ejpam-515	128	30	.	.	PUNCT
ejpam-515	129	1	these	these	DET
ejpam-515	129	2	results	result	VERB
ejpam-515	129	3	together	together	ADV
ejpam-515	129	4	with	with	ADP
ejpam-515	129	5	(	(	PUNCT
ejpam-515	129	6	s.2	s.2	NOUN
ejpam-515	129	7	)	)	PUNCT
ejpam-515	129	8	imply	imply	NOUN
ejpam-515	129	9	(	(	PUNCT
ejpam-515	129	10	c.1),(c.2	c.1),(c.2	PROPN
ejpam-515	129	11	)	)	PUNCT
ejpam-515	129	12	and	and	CCONJ
ejpam-515	129	13	(	(	PUNCT
ejpam-515	129	14	c.7	c.7	NOUN
ejpam-515	129	15	)	)	PUNCT
ejpam-515	129	16	.	.	PUNCT
ejpam-515	130	1	the	the	DET
ejpam-515	130	2	conditions	condition	NOUN
ejpam-515	130	3	(	(	PUNCT
ejpam-515	130	4	c.3	c.3	NOUN
ejpam-515	130	5	)	)	PUNCT
ejpam-515	130	6	to	to	PART
ejpam-515	130	7	(	(	PUNCT
ejpam-515	130	8	c.6	c.6	X
ejpam-515	130	9	)	)	PUNCT
ejpam-515	130	10	are	be	AUX
ejpam-515	130	11	implied	imply	VERB
ejpam-515	130	12	from	from	ADP
ejpam-515	130	13	(	(	PUNCT
ejpam-515	130	14	c.1	c.1	NOUN
ejpam-515	130	15	)	)	PUNCT
ejpam-515	130	16	,	,	PUNCT
ejpam-515	130	17	(	(	PUNCT
ejpam-515	130	18	c.2	c.2	NOUN
ejpam-515	130	19	)	)	PUNCT
ejpam-515	130	20	and	and	CCONJ
ejpam-515	130	21	the	the	DET
ejpam-515	130	22	fact	fact	NOUN
ejpam-515	130	23	that	that	SCONJ
ejpam-515	130	24	,	,	PUNCT
ejpam-515	130	25	under	under	ADP
ejpam-515	130	26	(	(	PUNCT
ejpam-515	130	27	s.1	s.1	NOUN
ejpam-515	130	28	)	)	PUNCT
ejpam-515	130	29	δ2(1+κ′	δ2(1+κ′	NOUN
ejpam-515	130	30	)	)	PUNCT
ejpam-515	130	31	n	n	PRON
ejpam-515	130	32	≤	≤	NOUN
ejpam-515	130	33	λ1{in(β0)}−1−κ′	λ1{in(β0)}−1−κ′	X
ejpam-515	130	34	n	n	CCONJ
ejpam-515	130	35	∑	∑	PUNCT
ejpam-515	130	36	j=1	j=1	PROPN
ejpam-515	130	37	e	e	X
ejpam-515	130	38	xt	xt	PROPN
ejpam-515	130	39	j	j	PROPN
ejpam-515	130	40	β0(1+κ	β0(1+κ	PROPN
ejpam-515	130	41	′	′	NUM
ejpam-515	130	42	)	)	PUNCT
ejpam-515	130	43	(	(	PUNCT
ejpam-515	130	44	xt	xt	PROPN
ejpam-515	130	45	jx	jx	PROPN
ejpam-515	130	46	j	j	PROPN
ejpam-515	130	47	)	)	PUNCT
ejpam-515	130	48	1+κ′	1+κ′	PROPN
ejpam-515	130	49	=	=	SYM
ejpam-515	130	50	o(n−κ	o(n−κ	PROPN
ejpam-515	130	51	′	′	NUM
ejpam-515	130	52	)	)	PUNCT
ejpam-515	131	1	a.s	a.s	PROPN
ejpam-515	131	2	.	.	PROPN
ejpam-515	131	3	and	and	CCONJ
ejpam-515	131	4	ξ2(1+κ′	ξ2(1+κ′	PROPN
ejpam-515	131	5	)	)	PUNCT
ejpam-515	131	6	n	n	CCONJ
ejpam-515	131	7	≤	≤	NOUN
ejpam-515	131	8	λ1{in(β0)}−1−κ′	λ1{in(β0)}−1−κ′	X
ejpam-515	131	9	n	n	CCONJ
ejpam-515	131	10	∑	∑	PROPN
ejpam-515	131	11	j=1	j=1	PROPN
ejpam-515	131	12	(	(	PUNCT
ejpam-515	131	13	xt	xt	PROPN
ejpam-515	131	14	jx	jx	PROPN
ejpam-515	131	15	j	j	PROPN
ejpam-515	131	16	)	)	PUNCT
ejpam-515	132	1	1+κ′	1+κ′	PROPN
ejpam-515	132	2	=	=	SYM
ejpam-515	132	3	o(n−κ	o(n−κ	PROPN
ejpam-515	132	4	′	′	NUM
ejpam-515	132	5	)	)	PUNCT
ejpam-515	133	1	a.s	a.s	PROPN
ejpam-515	133	2	.	.	PROPN
ejpam-515	133	3	for	for	ADP
ejpam-515	133	4	some	some	DET
ejpam-515	133	5	κ′	κ′	NOUN
ejpam-515	133	6	>	>	X
ejpam-515	133	7	0	0	X
ejpam-515	133	8	.	.	PUNCT
ejpam-515	134	1	in	in	ADP
ejpam-515	134	2	this	this	DET
ejpam-515	134	3	paper	paper	NOUN
ejpam-515	134	4	we	we	PRON
ejpam-515	134	5	will	will	AUX
ejpam-515	134	6	regard	regard	VERB
ejpam-515	134	7	the	the	DET
ejpam-515	134	8	observations	observation	NOUN
ejpam-515	134	9	x1	x1	PROPN
ejpam-515	134	10	,	,	PUNCT
ejpam-515	134	11	·	·	PUNCT
ejpam-515	134	12	·	·	PUNCT
ejpam-515	134	13	·	·	PUNCT
ejpam-515	134	14	,	,	PUNCT
ejpam-515	134	15	xn	xn	PUNCT
ejpam-515	134	16	as	as	ADP
ejpam-515	134	17	deterministic	deterministic	ADJ
ejpam-515	134	18	for	for	ADP
ejpam-515	134	19	simplicity	simplicity	NOUN
ejpam-515	134	20	of	of	ADP
ejpam-515	134	21	the	the	DET
ejpam-515	134	22	presentation	presentation	NOUN
ejpam-515	134	23	.	.	PUNCT
ejpam-515	135	1	there	there	PRON
ejpam-515	135	2	is	be	VERB
ejpam-515	135	3	no	no	DET
ejpam-515	135	4	essential	essential	ADJ
ejpam-515	135	5	complication	complication	NOUN
ejpam-515	135	6	with	with	ADP
ejpam-515	135	7	random	random	ADJ
ejpam-515	135	8	xi	xi	ADP
ejpam-515	135	9	’	'	PUNCT
ejpam-515	135	10	s.	s.	PROPN
ejpam-515	135	11	in	in	ADP
ejpam-515	135	12	this	this	DET
ejpam-515	135	13	paper	paper	NOUN
ejpam-515	135	14	we	we	PRON
ejpam-515	135	15	have	have	AUX
ejpam-515	135	16	obtained	obtain	VERB
ejpam-515	135	17	the	the	DET
ejpam-515	135	18	following	follow	VERB
ejpam-515	135	19	results	result	NOUN
ejpam-515	135	20	.	.	PUNCT
ejpam-515	136	1	theorem	theorem	NOUN
ejpam-515	136	2	1	1	NUM
ejpam-515	136	3	.	.	PUNCT
ejpam-515	136	4	suppose	suppose	VERB
ejpam-515	136	5	conditions	condition	NOUN
ejpam-515	136	6	(	(	PUNCT
ejpam-515	136	7	c	c	NOUN
ejpam-515	136	8	.1	.1	NUM
ejpam-515	136	9	)	)	PUNCT
ejpam-515	136	10	to	to	ADP
ejpam-515	136	11	(	(	PUNCT
ejpam-515	136	12	c	c	NOUN
ejpam-515	136	13	.6	.6	NUM
ejpam-515	136	14	)	)	PUNCT
ejpam-515	136	15	are	be	AUX
ejpam-515	136	16	satisfied	satisfied	ADJ
ejpam-515	136	17	.	.	PUNCT
ejpam-515	137	1	then	then	ADV
ejpam-515	137	2	for	for	ADP
ejpam-515	137	3	any	any	DET
ejpam-515	137	4	correct	correct	ADJ
ejpam-515	137	5	model	model	NOUN
ejpam-515	137	6	α	α	PROPN
ejpam-515	137	7	∈	∈	PROPN
ejpam-515	137	8	ac	ac	PROPN
ejpam-515	137	9	,	,	PUNCT
ejpam-515	137	10	||β̂(α)−	||β̂(α)−	PROPN
ejpam-515	137	11	β0(α)||=	β0(α)||=	ADJ
ejpam-515	138	1	o	o	NOUN
ejpam-515	138	2	(	(	PUNCT
ejpam-515	138	3	p	p	PROPN
ejpam-515	138	4	n−1	n−1	PROPN
ejpam-515	138	5	log	log	NOUN
ejpam-515	138	6	log	log	NOUN
ejpam-515	138	7	n	n	CCONJ
ejpam-515	138	8	)	)	PUNCT
ejpam-515	138	9	a.s	a.s	PROPN
ejpam-515	138	10	..	..	PUNCT
ejpam-515	138	11	(	(	PUNCT
ejpam-515	138	12	4	4	X
ejpam-515	138	13	)	)	PUNCT
ejpam-515	138	14	g.	g.	PROPN
ejpam-515	138	15	qian	qian	PROPN
ejpam-515	138	16	/	/	SYM
ejpam-515	138	17	eur	eur	PROPN
ejpam-515	138	18	.	.	PUNCT
ejpam-515	139	1	j.	j.	PROPN
ejpam-515	139	2	pure	pure	PROPN
ejpam-515	139	3	appl	appl	PROPN
ejpam-515	139	4	.	.	PROPN
ejpam-515	139	5	math	math	PROPN
ejpam-515	139	6	,	,	PUNCT
ejpam-515	139	7	3	3	NUM
ejpam-515	139	8	(	(	PUNCT
ejpam-515	139	9	2010	2010	NUM
ejpam-515	139	10	)	)	PUNCT
ejpam-515	139	11	,	,	PUNCT
ejpam-515	139	12	417	417	NUM
ejpam-515	139	13	-	-	SYM
ejpam-515	139	14	434	434	NUM
ejpam-515	139	15	422	422	NUM
ejpam-515	139	16	further	far	ADV
ejpam-515	139	17	,	,	PUNCT
ejpam-515	139	18	there	there	PRON
ejpam-515	139	19	exists	exist	VERB
ejpam-515	139	20	a	a	DET
ejpam-515	139	21	constant	constant	ADJ
ejpam-515	139	22	d	d	X
ejpam-515	139	23	>	>	X
ejpam-515	139	24	0	0	NUM
ejpam-515	139	25	such	such	ADJ
ejpam-515	139	26	that	that	PRON
ejpam-515	139	27	for	for	ADP
ejpam-515	139	28	any	any	DET
ejpam-515	139	29	α	α	PRON
ejpam-515	139	30	∈ac	∈ac	ADJ
ejpam-515	139	31	lim	lim	NOUN
ejpam-515	139	32	sup	sup	NOUN
ejpam-515	139	33	n→∞	n→∞	NUM
ejpam-515	139	34	||β̂(α)−β0(α)||	||β̂(α)−β0(α)||	PROPN
ejpam-515	139	35	p	p	DET
ejpam-515	139	36	n−1	n−1	PROPN
ejpam-515	139	37	log	log	NOUN
ejpam-515	139	38	log	log	NOUN
ejpam-515	139	39	n	n	NOUN
ejpam-515	139	40	=	=	SYM
ejpam-515	140	1	d	d	X
ejpam-515	140	2	a.s	a.s	PROPN
ejpam-515	140	3	..	..	PUNCT
ejpam-515	140	4	(	(	PUNCT
ejpam-515	140	5	5	5	NUM
ejpam-515	140	6	)	)	PUNCT
ejpam-515	140	7	hence	hence	ADV
ejpam-515	140	8	the	the	DET
ejpam-515	140	9	m	m	PROPN
ejpam-515	140	10	le	le	X
ejpam-515	140	11	β̂(α	β̂(α	PUNCT
ejpam-515	140	12	)	)	PUNCT
ejpam-515	140	13	follows	follow	VERB
ejpam-515	140	14	the	the	DET
ejpam-515	140	15	law	law	NOUN
ejpam-515	140	16	of	of	ADP
ejpam-515	140	17	iterated	iterated	ADJ
ejpam-515	140	18	logarithm	logarithm	NOUN
ejpam-515	140	19	.	.	PUNCT
ejpam-515	141	1	theorem	theorem	NOUN
ejpam-515	141	2	2	2	NUM
ejpam-515	141	3	.	.	PUNCT
ejpam-515	142	1	under	under	ADP
ejpam-515	142	2	conditions	condition	NOUN
ejpam-515	142	3	(	(	PUNCT
ejpam-515	142	4	c	c	NOUN
ejpam-515	142	5	.1	.1	NUM
ejpam-515	142	6	)	)	PUNCT
ejpam-515	142	7	to	to	ADP
ejpam-515	142	8	(	(	PUNCT
ejpam-515	142	9	c	c	NOUN
ejpam-515	142	10	.6	.6	NUM
ejpam-515	142	11	)	)	PUNCT
ejpam-515	142	12	,	,	PUNCT
ejpam-515	142	13	for	for	ADP
ejpam-515	142	14	any	any	DET
ejpam-515	142	15	correct	correct	ADJ
ejpam-515	142	16	model	model	NOUN
ejpam-515	142	17	α	α	PRON
ejpam-515	142	18	∈ac	∈ac	ADJ
ejpam-515	142	19	,	,	PUNCT
ejpam-515	142	20	0≤	0≤	SYM
ejpam-515	142	21	ℓ(β̂(α)|yn	ℓ(β̂(α)|yn	PROPN
ejpam-515	142	22	,	,	PUNCT
ejpam-515	142	23	xnα)−	xnα)−	PROPN
ejpam-515	142	24	ℓ(β0|yn	ℓ(β0|yn	PROPN
ejpam-515	142	25	,	,	PUNCT
ejpam-515	142	26	xn	xn	PROPN
ejpam-515	142	27	)	)	PUNCT
ejpam-515	142	28	=	=	NOUN
ejpam-515	142	29	o(log	o(log	PROPN
ejpam-515	142	30	log	log	PROPN
ejpam-515	142	31	n	n	CCONJ
ejpam-515	142	32	)	)	PUNCT
ejpam-515	142	33	a.s	a.s	PROPN
ejpam-515	142	34	..	..	PUNCT
ejpam-515	142	35	(	(	PUNCT
ejpam-515	142	36	6	6	NUM
ejpam-515	142	37	)	)	PUNCT
ejpam-515	142	38	theorem	theorem	NOUN
ejpam-515	142	39	3	3	NUM
ejpam-515	142	40	.	.	PUNCT
ejpam-515	143	1	under	under	ADP
ejpam-515	143	2	conditions	condition	NOUN
ejpam-515	143	3	(	(	PUNCT
ejpam-515	143	4	c	c	NOUN
ejpam-515	143	5	.1	.1	NUM
ejpam-515	143	6	)	)	PUNCT
ejpam-515	143	7	to	to	ADP
ejpam-515	143	8	(	(	PUNCT
ejpam-515	143	9	c	c	PROPN
ejpam-515	143	10	.7	.7	PROPN
ejpam-515	143	11	)	)	PUNCT
ejpam-515	143	12	,	,	PUNCT
ejpam-515	143	13	for	for	ADP
ejpam-515	143	14	any	any	DET
ejpam-515	143	15	incorrect	incorrect	ADJ
ejpam-515	143	16	model	model	NOUN
ejpam-515	143	17	α	α	X
ejpam-515	143	18	∈	∈	PROPN
ejpam-515	143	19	aw	aw	INTJ
ejpam-515	143	20	,	,	PUNCT
ejpam-515	143	21	we	we	PRON
ejpam-515	143	22	have	have	VERB
ejpam-515	143	23	lim	lim	PROPN
ejpam-515	143	24	sup	sup	VERB
ejpam-515	143	25	n→∞	n→∞	NUM
ejpam-515	143	26	1	1	NUM
ejpam-515	143	27	n	n	PROPN
ejpam-515	143	28	{	{	PUNCT
ejpam-515	143	29	ℓ(β̂(α)|yn	ℓ(β̂(α)|yn	PROPN
ejpam-515	143	30	,	,	PUNCT
ejpam-515	143	31	xnα)−	xnα)−	PROPN
ejpam-515	143	32	ℓ(β0|yn	ℓ(β0|yn	PROPN
ejpam-515	143	33	,	,	PUNCT
ejpam-515	143	34	xn	xn	PROPN
ejpam-515	143	35	)	)	PUNCT
ejpam-515	143	36	}	}	PUNCT
ejpam-515	143	37	<	<	X
ejpam-515	143	38	0	0	PUNCT
ejpam-515	144	1	a.s	a.s	PROPN
ejpam-515	144	2	..	..	PUNCT
ejpam-515	144	3	(	(	PUNCT
ejpam-515	144	4	7	7	NUM
ejpam-515	144	5	)	)	PUNCT
ejpam-515	144	6	from	from	ADP
ejpam-515	144	7	theorems	theorem	NOUN
ejpam-515	144	8	2	2	NUM
ejpam-515	144	9	and	and	CCONJ
ejpam-515	144	10	3	3	NUM
ejpam-515	144	11	we	we	PRON
ejpam-515	144	12	know	know	VERB
ejpam-515	144	13	that	that	SCONJ
ejpam-515	144	14	the	the	DET
ejpam-515	144	15	maximum	maximum	ADJ
ejpam-515	144	16	log	log	NOUN
ejpam-515	144	17	-	-	PUNCT
ejpam-515	144	18	likelihood	likelihood	NOUN
ejpam-515	144	19	of	of	ADP
ejpam-515	144	20	any	any	DET
ejpam-515	144	21	correct	correct	ADJ
ejpam-515	144	22	model	model	NOUN
ejpam-515	144	23	is	be	AUX
ejpam-515	144	24	almost	almost	ADV
ejpam-515	144	25	surely	surely	ADV
ejpam-515	144	26	greater	great	ADJ
ejpam-515	144	27	than	than	ADP
ejpam-515	144	28	the	the	DET
ejpam-515	144	29	unknown	unknown	ADJ
ejpam-515	144	30	true	true	ADJ
ejpam-515	144	31	log	log	NOUN
ejpam-515	144	32	-	-	PUNCT
ejpam-515	144	33	likelihood	likelihood	NOUN
ejpam-515	144	34	of	of	ADP
ejpam-515	144	35	the	the	DET
ejpam-515	144	36	full	full	ADJ
ejpam-515	144	37	model	model	NOUN
ejpam-515	144	38	by	by	ADP
ejpam-515	144	39	an	an	DET
ejpam-515	144	40	amount	amount	NOUN
ejpam-515	144	41	bounded	bound	VERB
ejpam-515	144	42	by	by	ADP
ejpam-515	144	43	|o(log	|o(log	PUNCT
ejpam-515	144	44	log	log	PROPN
ejpam-515	144	45	n)|	n)|	PROPN
ejpam-515	144	46	.	.	PUNCT
ejpam-515	145	1	on	on	ADP
ejpam-515	145	2	the	the	DET
ejpam-515	145	3	other	other	ADJ
ejpam-515	145	4	hand	hand	NOUN
ejpam-515	145	5	,	,	PUNCT
ejpam-515	145	6	the	the	DET
ejpam-515	145	7	maximum	maximum	ADJ
ejpam-515	145	8	log	log	NOUN
ejpam-515	145	9	-	-	PUNCT
ejpam-515	145	10	likelihood	likelihood	NOUN
ejpam-515	145	11	of	of	ADP
ejpam-515	145	12	any	any	DET
ejpam-515	145	13	incorrect	incorrect	ADJ
ejpam-515	145	14	model	model	NOUN
ejpam-515	145	15	in	in	ADP
ejpam-515	145	16	aw	aw	INTJ
ejpam-515	145	17	is	be	AUX
ejpam-515	145	18	almost	almost	ADV
ejpam-515	145	19	surely	surely	ADV
ejpam-515	145	20	smaller	small	ADJ
ejpam-515	145	21	than	than	ADP
ejpam-515	145	22	the	the	DET
ejpam-515	145	23	true	true	ADJ
ejpam-515	145	24	log	log	NOUN
ejpam-515	145	25	-	-	PUNCT
ejpam-515	145	26	likelihood	likelihood	NOUN
ejpam-515	145	27	of	of	ADP
ejpam-515	145	28	the	the	DET
ejpam-515	145	29	full	full	ADJ
ejpam-515	145	30	model	model	NOUN
ejpam-515	145	31	by	by	ADP
ejpam-515	145	32	an	an	DET
ejpam-515	145	33	amount	amount	NOUN
ejpam-515	145	34	greater	great	ADJ
ejpam-515	145	35	than	than	ADP
ejpam-515	145	36	τn	τn	VERB
ejpam-515	145	37	with	with	ADP
ejpam-515	145	38	τ	τ	PROPN
ejpam-515	145	39	>	>	X
ejpam-515	145	40	0	0	PUNCT
ejpam-515	146	1	when	when	SCONJ
ejpam-515	146	2	n	n	PRON
ejpam-515	146	3	is	be	AUX
ejpam-515	146	4	sufficiently	sufficiently	ADV
ejpam-515	146	5	large	large	ADJ
ejpam-515	146	6	.	.	PUNCT
ejpam-515	147	1	therefore	therefore	ADV
ejpam-515	147	2	,	,	PUNCT
ejpam-515	147	3	if	if	SCONJ
ejpam-515	147	4	we	we	PRON
ejpam-515	147	5	use	use	VERB
ejpam-515	147	6	a	a	DET
ejpam-515	147	7	penalised	penalise	VERB
ejpam-515	147	8	log	log	NOUN
ejpam-515	147	9	-	-	PUNCT
ejpam-515	147	10	likelihood	likelihood	NOUN
ejpam-515	147	11	based	base	VERB
ejpam-515	147	12	criterion	criterion	NOUN
ejpam-515	147	13	of	of	ADP
ejpam-515	147	14	form	form	NOUN
ejpam-515	147	15	(	(	PUNCT
ejpam-515	147	16	3	3	NUM
ejpam-515	147	17	)	)	PUNCT
ejpam-515	147	18	for	for	ADP
ejpam-515	147	19	model	model	NOUN
ejpam-515	147	20	selection	selection	NOUN
ejpam-515	147	21	,	,	PUNCT
ejpam-515	147	22	we	we	PRON
ejpam-515	147	23	will	will	AUX
ejpam-515	147	24	almost	almost	ADV
ejpam-515	147	25	surely	surely	ADV
ejpam-515	147	26	select	select	VERB
ejpam-515	147	27	the	the	DET
ejpam-515	147	28	simplest	simple	ADJ
ejpam-515	147	29	correct	correct	ADJ
ejpam-515	147	30	model	model	NOUN
ejpam-515	147	31	inac	inac	PROPN
ejpam-515	147	32	if	if	SCONJ
ejpam-515	147	33	the	the	DET
ejpam-515	147	34	penalty	penalty	NOUN
ejpam-515	147	35	term	term	NOUN
ejpam-515	147	36	c(n	c(n	PROPN
ejpam-515	147	37	,	,	PUNCT
ejpam-515	147	38	β̂(α	β̂(α	PRON
ejpam-515	147	39	)	)	PUNCT
ejpam-515	147	40	)	)	PUNCT
ejpam-515	147	41	is	be	AUX
ejpam-515	147	42	an	an	DET
ejpam-515	147	43	increasing	increase	VERB
ejpam-515	147	44	function	function	NOUN
ejpam-515	147	45	of	of	ADP
ejpam-515	147	46	the	the	DET
ejpam-515	147	47	model	model	NOUN
ejpam-515	147	48	dimension	dimension	NOUN
ejpam-515	147	49	pα	pα	INTJ
ejpam-515	147	50	and	and	CCONJ
ejpam-515	147	51	is	be	AUX
ejpam-515	147	52	of	of	ADP
ejpam-515	147	53	an	an	DET
ejpam-515	147	54	order	order	NOUN
ejpam-515	147	55	in	in	ADP
ejpam-515	147	56	between	between	ADP
ejpam-515	147	57	o(log	o(log	PROPN
ejpam-515	147	58	log	log	PROPN
ejpam-515	147	59	n	n	CCONJ
ejpam-515	147	60	)	)	PUNCT
ejpam-515	147	61	and	and	CCONJ
ejpam-515	147	62	o(n	o(n	NUM
ejpam-515	147	63	)	)	PUNCT
ejpam-515	147	64	.	.	PUNCT
ejpam-515	148	1	we	we	PRON
ejpam-515	148	2	call	call	VERB
ejpam-515	148	3	a	a	DET
ejpam-515	148	4	model	model	NOUN
ejpam-515	148	5	selection	selection	NOUN
ejpam-515	148	6	criterion	criterion	NOUN
ejpam-515	148	7	strongly	strongly	ADV
ejpam-515	148	8	consistent	consistent	ADJ
ejpam-515	148	9	if	if	SCONJ
ejpam-515	148	10	it	it	PRON
ejpam-515	148	11	selects	select	VERB
ejpam-515	148	12	the	the	DET
ejpam-515	148	13	simplest	simple	ADJ
ejpam-515	148	14	correct	correct	ADJ
ejpam-515	148	15	model	model	NOUN
ejpam-515	148	16	almost	almost	ADV
ejpam-515	148	17	surely	surely	ADV
ejpam-515	148	18	;	;	PUNCT
ejpam-515	148	19	and	and	CCONJ
ejpam-515	148	20	consistent	consistent	ADJ
ejpam-515	148	21	if	if	SCONJ
ejpam-515	148	22	almost	almost	ADV
ejpam-515	148	23	surely	surely	ADV
ejpam-515	148	24	it	it	PRON
ejpam-515	148	25	only	only	ADV
ejpam-515	148	26	selects	select	VERB
ejpam-515	148	27	one	one	NUM
ejpam-515	148	28	of	of	ADP
ejpam-515	148	29	the	the	DET
ejpam-515	148	30	correct	correct	ADJ
ejpam-515	148	31	models	model	NOUN
ejpam-515	148	32	.	.	PUNCT
ejpam-515	149	1	from	from	ADP
ejpam-515	149	2	this	this	DET
ejpam-515	149	3	discussion	discussion	NOUN
ejpam-515	149	4	we	we	PRON
ejpam-515	149	5	have	have	VERB
ejpam-515	149	6	the	the	DET
ejpam-515	149	7	following	following	NOUN
ejpam-515	149	8	:	:	PUNCT
ejpam-515	149	9	theorem	theorem	NOUN
ejpam-515	149	10	4	4	NUM
ejpam-515	149	11	.	.	X
ejpam-515	150	1	for	for	ADP
ejpam-515	150	2	a	a	DET
ejpam-515	150	3	poisson	poisson	NOUN
ejpam-515	150	4	regression	regression	NOUN
ejpam-515	150	5	model	model	NOUN
ejpam-515	150	6	satisfying	satisfy	VERB
ejpam-515	150	7	conditions	condition	NOUN
ejpam-515	150	8	(	(	PUNCT
ejpam-515	150	9	c	c	NOUN
ejpam-515	150	10	.1	.1	NUM
ejpam-515	150	11	)	)	PUNCT
ejpam-515	150	12	to	to	ADP
ejpam-515	150	13	(	(	PUNCT
ejpam-515	150	14	c	c	PROPN
ejpam-515	150	15	.7	.7	PROPN
ejpam-515	150	16	)	)	PUNCT
ejpam-515	150	17	,	,	PUNCT
ejpam-515	150	18	both	both	DET
ejpam-515	150	19	model	model	NOUN
ejpam-515	150	20	selection	selection	NOUN
ejpam-515	150	21	criteria	criterion	NOUN
ejpam-515	150	22	bic	bic	PROPN
ejpam-515	150	23	and	and	CCONJ
ejpam-515	150	24	scc	scc	PROPN
ejpam-515	150	25	are	be	AUX
ejpam-515	150	26	strongly	strongly	ADV
ejpam-515	150	27	consistent	consistent	ADJ
ejpam-515	150	28	,	,	PUNCT
ejpam-515	150	29	while	while	SCONJ
ejpam-515	150	30	aic	aic	PROPN
ejpam-515	150	31	is	be	AUX
ejpam-515	150	32	consistent	consistent	ADJ
ejpam-515	150	33	but	but	CCONJ
ejpam-515	150	34	not	not	PART
ejpam-515	150	35	necessarily	necessarily	ADV
ejpam-515	150	36	strongly	strongly	ADV
ejpam-515	150	37	consistent	consistent	ADJ
ejpam-515	150	38	.	.	PUNCT
ejpam-515	151	1	proof	proof	NOUN
ejpam-515	151	2	.	.	PUNCT
ejpam-515	152	1	as	as	SCONJ
ejpam-515	152	2	the	the	DET
ejpam-515	152	3	fisher	fisher	PROPN
ejpam-515	152	4	information	information	NOUN
ejpam-515	152	5	’s	’s	PART
ejpam-515	152	6	determinant	determinant	ADJ
ejpam-515	152	7	|i(β(α))|	|i(β(α))|	PROPN
ejpam-515	152	8	is	be	AUX
ejpam-515	152	9	typically	typically	ADV
ejpam-515	152	10	of	of	ADP
ejpam-515	152	11	order	order	NOUN
ejpam-515	152	12	o(npα	o(npα	ADV
ejpam-515	152	13	)	)	PUNCT
ejpam-515	152	14	,	,	PUNCT
ejpam-515	152	15	both	both	CCONJ
ejpam-515	152	16	the	the	DET
ejpam-515	152	17	penalty	penalty	NOUN
ejpam-515	152	18	terms	term	NOUN
ejpam-515	152	19	of	of	ADP
ejpam-515	152	20	scc	scc	NOUN
ejpam-515	152	21	and	and	CCONJ
ejpam-515	152	22	bic	bic	PROPN
ejpam-515	152	23	are	be	AUX
ejpam-515	152	24	increasing	increase	VERB
ejpam-515	152	25	functions	function	NOUN
ejpam-515	152	26	of	of	ADP
ejpam-515	152	27	the	the	DET
ejpam-515	152	28	model	model	NOUN
ejpam-515	152	29	dimension	dimension	NOUN
ejpam-515	152	30	pα	pα	INTJ
ejpam-515	152	31	and	and	CCONJ
ejpam-515	152	32	are	be	AUX
ejpam-515	152	33	of	of	ADP
ejpam-515	152	34	order	order	NOUN
ejpam-515	152	35	o(log	o(log	PROPN
ejpam-515	152	36	n	n	CCONJ
ejpam-515	152	37	)	)	PUNCT
ejpam-515	152	38	,	,	PUNCT
ejpam-515	152	39	it	it	PRON
ejpam-515	152	40	follows	follow	VERB
ejpam-515	152	41	from	from	ADP
ejpam-515	152	42	theorem	theorem	ADJ
ejpam-515	152	43	2	2	NUM
ejpam-515	152	44	and	and	CCONJ
ejpam-515	152	45	theorem	theorem	VERB
ejpam-515	152	46	3	3	NUM
ejpam-515	152	47	that	that	SCONJ
ejpam-515	152	48	both	both	CCONJ
ejpam-515	152	49	scc	scc	NOUN
ejpam-515	152	50	and	and	CCONJ
ejpam-515	152	51	bic	bic	PROPN
ejpam-515	152	52	are	be	AUX
ejpam-515	152	53	strongly	strongly	ADV
ejpam-515	152	54	consistent	consistent	ADJ
ejpam-515	152	55	.	.	PUNCT
ejpam-515	153	1	aic	aic	PROPN
ejpam-515	153	2	is	be	AUX
ejpam-515	153	3	not	not	PART
ejpam-515	153	4	necessarily	necessarily	ADV
ejpam-515	153	5	strongly	strongly	ADV
ejpam-515	153	6	consistent	consistent	ADJ
ejpam-515	153	7	because	because	SCONJ
ejpam-515	153	8	its	its	PRON
ejpam-515	153	9	penalty	penalty	NOUN
ejpam-515	153	10	term	term	NOUN
ejpam-515	153	11	is	be	AUX
ejpam-515	153	12	of	of	ADP
ejpam-515	153	13	order	order	NOUN
ejpam-515	153	14	o(1	o(1	NOUN
ejpam-515	153	15	)	)	PUNCT
ejpam-515	153	16	.	.	PUNCT
ejpam-515	154	1	but	but	CCONJ
ejpam-515	154	2	aic	aic	PROPN
ejpam-515	154	3	is	be	AUX
ejpam-515	154	4	clearly	clearly	ADV
ejpam-515	154	5	consistent	consistent	ADJ
ejpam-515	154	6	because	because	SCONJ
ejpam-515	154	7	its	its	PRON
ejpam-515	154	8	criterion	criterion	NOUN
ejpam-515	154	9	value	value	NOUN
ejpam-515	154	10	for	for	ADP
ejpam-515	154	11	a	a	DET
ejpam-515	154	12	correct	correct	ADJ
ejpam-515	154	13	model	model	NOUN
ejpam-515	154	14	is	be	AUX
ejpam-515	154	15	almost	almost	ADV
ejpam-515	154	16	surely	surely	ADV
ejpam-515	154	17	smaller	small	ADJ
ejpam-515	154	18	than	than	ADP
ejpam-515	154	19	that	that	PRON
ejpam-515	154	20	for	for	ADP
ejpam-515	154	21	any	any	DET
ejpam-515	154	22	incorrect	incorrect	ADJ
ejpam-515	154	23	model	model	NOUN
ejpam-515	154	24	by	by	ADP
ejpam-515	154	25	an	an	DET
ejpam-515	154	26	amount	amount	NOUN
ejpam-515	154	27	greater	great	ADJ
ejpam-515	154	28	than	than	ADP
ejpam-515	154	29	τn	τn	ADP
ejpam-515	154	30	when	when	SCONJ
ejpam-515	154	31	n	n	PRON
ejpam-515	154	32	is	be	AUX
ejpam-515	154	33	sufficiently	sufficiently	ADV
ejpam-515	154	34	large	large	ADJ
ejpam-515	154	35	.	.	PUNCT
ejpam-515	155	1	the	the	DET
ejpam-515	155	2	proof	proof	NOUN
ejpam-515	155	3	of	of	ADP
ejpam-515	155	4	theorems	theorem	NOUN
ejpam-515	155	5	1	1	NUM
ejpam-515	155	6	to	to	PART
ejpam-515	155	7	3	3	NUM
ejpam-515	155	8	will	will	AUX
ejpam-515	155	9	be	be	AUX
ejpam-515	155	10	the	the	DET
ejpam-515	155	11	focus	focus	NOUN
ejpam-515	155	12	of	of	ADP
ejpam-515	155	13	the	the	DET
ejpam-515	155	14	next	next	ADJ
ejpam-515	155	15	section	section	NOUN
ejpam-515	155	16	.	.	PUNCT
ejpam-515	156	1	4	4	X
ejpam-515	156	2	.	.	X
ejpam-515	156	3	proof	proof	NOUN
ejpam-515	156	4	of	of	ADP
ejpam-515	156	5	the	the	DET
ejpam-515	156	6	results	result	NOUN
ejpam-515	156	7	the	the	DET
ejpam-515	156	8	key	key	NOUN
ejpam-515	156	9	to	to	ADP
ejpam-515	156	10	proving	prove	VERB
ejpam-515	156	11	our	our	PRON
ejpam-515	156	12	main	main	ADJ
ejpam-515	156	13	results	result	NOUN
ejpam-515	156	14	lies	lie	VERB
ejpam-515	156	15	on	on	ADP
ejpam-515	156	16	the	the	DET
ejpam-515	156	17	convexity	convexity	NOUN
ejpam-515	156	18	and	and	CCONJ
ejpam-515	156	19	quadratic	quadratic	ADJ
ejpam-515	156	20	approximation	approximation	NOUN
ejpam-515	156	21	of	of	ADP
ejpam-515	156	22	the	the	DET
ejpam-515	156	23	negative	negative	ADJ
ejpam-515	156	24	log	log	NOUN
ejpam-515	156	25	-	-	PUNCT
ejpam-515	156	26	likelihood	likelihood	NOUN
ejpam-515	156	27	function	function	NOUN
ejpam-515	156	28	,	,	PUNCT
ejpam-515	156	29	the	the	DET
ejpam-515	156	30	normal	normal	ADJ
ejpam-515	156	31	and	and	CCONJ
ejpam-515	156	32	gamma	gamma	NOUN
ejpam-515	156	33	approximations	approximation	NOUN
ejpam-515	156	34	of	of	ADP
ejpam-515	156	35	the	the	DET
ejpam-515	156	36	poisson	poisson	PROPN
ejpam-515	156	37	g.	g.	PROPN
ejpam-515	156	38	qian	qian	PROPN
ejpam-515	156	39	/	/	SYM
ejpam-515	156	40	eur	eur	PROPN
ejpam-515	156	41	.	.	PUNCT
ejpam-515	157	1	j.	j.	PROPN
ejpam-515	157	2	pure	pure	PROPN
ejpam-515	157	3	appl	appl	PROPN
ejpam-515	157	4	.	.	PROPN
ejpam-515	157	5	math	math	PROPN
ejpam-515	157	6	,	,	PUNCT
ejpam-515	157	7	3	3	NUM
ejpam-515	157	8	(	(	PUNCT
ejpam-515	157	9	2010	2010	NUM
ejpam-515	157	10	)	)	PUNCT
ejpam-515	157	11	,	,	PUNCT
ejpam-515	157	12	417	417	NUM
ejpam-515	157	13	-	-	SYM
ejpam-515	157	14	434	434	NUM
ejpam-515	157	15	423	423	NUM
ejpam-515	157	16	probabilities	probability	NOUN
ejpam-515	157	17	,	,	PUNCT
ejpam-515	157	18	and	and	CCONJ
ejpam-515	157	19	the	the	DET
ejpam-515	157	20	law	law	NOUN
ejpam-515	157	21	of	of	ADP
ejpam-515	157	22	iterated	iterated	ADJ
ejpam-515	157	23	logarithm	logarithm	NOUN
ejpam-515	157	24	for	for	ADP
ejpam-515	157	25	independent	independent	ADJ
ejpam-515	157	26	random	random	ADJ
ejpam-515	157	27	variables	variable	NOUN
ejpam-515	157	28	.	.	PUNCT
ejpam-515	158	1	the	the	DET
ejpam-515	158	2	idea	idea	NOUN
ejpam-515	158	3	of	of	ADP
ejpam-515	158	4	using	use	VERB
ejpam-515	158	5	the	the	DET
ejpam-515	158	6	convexity	convexity	NOUN
ejpam-515	158	7	property	property	NOUN
ejpam-515	158	8	is	be	AUX
ejpam-515	158	9	broadly	broadly	ADV
ejpam-515	158	10	seen	see	VERB
ejpam-515	158	11	in	in	ADP
ejpam-515	158	12	establishing	establish	VERB
ejpam-515	158	13	asymptotic	asymptotic	ADJ
ejpam-515	158	14	representations	representation	NOUN
ejpam-515	158	15	of	of	ADP
ejpam-515	158	16	the	the	DET
ejpam-515	158	17	m	m	NOUN
ejpam-515	158	18	-	-	NOUN
ejpam-515	158	19	estimators	estimator	NOUN
ejpam-515	158	20	in	in	ADP
ejpam-515	158	21	linear	linear	PROPN
ejpam-515	158	22	models	model	NOUN
ejpam-515	158	23	,	,	PUNCT
ejpam-515	158	24	see	see	VERB
ejpam-515	159	1	e.g.	e.g.	ADV
ejpam-515	159	2	[	[	X
ejpam-515	159	3	15	15	NUM
ejpam-515	159	4	,	,	PUNCT
ejpam-515	159	5	20	20	NUM
ejpam-515	159	6	,	,	PUNCT
ejpam-515	159	7	12	12	NUM
ejpam-515	159	8	]	]	PUNCT
ejpam-515	159	9	among	among	ADP
ejpam-515	159	10	the	the	DET
ejpam-515	159	11	others	other	NOUN
ejpam-515	159	12	.	.	PUNCT
ejpam-515	160	1	by	by	ADP
ejpam-515	160	2	the	the	DET
ejpam-515	160	3	definition	definition	NOUN
ejpam-515	160	4	of	of	ADP
ejpam-515	160	5	ξn	ξn	NOUN
ejpam-515	160	6	and	and	CCONJ
ejpam-515	160	7	conditions	condition	NOUN
ejpam-515	160	8	(	(	PUNCT
ejpam-515	160	9	c.2	c.2	NOUN
ejpam-515	160	10	)	)	PUNCT
ejpam-515	160	11	and	and	CCONJ
ejpam-515	160	12	(	(	PUNCT
ejpam-515	160	13	c.3	c.3	X
ejpam-515	160	14	)	)	PUNCT
ejpam-515	160	15	it	it	PRON
ejpam-515	160	16	is	be	AUX
ejpam-515	160	17	easy	easy	ADJ
ejpam-515	160	18	to	to	PART
ejpam-515	160	19	see	see	VERB
ejpam-515	160	20	that	that	SCONJ
ejpam-515	160	21	there	there	PRON
ejpam-515	160	22	exists	exist	VERB
ejpam-515	160	23	a	a	DET
ejpam-515	160	24	sequence	sequence	NOUN
ejpam-515	160	25	of	of	ADP
ejpam-515	160	26	positive	positive	ADJ
ejpam-515	160	27	numbers	number	NOUN
ejpam-515	160	28	{	{	PUNCT
ejpam-515	160	29	τn	τn	AUX
ejpam-515	160	30	}	}	PUNCT
ejpam-515	160	31	satisfying	satisfy	VERB
ejpam-515	160	32	:	:	PUNCT
ejpam-515	160	33	τn	τn	VERB
ejpam-515	160	34	↑	↑	PROPN
ejpam-515	160	35	∞	∞	PROPN
ejpam-515	160	36	,	,	PUNCT
ejpam-515	160	37	τnξn	τnξn	VERB
ejpam-515	160	38	p	p	NOUN
ejpam-515	160	39	log	log	NOUN
ejpam-515	160	40	log	log	NOUN
ejpam-515	160	41	n→	n→	ADV
ejpam-515	160	42	0	0	NUM
ejpam-515	160	43	and	and	CCONJ
ejpam-515	160	44	τn	τn	ADP
ejpam-515	160	45	p	p	PRON
ejpam-515	160	46	n−1	n−1	PROPN
ejpam-515	160	47	log	log	NOUN
ejpam-515	160	48	log	log	NOUN
ejpam-515	160	49	n	n	CCONJ
ejpam-515	160	50	↓	↓	NOUN
ejpam-515	160	51	0	0	NUM
ejpam-515	160	52	.	.	PUNCT
ejpam-515	161	1	using	use	VERB
ejpam-515	161	2	τn	τn	ADP
ejpam-515	161	3	we	we	PRON
ejpam-515	161	4	introduce	introduce	VERB
ejpam-515	161	5	two	two	NUM
ejpam-515	161	6	sequences	sequence	NOUN
ejpam-515	161	7	of	of	ADP
ejpam-515	161	8	subsets	subset	NOUN
ejpam-515	161	9	:	:	PUNCT
ejpam-515	161	10	an	an	DET
ejpam-515	161	11	=	=	X
ejpam-515	161	12	{	{	PUNCT
ejpam-515	161	13	β	β	NOUN
ejpam-515	161	14	:	:	PUNCT
ejpam-515	161	15	||β	||β	NOUN
ejpam-515	161	16	−	−	PROPN
ejpam-515	161	17	β0||	β0||	PROPN
ejpam-515	161	18	≤	≤	ADV
ejpam-515	161	19	τn	τn	ADP
ejpam-515	161	20	p	p	PROPN
ejpam-515	161	21	n−1	n−1	PROPN
ejpam-515	161	22	log	log	NOUN
ejpam-515	161	23	log	log	NOUN
ejpam-515	161	24	n	n	CCONJ
ejpam-515	161	25	}	}	SYM
ejpam-515	161	26	∂	∂	NOUN
ejpam-515	161	27	an	an	DET
ejpam-515	161	28	=	=	X
ejpam-515	161	29	{	{	PUNCT
ejpam-515	161	30	β	β	NOUN
ejpam-515	161	31	:	:	PUNCT
ejpam-515	161	32	||β	||β	NOUN
ejpam-515	161	33	−	−	PROPN
ejpam-515	161	34	β0||=	β0||=	NOUN
ejpam-515	161	35	τn	τn	ADP
ejpam-515	161	36	p	p	PROPN
ejpam-515	161	37	n−1	n−1	PROPN
ejpam-515	161	38	log	log	NOUN
ejpam-515	161	39	log	log	NOUN
ejpam-515	161	40	n	n	CCONJ
ejpam-515	161	41	}	}	PUNCT
ejpam-515	161	42	.	.	PUNCT
ejpam-515	162	1	it	it	PRON
ejpam-515	162	2	is	be	AUX
ejpam-515	162	3	clear	clear	ADJ
ejpam-515	162	4	that	that	SCONJ
ejpam-515	162	5	a1	a1	PROPN
ejpam-515	162	6	⊃	⊃	PROPN
ejpam-515	162	7	a2	a2	PROPN
ejpam-515	162	8	⊃	⊃	PROPN
ejpam-515	162	9	a3	a3	PROPN
ejpam-515	162	10	⊃	⊃	PROPN
ejpam-515	162	11	·	·	PUNCT
ejpam-515	162	12	·	·	PUNCT
ejpam-515	162	13	·	·	PUNCT
ejpam-515	163	1	⊃	⊃	PROPN
ejpam-515	163	2	an	an	X
ejpam-515	163	3	.	.	PUNCT
ejpam-515	163	4	further	far	ADV
ejpam-515	163	5	we	we	PRON
ejpam-515	163	6	define	define	VERB
ejpam-515	163	7	h(β	h(β	PROPN
ejpam-515	163	8	,	,	PUNCT
ejpam-515	163	9	n	n	CCONJ
ejpam-515	163	10	)	)	PUNCT
ejpam-515	163	11	=	=	SYM
ejpam-515	163	12	ℓ(β0|yn	ℓ(β0|yn	PROPN
ejpam-515	163	13	,	,	PUNCT
ejpam-515	163	14	xn)−	xn)−	PUNCT
ejpam-515	163	15	ℓ(β	ℓ(β	PROPN
ejpam-515	163	16	|yn	|yn	NUM
ejpam-515	163	17	,	,	PUNCT
ejpam-515	163	18	xn	xn	PROPN
ejpam-515	163	19	)	)	PUNCT
ejpam-515	163	20	=	=	SYM
ejpam-515	164	1	n	n	PROPN
ejpam-515	164	2	∑	∑	PUNCT
ejpam-515	164	3	k=1	k=1	X
ejpam-515	164	4	{	{	PUNCT
ejpam-515	164	5	ext	ext	PROPN
ejpam-515	164	6	k	k	PROPN
ejpam-515	164	7	β	β	PROPN
ejpam-515	164	8	−	−	PROPN
ejpam-515	164	9	ext	ext	PROPN
ejpam-515	164	10	k	k	PROPN
ejpam-515	164	11	β0	β0	PROPN
ejpam-515	164	12	−	−	PROPN
ejpam-515	164	13	ykxt	ykxt	ADJ
ejpam-515	164	14	k(β	k(β	PROPN
ejpam-515	164	15	−	−	PROPN
ejpam-515	164	16	β0	β0	NOUN
ejpam-515	164	17	)	)	PUNCT
ejpam-515	164	18	}	}	PUNCT
ejpam-515	164	19	,	,	PUNCT
ejpam-515	164	20	and	and	CCONJ
ejpam-515	164	21	k(t	k(t	PROPN
ejpam-515	164	22	,	,	PUNCT
ejpam-515	164	23	s	s	X
ejpam-515	164	24	)	)	PUNCT
ejpam-515	164	25	=	=	SYM
ejpam-515	164	26	et	et	NOUN
ejpam-515	164	27	−	−	NOUN
ejpam-515	164	28	es	es	INTJ
ejpam-515	164	29	−	−	NOUN
ejpam-515	164	30	es(t	es(t	PUNCT
ejpam-515	164	31	−	−	PROPN
ejpam-515	164	32	s	s	PART
ejpam-515	164	33	)	)	PUNCT
ejpam-515	164	34	.	.	PUNCT
ejpam-515	165	1	by	by	ADP
ejpam-515	165	2	these	these	DET
ejpam-515	165	3	definitions	definition	NOUN
ejpam-515	165	4	it	it	PRON
ejpam-515	165	5	follows	follow	VERB
ejpam-515	165	6	that	that	SCONJ
ejpam-515	165	7	h(β	h(β	PROPN
ejpam-515	165	8	,	,	PUNCT
ejpam-515	165	9	n	n	CCONJ
ejpam-515	165	10	)	)	PUNCT
ejpam-515	165	11	=	=	SYM
ejpam-515	166	1	n	n	PROPN
ejpam-515	166	2	∑	∑	PUNCT
ejpam-515	166	3	k=1	k=1	PROPN
ejpam-515	166	4	k(xt	k(xt	NOUN
ejpam-515	166	5	kβ	kβ	ADV
ejpam-515	166	6	,	,	PUNCT
ejpam-515	166	7	xt	xt	PROPN
ejpam-515	166	8	kβ0)−	kβ0)−	PROPN
ejpam-515	166	9	n	n	PROPN
ejpam-515	166	10	∑	∑	PROPN
ejpam-515	166	11	k=1	k=1	PROPN
ejpam-515	167	1	(	(	PUNCT
ejpam-515	167	2	yk	yk	PROPN
ejpam-515	167	3	−µ0k)x	−µ0k)x	PROPN
ejpam-515	167	4	t	t	PROPN
ejpam-515	167	5	k(β	k(β	PROPN
ejpam-515	167	6	−	−	PROPN
ejpam-515	167	7	β0	β0	PROPN
ejpam-515	167	8	)	)	PUNCT
ejpam-515	167	9	def	def	NOUN
ejpam-515	167	10	=	=	SYM
ejpam-515	167	11	r1(β	r1(β	PROPN
ejpam-515	167	12	,	,	PUNCT
ejpam-515	167	13	n	n	CCONJ
ejpam-515	167	14	)	)	PUNCT
ejpam-515	168	1	+	+	CCONJ
ejpam-515	168	2	r2(β	r2(β	PROPN
ejpam-515	168	3	,	,	PUNCT
ejpam-515	168	4	n	n	CCONJ
ejpam-515	168	5	)	)	PUNCT
ejpam-515	168	6	.	.	PUNCT
ejpam-515	169	1	(	(	PUNCT
ejpam-515	169	2	8)	8)	NUM
ejpam-515	169	3	before	before	ADP
ejpam-515	169	4	proving	prove	VERB
ejpam-515	169	5	the	the	DET
ejpam-515	169	6	main	main	ADJ
ejpam-515	169	7	results	result	NOUN
ejpam-515	169	8	we	we	PRON
ejpam-515	169	9	need	need	VERB
ejpam-515	169	10	to	to	PART
ejpam-515	169	11	establish	establish	VERB
ejpam-515	169	12	some	some	DET
ejpam-515	169	13	preliminary	preliminary	ADJ
ejpam-515	169	14	results	result	NOUN
ejpam-515	169	15	.	.	PUNCT
ejpam-515	170	1	lemma	lemma	PROPN
ejpam-515	170	2	1	1	NUM
ejpam-515	170	3	.	.	PUNCT
ejpam-515	171	1	the	the	DET
ejpam-515	171	2	function	function	NOUN
ejpam-515	171	3	k(t	k(t	PROPN
ejpam-515	171	4	,	,	PUNCT
ejpam-515	171	5	s	s	X
ejpam-515	171	6	)	)	PUNCT
ejpam-515	171	7	defined	define	VERB
ejpam-515	171	8	has	have	VERB
ejpam-515	171	9	the	the	DET
ejpam-515	171	10	following	follow	VERB
ejpam-515	171	11	properties	property	NOUN
ejpam-515	171	12	:	:	PUNCT
ejpam-515	171	13	(	(	PUNCT
ejpam-515	171	14	i	i	NOUN
ejpam-515	171	15	)	)	PUNCT
ejpam-515	171	16	.	.	PUNCT
ejpam-515	172	1	k(t	k(t	PROPN
ejpam-515	172	2	,	,	PUNCT
ejpam-515	172	3	s	s	X
ejpam-515	172	4	)	)	PUNCT
ejpam-515	172	5	≥	≥	NOUN
ejpam-515	172	6	0	0	NUM
ejpam-515	172	7	for	for	ADP
ejpam-515	172	8	any	any	DET
ejpam-515	172	9	real	real	ADJ
ejpam-515	172	10	numbers	number	NOUN
ejpam-515	172	11	t	t	PROPN
ejpam-515	172	12	and	and	CCONJ
ejpam-515	172	13	s.	s.	PROPN
ejpam-515	172	14	(	(	PUNCT
ejpam-515	172	15	ii	ii	PROPN
ejpam-515	172	16	)	)	PUNCT
ejpam-515	172	17	.	.	PUNCT
ejpam-515	173	1	k(t	k(t	PROPN
ejpam-515	173	2	,	,	PUNCT
ejpam-515	173	3	s	s	PART
ejpam-515	173	4	)	)	PUNCT
ejpam-515	173	5	is	be	AUX
ejpam-515	173	6	strictly	strictly	ADV
ejpam-515	173	7	convex	convex	ADJ
ejpam-515	173	8	with	with	ADP
ejpam-515	173	9	respect	respect	NOUN
ejpam-515	173	10	to	to	ADP
ejpam-515	173	11	t.	t.	PROPN
ejpam-515	173	12	(	(	PUNCT
ejpam-515	173	13	iii	iii	PROPN
ejpam-515	173	14	)	)	PUNCT
ejpam-515	173	15	.	.	PUNCT
ejpam-515	174	1	for	for	ADP
ejpam-515	174	2	any	any	DET
ejpam-515	174	3	∆	∆	PROPN
ejpam-515	174	4	>	>	X
ejpam-515	174	5	0	0	NUM
ejpam-515	174	6	,	,	PUNCT
ejpam-515	174	7	1	1	NUM
ejpam-515	174	8	2	2	NUM
ejpam-515	174	9	es−2∆(t	es−2∆(t	NOUN
ejpam-515	174	10	−	−	PROPN
ejpam-515	174	11	s)2	s)2	NOUN
ejpam-515	174	12	≤	≤	PROPN
ejpam-515	174	13	k(t	k(t	PROPN
ejpam-515	174	14	,	,	PUNCT
ejpam-515	174	15	s	s	X
ejpam-515	174	16	)	)	PUNCT
ejpam-515	174	17	≤	≤	NUM
ejpam-515	174	18	1	1	NUM
ejpam-515	174	19	2	2	NUM
ejpam-515	174	20	es+2∆(t	es+2∆(t	NOUN
ejpam-515	174	21	−	−	NOUN
ejpam-515	174	22	s)2	s)2	NOUN
ejpam-515	175	1	if	if	SCONJ
ejpam-515	175	2	|t	|t	PROPN
ejpam-515	175	3	−	−	PROPN
ejpam-515	175	4	s|	s|	AUX
ejpam-515	175	5	≤∆.	≤∆.	VERB
ejpam-515	175	6	the	the	DET
ejpam-515	175	7	proof	proof	NOUN
ejpam-515	175	8	of	of	ADP
ejpam-515	175	9	lemma	lemma	PROPN
ejpam-515	175	10	1	1	NUM
ejpam-515	175	11	will	will	AUX
ejpam-515	175	12	be	be	AUX
ejpam-515	175	13	give	give	VERB
ejpam-515	175	14	in	in	ADP
ejpam-515	175	15	appendix	appendix	NOUN
ejpam-515	175	16	.	.	PUNCT
ejpam-515	176	1	lemma	lemma	PROPN
ejpam-515	176	2	2	2	X
ejpam-515	176	3	.	.	PUNCT
ejpam-515	177	1	let	let	VERB
ejpam-515	177	2	w	w	NOUN
ejpam-515	177	3	be	be	AUX
ejpam-515	177	4	a	a	DET
ejpam-515	177	5	poisson(θ	poisson(θ	NOUN
ejpam-515	177	6	)	)	PUNCT
ejpam-515	177	7	random	random	ADJ
ejpam-515	177	8	variable	variable	NOUN
ejpam-515	177	9	.	.	PUNCT
ejpam-515	178	1	then	then	ADV
ejpam-515	178	2	for	for	ADP
ejpam-515	178	3	any	any	DET
ejpam-515	178	4	w	w	NOUN
ejpam-515	178	5	≥	≥	NOUN
ejpam-515	178	6	0	0	NUM
ejpam-515	178	7	the	the	DET
ejpam-515	178	8	following	follow	VERB
ejpam-515	178	9	inequalities	inequality	NOUN
ejpam-515	178	10	hold	hold	VERB
ejpam-515	178	11	:	:	PUNCT
ejpam-515	178	12	p(w	p(w	PROPN
ejpam-515	178	13	≤	≤	NUM
ejpam-515	178	14	w)≤	w)≤	X
ejpam-515	178	15	(	(	PUNCT
ejpam-515	178	16	2π)−	2π)−	NUM
ejpam-515	178	17	1	1	NUM
ejpam-515	178	18	2	2	NUM
ejpam-515	178	19	∫	∫	NOUN
ejpam-515	178	20	(	(	PUNCT
ejpam-515	178	21	w+1−θ	w+1−θ	PROPN
ejpam-515	178	22	)	)	PUNCT
ejpam-515	178	23	/pθ	/pθ	PROPN
ejpam-515	179	1	−∞	−∞	PUNCT
ejpam-515	179	2	e−	e−	PROPN
ejpam-515	179	3	1	1	NUM
ejpam-515	179	4	2	2	NUM
ejpam-515	179	5	t2	t2	NOUN
ejpam-515	179	6	d	d	PROPN
ejpam-515	179	7	t	t	PROPN
ejpam-515	179	8	,	,	PUNCT
ejpam-515	179	9	(	(	PUNCT
ejpam-515	179	10	9	9	X
ejpam-515	179	11	)	)	PUNCT
ejpam-515	179	12	p(w	p(w	PROPN
ejpam-515	179	13	≤	≤	NUM
ejpam-515	179	14	w	w	NOUN
ejpam-515	179	15	)	)	PUNCT
ejpam-515	179	16	≥	≥	NOUN
ejpam-515	180	1	[	[	X
ejpam-515	180	2	γ(θ	γ(θ	PROPN
ejpam-515	180	3	+	+	CCONJ
ejpam-515	180	4	1)]−1	1)]−1	NUM
ejpam-515	180	5	∫	∫	NOUN
ejpam-515	180	6	w	w	NOUN
ejpam-515	180	7	0	0	NUM
ejpam-515	180	8	tθ	tθ	NOUN
ejpam-515	180	9	e−t	e−t	NOUN
ejpam-515	180	10	d	d	X
ejpam-515	180	11	t.	t.	NOUN
ejpam-515	180	12	(	(	PUNCT
ejpam-515	180	13	10	10	NUM
ejpam-515	180	14	)	)	PUNCT
ejpam-515	180	15	g.	g.	PROPN
ejpam-515	180	16	qian	qian	PROPN
ejpam-515	180	17	/	/	SYM
ejpam-515	180	18	eur	eur	PROPN
ejpam-515	180	19	.	.	PUNCT
ejpam-515	181	1	j.	j.	PROPN
ejpam-515	181	2	pure	pure	PROPN
ejpam-515	181	3	appl	appl	PROPN
ejpam-515	181	4	.	.	PROPN
ejpam-515	181	5	math	math	PROPN
ejpam-515	181	6	,	,	PUNCT
ejpam-515	181	7	3	3	NUM
ejpam-515	181	8	(	(	PUNCT
ejpam-515	181	9	2010	2010	NUM
ejpam-515	181	10	)	)	PUNCT
ejpam-515	181	11	,	,	PUNCT
ejpam-515	181	12	417	417	NUM
ejpam-515	181	13	-	-	SYM
ejpam-515	181	14	434	434	NUM
ejpam-515	181	15	424	424	NUM
ejpam-515	181	16	the	the	DET
ejpam-515	181	17	results	result	NOUN
ejpam-515	181	18	of	of	ADP
ejpam-515	181	19	lemma	lemma	PROPN
ejpam-515	181	20	2	2	NUM
ejpam-515	181	21	were	be	AUX
ejpam-515	181	22	obtained	obtain	VERB
ejpam-515	181	23	by	by	ADP
ejpam-515	181	24	[	[	X
ejpam-515	181	25	2	2	NUM
ejpam-515	181	26	]	]	PUNCT
ejpam-515	181	27	which	which	PRON
ejpam-515	181	28	can	can	AUX
ejpam-515	181	29	also	also	ADV
ejpam-515	181	30	be	be	AUX
ejpam-515	181	31	found	find	VERB
ejpam-515	181	32	in	in	ADP
ejpam-515	181	33	[	[	X
ejpam-515	181	34	6	6	NUM
ejpam-515	181	35	,	,	PUNCT
ejpam-515	181	36	p.	p.	NOUN
ejpam-515	181	37	102	102	NUM
ejpam-515	181	38	]	]	PUNCT
ejpam-515	181	39	.	.	PUNCT
ejpam-515	182	1	lemma	lemma	PROPN
ejpam-515	182	2	3	3	NUM
ejpam-515	182	3	(	(	PUNCT
ejpam-515	182	4	law	law	NOUN
ejpam-515	182	5	of	of	ADP
ejpam-515	182	6	the	the	DET
ejpam-515	182	7	iterated	iterated	ADJ
ejpam-515	182	8	logarithm	logarithm	NOUN
ejpam-515	182	9	)	)	PUNCT
ejpam-515	182	10	.	.	PUNCT
ejpam-515	183	1	let	let	VERB
ejpam-515	183	2	{	{	PUNCT
ejpam-515	183	3	zn	zn	PROPN
ejpam-515	183	4	,	,	PUNCT
ejpam-515	183	5	n	n	PRON
ejpam-515	183	6	≥	≥	NOUN
ejpam-515	183	7	1	1	NUM
ejpam-515	183	8	}	}	PUNCT
ejpam-515	183	9	be	be	AUX
ejpam-515	183	10	independent	independent	ADJ
ejpam-515	183	11	random	random	ADJ
ejpam-515	183	12	variables	variable	NOUN
ejpam-515	183	13	with	with	ADP
ejpam-515	183	14	ezn	ezn	ADJ
ejpam-515	183	15	=	=	SYM
ejpam-515	183	16	0	0	NUM
ejpam-515	183	17	,	,	PUNCT
ejpam-515	183	18	ez2	ez2	NOUN
ejpam-515	183	19	n	n	NOUN
ejpam-515	183	20	=	=	SYM
ejpam-515	183	21	σ	σ	PROPN
ejpam-515	183	22	2	2	NUM
ejpam-515	183	23	n	n	NOUN
ejpam-515	183	24	and	and	CCONJ
ejpam-515	183	25	s2	s2	VERB
ejpam-515	183	26	n	n	NOUN
ejpam-515	183	27	=	=	SYM
ejpam-515	183	28	∑n	∑n	PROPN
ejpam-515	183	29	k=1σ	k=1σ	NOUN
ejpam-515	183	30	2	2	NUM
ejpam-515	183	31	k	k	PROPN
ejpam-515	183	32	→	→	SYM
ejpam-515	183	33	∞.	∞.	PROPN
ejpam-515	183	34	if	if	SCONJ
ejpam-515	183	35	|zn|	|zn|	ADJ
ejpam-515	183	36	≤	≤	NUM
ejpam-515	183	37	dn	dn	ADP
ejpam-515	184	1	a.s	a.s	PROPN
ejpam-515	184	2	.	.	PROPN
ejpam-515	184	3	,	,	PUNCT
ejpam-515	184	4	where	where	SCONJ
ejpam-515	184	5	dn	dn	PROPN
ejpam-515	184	6	=	=	PUNCT
ejpam-515	184	7	o((s2	o((s2	X
ejpam-515	184	8	n/	n/	ADV
ejpam-515	184	9	log	log	VERB
ejpam-515	184	10	log	log	NOUN
ejpam-515	184	11	s2	s2	PROPN
ejpam-515	184	12	n	n	CCONJ
ejpam-515	184	13	)	)	PUNCT
ejpam-515	184	14	1/2	1/2	NUM
ejpam-515	184	15	)	)	PUNCT
ejpam-515	184	16	,	,	PUNCT
ejpam-515	184	17	then	then	ADV
ejpam-515	184	18	lim	lim	PROPN
ejpam-515	184	19	sup	sup	PROPN
ejpam-515	184	20	n→∞	n→∞	PRON
ejpam-515	184	21	±∑n	±∑n	NOUN
ejpam-515	184	22	k=1	k=1	PROPN
ejpam-515	184	23	zk	zk	PROPN
ejpam-515	185	1	p	p	NOUN
ejpam-515	185	2	2s2	2s2	NUM
ejpam-515	185	3	n	n	CCONJ
ejpam-515	185	4	log	log	NOUN
ejpam-515	185	5	log	log	NOUN
ejpam-515	185	6	s2	s2	NOUN
ejpam-515	185	7	n	n	CCONJ
ejpam-515	185	8	=	=	SYM
ejpam-515	185	9	1	1	NUM
ejpam-515	185	10	a.s	a.s	PROPN
ejpam-515	185	11	..	..	PROPN
ejpam-515	185	12	this	this	DET
ejpam-515	185	13	lemma	lemma	PROPN
ejpam-515	185	14	and	and	CCONJ
ejpam-515	185	15	its	its	PRON
ejpam-515	185	16	proof	proof	NOUN
ejpam-515	185	17	can	can	AUX
ejpam-515	185	18	be	be	AUX
ejpam-515	185	19	found	find	VERB
ejpam-515	185	20	in	in	ADP
ejpam-515	185	21	e.g.	e.g.	ADV
ejpam-515	185	22	[	[	X
ejpam-515	185	23	3	3	NUM
ejpam-515	185	24	,	,	PUNCT
ejpam-515	185	25	pp	pp	ADJ
ejpam-515	185	26	.	.	PUNCT
ejpam-515	186	1	373	373	NUM
ejpam-515	186	2	-	-	SYM
ejpam-515	186	3	374	374	NUM
ejpam-515	186	4	]	]	PUNCT
ejpam-515	186	5	and	and	CCONJ
ejpam-515	186	6	[	[	X
ejpam-515	186	7	9	9	NUM
ejpam-515	186	8	,	,	PUNCT
ejpam-515	186	9	pp	pp	ADJ
ejpam-515	186	10	.	.	PUNCT
ejpam-515	187	1	239	239	NUM
ejpam-515	187	2	-	-	SYM
ejpam-515	187	3	246	246	NUM
ejpam-515	187	4	]	]	PUNCT
ejpam-515	187	5	.	.	PUNCT
ejpam-515	188	1	lemma	lemma	PROPN
ejpam-515	188	2	4	4	NUM
ejpam-515	188	3	.	.	PUNCT
ejpam-515	189	1	under	under	ADP
ejpam-515	189	2	conditions	condition	NOUN
ejpam-515	189	3	(	(	PUNCT
ejpam-515	189	4	c.1	c.1	NOUN
ejpam-515	189	5	)	)	PUNCT
ejpam-515	189	6	,	,	PUNCT
ejpam-515	189	7	(	(	PUNCT
ejpam-515	189	8	c.2	c.2	NOUN
ejpam-515	189	9	)	)	PUNCT
ejpam-515	189	10	and	and	CCONJ
ejpam-515	189	11	(	(	PUNCT
ejpam-515	189	12	c.4	c.4	X
ejpam-515	189	13	)	)	PUNCT
ejpam-515	189	14	to	to	PART
ejpam-515	189	15	(	(	PUNCT
ejpam-515	189	16	c.6	c.6	NOUN
ejpam-515	189	17	)	)	PUNCT
ejpam-515	189	18	,	,	PUNCT
ejpam-515	189	19	we	we	PRON
ejpam-515	189	20	have	have	VERB
ejpam-515	189	21	lim	lim	PROPN
ejpam-515	189	22	sup	sup	PROPN
ejpam-515	189	23	n→∞	n→∞	PRON
ejpam-515	189	24	±∑n	±∑n	PROPN
ejpam-515	189	25	k=1(yk	k=1(yk	PROPN
ejpam-515	190	1	−µ0k)xk	−µ0k)xk	PROPN
ejpam-515	190	2	j	j	PROPN
ejpam-515	190	3	p	p	PROPN
ejpam-515	190	4	2in(β0	2in(β0	PROPN
ejpam-515	190	5	)	)	PUNCT
ejpam-515	190	6	(	(	PUNCT
ejpam-515	190	7	j	j	PROPN
ejpam-515	190	8	,	,	PUNCT
ejpam-515	190	9	j	j	PROPN
ejpam-515	190	10	)	)	PUNCT
ejpam-515	190	11	log	log	VERB
ejpam-515	190	12	log	log	NOUN
ejpam-515	190	13	in(β0	in(β0	NOUN
ejpam-515	190	14	)	)	PUNCT
ejpam-515	191	1	(	(	PUNCT
ejpam-515	191	2	j	j	PROPN
ejpam-515	191	3	,	,	PUNCT
ejpam-515	191	4	j	j	PROPN
ejpam-515	191	5	)	)	PUNCT
ejpam-515	191	6	=	=	NOUN
ejpam-515	191	7	1	1	NUM
ejpam-515	191	8	a.s	a.s	PROPN
ejpam-515	191	9	.	.	PROPN
ejpam-515	191	10	for	for	ADP
ejpam-515	191	11	j	j	PROPN
ejpam-515	191	12	=	=	SYM
ejpam-515	191	13	1	1	NUM
ejpam-515	191	14	,	,	PUNCT
ejpam-515	191	15	·	·	PUNCT
ejpam-515	191	16	·	·	PUNCT
ejpam-515	191	17	·	·	PUNCT
ejpam-515	191	18	,	,	PUNCT
ejpam-515	191	19	p.	p.	NOUN
ejpam-515	191	20	(	(	PUNCT
ejpam-515	191	21	11	11	NUM
ejpam-515	191	22	)	)	PUNCT
ejpam-515	191	23	here	here	ADV
ejpam-515	191	24	xk	xk	PROPN
ejpam-515	191	25	j	j	PROPN
ejpam-515	191	26	is	be	AUX
ejpam-515	191	27	the	the	DET
ejpam-515	191	28	j	j	PROPN
ejpam-515	191	29	-	-	PUNCT
ejpam-515	191	30	th	th	VERB
ejpam-515	191	31	element	element	NOUN
ejpam-515	191	32	of	of	ADP
ejpam-515	191	33	xk	xk	PROPN
ejpam-515	191	34	and	and	CCONJ
ejpam-515	191	35	in(β0	in(β0	PROPN
ejpam-515	191	36	)	)	PUNCT
ejpam-515	191	37	(	(	PUNCT
ejpam-515	191	38	j	j	PROPN
ejpam-515	191	39	,	,	PUNCT
ejpam-515	191	40	j	j	PROPN
ejpam-515	191	41	)	)	PUNCT
ejpam-515	191	42	is	be	AUX
ejpam-515	191	43	the	the	DET
ejpam-515	191	44	(	(	PUNCT
ejpam-515	191	45	j	j	PROPN
ejpam-515	191	46	,	,	PUNCT
ejpam-515	191	47	j)-th	j)-th	PROPN
ejpam-515	191	48	element	element	NOUN
ejpam-515	191	49	of	of	ADP
ejpam-515	191	50	in(β0	in(β0	NOUN
ejpam-515	191	51	)	)	PUNCT
ejpam-515	191	52	.	.	PUNCT
ejpam-515	192	1	equation	equation	NOUN
ejpam-515	192	2	(	(	PUNCT
ejpam-515	192	3	11	11	NUM
ejpam-515	192	4	)	)	PUNCT
ejpam-515	192	5	suggests	suggest	VERB
ejpam-515	192	6	that	that	SCONJ
ejpam-515	192	7	{	{	PUNCT
ejpam-515	192	8	(	(	PUNCT
ejpam-515	192	9	yk−µ0k)xk	yk−µ0k)xk	PROPN
ejpam-515	192	10	j	j	PROPN
ejpam-515	192	11	,	,	PUNCT
ejpam-515	192	12	k	k	PROPN
ejpam-515	192	13	=	=	SYM
ejpam-515	192	14	1,2	1,2	NUM
ejpam-515	192	15	,	,	PUNCT
ejpam-515	192	16	·	·	PUNCT
ejpam-515	192	17	·	·	PUNCT
ejpam-515	192	18	·	·	PUNCT
ejpam-515	192	19	}	}	PUNCT
ejpam-515	192	20	obeys	obey	VERB
ejpam-515	192	21	the	the	DET
ejpam-515	192	22	law	law	NOUN
ejpam-515	192	23	of	of	ADP
ejpam-515	192	24	iterated	iterated	ADJ
ejpam-515	192	25	logarithm	logarithm	NOUN
ejpam-515	192	26	.	.	PUNCT
ejpam-515	193	1	accordingly	accordingly	ADV
ejpam-515	193	2	,	,	PUNCT
ejpam-515	193	3	we	we	PRON
ejpam-515	193	4	have	have	VERB
ejpam-515	193	5	∂	∂	NUM
ejpam-515	193	6	ℓ	ℓ	NOUN
ejpam-515	193	7	∂	∂	PROPN
ejpam-515	193	8	β	β	X
ejpam-515	193	9	|β	|β	PROPN
ejpam-515	193	10	=	=	PROPN
ejpam-515	193	11	β0	β0	NOUN
ejpam-515	193	12	=	=	NOUN
ejpam-515	193	13	n	n	NOUN
ejpam-515	193	14	∑	∑	PUNCT
ejpam-515	193	15	k=1	k=1	PROPN
ejpam-515	193	16	(	(	PUNCT
ejpam-515	193	17	yk	yk	INTJ
ejpam-515	193	18	−µ0k)xk	−µ0k)xk	PROPN
ejpam-515	193	19	=	=	PUNCT
ejpam-515	193	20	x	x	SYM
ejpam-515	193	21	t	t	NOUN
ejpam-515	193	22	n(yn−µ0	n(yn−µ0	PROPN
ejpam-515	193	23	)	)	PUNCT
ejpam-515	194	1	=	=	PUNCT
ejpam-515	195	1	o	o	NOUN
ejpam-515	195	2	(	(	PUNCT
ejpam-515	195	3	p	p	NOUN
ejpam-515	195	4	n	n	CCONJ
ejpam-515	195	5	log	log	VERB
ejpam-515	195	6	log	log	NOUN
ejpam-515	195	7	n	n	CCONJ
ejpam-515	195	8	)	)	PUNCT
ejpam-515	195	9	a.s	a.s	PROPN
ejpam-515	195	10	.	.	PROPN
ejpam-515	195	11	(	(	PUNCT
ejpam-515	195	12	12	12	NUM
ejpam-515	195	13	)	)	PUNCT
ejpam-515	195	14	where	where	SCONJ
ejpam-515	195	15	µ0	µ0	NOUN
ejpam-515	195	16	=	=	SYM
ejpam-515	195	17	(	(	PUNCT
ejpam-515	195	18	µ01	µ01	ADV
ejpam-515	195	19	,	,	PUNCT
ejpam-515	195	20	·	·	PUNCT
ejpam-515	195	21	·	·	PUNCT
ejpam-515	195	22	·	·	PUNCT
ejpam-515	195	23	,	,	PUNCT
ejpam-515	195	24	µ0n	µ0n	NOUN
ejpam-515	195	25	)	)	PUNCT
ejpam-515	195	26	t	t	PROPN
ejpam-515	195	27	is	be	AUX
ejpam-515	195	28	the	the	DET
ejpam-515	195	29	true	true	ADJ
ejpam-515	195	30	mean	mean	NOUN
ejpam-515	195	31	vector	vector	NOUN
ejpam-515	195	32	.	.	PUNCT
ejpam-515	196	1	proof	proof	NOUN
ejpam-515	196	2	.	.	PUNCT
ejpam-515	197	1	the	the	DET
ejpam-515	197	2	result	result	NOUN
ejpam-515	197	3	(	(	PUNCT
ejpam-515	197	4	12	12	NUM
ejpam-515	197	5	)	)	PUNCT
ejpam-515	197	6	is	be	AUX
ejpam-515	197	7	obvious	obvious	ADJ
ejpam-515	197	8	from	from	ADP
ejpam-515	197	9	(	(	PUNCT
ejpam-515	197	10	11	11	NUM
ejpam-515	197	11	)	)	PUNCT
ejpam-515	197	12	and	and	CCONJ
ejpam-515	197	13	condition	condition	NOUN
ejpam-515	197	14	(	(	PUNCT
ejpam-515	197	15	c.2	c.2	NOUN
ejpam-515	197	16	)	)	PUNCT
ejpam-515	197	17	.	.	PUNCT
ejpam-515	198	1	hence	hence	ADV
ejpam-515	198	2	we	we	PRON
ejpam-515	198	3	only	only	ADV
ejpam-515	198	4	need	need	VERB
ejpam-515	198	5	to	to	PART
ejpam-515	198	6	prove	prove	VERB
ejpam-515	198	7	(	(	PUNCT
ejpam-515	198	8	11	11	NUM
ejpam-515	198	9	)	)	PUNCT
ejpam-515	198	10	.	.	PUNCT
ejpam-515	199	1	without	without	ADP
ejpam-515	199	2	losing	lose	VERB
ejpam-515	199	3	generality	generality	NOUN
ejpam-515	199	4	we	we	PRON
ejpam-515	199	5	assume	assume	VERB
ejpam-515	199	6	all	all	PRON
ejpam-515	199	7	xk	xk	PROPN
ejpam-515	200	1	j	j	PROPN
ejpam-515	200	2	>	>	X
ejpam-515	200	3	0	0	X
ejpam-515	200	4	.	.	PUNCT
ejpam-515	201	1	using	use	VERB
ejpam-515	201	2	the	the	DET
ejpam-515	201	3	information	information	NOUN
ejpam-515	201	4	that	that	PRON
ejpam-515	201	5	yk	yk	PROPN
ejpam-515	201	6	∼	∼	NOUN
ejpam-515	201	7	poisson(µ0k	poisson(µ0k	PROPN
ejpam-515	201	8	)	)	PUNCT
ejpam-515	201	9	and	and	CCONJ
ejpam-515	201	10	the	the	DET
ejpam-515	201	11	definition	definition	NOUN
ejpam-515	201	12	of	of	ADP
ejpam-515	201	13	in(β0	in(β0	NOUN
ejpam-515	201	14	)	)	PUNCT
ejpam-515	201	15	it	it	PRON
ejpam-515	201	16	is	be	AUX
ejpam-515	201	17	easy	easy	ADJ
ejpam-515	201	18	to	to	PART
ejpam-515	201	19	verify	verify	VERB
ejpam-515	201	20	that	that	PRON
ejpam-515	201	21	for	for	ADP
ejpam-515	201	22	j	j	PROPN
ejpam-515	201	23	=	=	SYM
ejpam-515	201	24	1	1	NUM
ejpam-515	201	25	,	,	PUNCT
ejpam-515	201	26	·	·	PUNCT
ejpam-515	201	27	·	·	PUNCT
ejpam-515	201	28	·	·	PUNCT
ejpam-515	201	29	,	,	PUNCT
ejpam-515	202	1	p	p	NOUN
ejpam-515	202	2	e(yk	e(yk	PROPN
ejpam-515	202	3	−µ0k)xk	−µ0k)xk	PROPN
ejpam-515	202	4	j	j	PROPN
ejpam-515	202	5	=	=	SYM
ejpam-515	202	6	0	0	PROPN
ejpam-515	202	7	,	,	PUNCT
ejpam-515	202	8	(	(	PUNCT
ejpam-515	202	9	13	13	NUM
ejpam-515	202	10	)	)	PUNCT
ejpam-515	202	11	n	n	NOUN
ejpam-515	202	12	∑	∑	PUNCT
ejpam-515	202	13	k=1	k=1	PUNCT
ejpam-515	202	14	e((yk	e((yk	PUNCT
ejpam-515	202	15	−µ0k)xk	−µ0k)xk	NUM
ejpam-515	202	16	j	j	NOUN
ejpam-515	202	17	)	)	PUNCT
ejpam-515	202	18	2	2	NUM
ejpam-515	202	19	=	=	SYM
ejpam-515	202	20	n	n	PROPN
ejpam-515	202	21	∑	∑	PUNCT
ejpam-515	202	22	k=1	k=1	PROPN
ejpam-515	202	23	µ0k	µ0k	PROPN
ejpam-515	202	24	x2	x2	PROPN
ejpam-515	202	25	k	k	PROPN
ejpam-515	202	26	j	j	PROPN
ejpam-515	202	27	=	=	SYM
ejpam-515	202	28	in(β0	in(β0	PROPN
ejpam-515	202	29	)	)	PUNCT
ejpam-515	202	30	(	(	PUNCT
ejpam-515	202	31	j	j	NOUN
ejpam-515	202	32	,	,	PUNCT
ejpam-515	202	33	j)→∞	j)→∞	NUM
ejpam-515	202	34	(	(	PUNCT
ejpam-515	202	35	14	14	NUM
ejpam-515	202	36	)	)	PUNCT
ejpam-515	202	37	as	as	ADP
ejpam-515	202	38	n→∞	n→∞	NUM
ejpam-515	202	39	by	by	ADP
ejpam-515	202	40	condition	condition	NOUN
ejpam-515	202	41	(	(	PUNCT
ejpam-515	202	42	c.1	c.1	NOUN
ejpam-515	202	43	)	)	PUNCT
ejpam-515	202	44	.	.	PUNCT
ejpam-515	203	1	from	from	ADP
ejpam-515	203	2	now	now	ADV
ejpam-515	203	3	on	on	ADV
ejpam-515	203	4	we	we	PRON
ejpam-515	203	5	proceed	proceed	VERB
ejpam-515	203	6	to	to	PART
ejpam-515	203	7	show	show	VERB
ejpam-515	203	8	that	that	SCONJ
ejpam-515	203	9	for	for	ADP
ejpam-515	203	10	j	j	PROPN
ejpam-515	203	11	=	=	SYM
ejpam-515	203	12	1	1	NUM
ejpam-515	203	13	,	,	PUNCT
ejpam-515	203	14	·	·	PUNCT
ejpam-515	203	15	·	·	PUNCT
ejpam-515	203	16	·	·	PUNCT
ejpam-515	203	17	,	,	PUNCT
ejpam-515	203	18	p	p	NOUN
ejpam-515	203	19	|(yn−µ0n)xnj|	|(yn−µ0n)xnj|	X
ejpam-515	203	20	≤	≤	ADJ
ejpam-515	203	21	o(dnj	o(dnj	PROPN
ejpam-515	203	22	)	)	PUNCT
ejpam-515	203	23	a.s	a.s	PROPN
ejpam-515	203	24	.	.	PROPN
ejpam-515	203	25	(	(	PUNCT
ejpam-515	203	26	15	15	NUM
ejpam-515	203	27	)	)	PUNCT
ejpam-515	203	28	where	where	SCONJ
ejpam-515	203	29	dnj	dnj	VERB
ejpam-515	203	30	=	=	SYM
ejpam-515	203	31	p	p	PROPN
ejpam-515	203	32	in(β0	in(β0	NOUN
ejpam-515	203	33	)	)	PUNCT
ejpam-515	203	34	(	(	PUNCT
ejpam-515	203	35	j	j	PROPN
ejpam-515	203	36	,	,	PUNCT
ejpam-515	203	37	j)/	j)/	NOUN
ejpam-515	203	38	log	log	VERB
ejpam-515	203	39	log	log	NOUN
ejpam-515	203	40	in(β0	in(β0	NOUN
ejpam-515	203	41	)	)	PUNCT
ejpam-515	203	42	(	(	PUNCT
ejpam-515	203	43	j	j	PROPN
ejpam-515	203	44	,	,	PUNCT
ejpam-515	203	45	j)→	j)→	PROPN
ejpam-515	203	46	∞	∞	NUM
ejpam-515	203	47	by	by	ADP
ejpam-515	203	48	(	(	PUNCT
ejpam-515	203	49	14	14	NUM
ejpam-515	203	50	)	)	PUNCT
ejpam-515	203	51	.	.	PUNCT
ejpam-515	204	1	for	for	ADP
ejpam-515	204	2	any	any	DET
ejpam-515	204	3	ǫ	ǫ	NOUN
ejpam-515	204	4	>	>	X
ejpam-515	204	5	0	0	NUM
ejpam-515	204	6	,	,	PUNCT
ejpam-515	204	7	it	it	PRON
ejpam-515	204	8	is	be	AUX
ejpam-515	204	9	easy	easy	ADJ
ejpam-515	204	10	to	to	PART
ejpam-515	204	11	see	see	VERB
ejpam-515	204	12	that	that	SCONJ
ejpam-515	204	13	p{|(yn	p{|(yn	ADJ
ejpam-515	204	14	−µ0n)xnj|	−µ0n)xnj|	PROPN
ejpam-515	204	15	>	>	X
ejpam-515	204	16	ǫdnj	ǫdnj	PROPN
ejpam-515	204	17	}	}	PUNCT
ejpam-515	204	18	≤	≤	PROPN
ejpam-515	205	1	p{yn	p{yn	PROPN
ejpam-515	205	2	>	>	X
ejpam-515	205	3	µ0n+	µ0n+	PROPN
ejpam-515	205	4	ǫdnj	ǫdnj	PROPN
ejpam-515	205	5	x	x	SYM
ejpam-515	205	6	−1	−1	NOUN
ejpam-515	205	7	nj	nj	PROPN
ejpam-515	205	8	}	}	PUNCT
ejpam-515	205	9	+	+	CCONJ
ejpam-515	206	1	p{yn	p{yn	PROPN
ejpam-515	206	2	<	<	X
ejpam-515	206	3	µ0n	µ0n	NOUN
ejpam-515	206	4	−	−	PROPN
ejpam-515	206	5	ǫdnj	ǫdnj	NOUN
ejpam-515	206	6	x	x	SYM
ejpam-515	206	7	−1	−1	NOUN
ejpam-515	206	8	nj	nj	PROPN
ejpam-515	206	9	}	}	PUNCT
ejpam-515	206	10	.	.	PUNCT
ejpam-515	207	1	(	(	PUNCT
ejpam-515	207	2	16	16	X
ejpam-515	207	3	)	)	PUNCT
ejpam-515	207	4	applying	apply	VERB
ejpam-515	207	5	(	(	PUNCT
ejpam-515	207	6	9	9	NUM
ejpam-515	207	7	)	)	PUNCT
ejpam-515	207	8	of	of	ADP
ejpam-515	207	9	lemma	lemma	PROPN
ejpam-515	207	10	2	2	NUM
ejpam-515	207	11	we	we	PRON
ejpam-515	207	12	have	have	VERB
ejpam-515	207	13	p{yn	p{yn	PROPN
ejpam-515	207	14	<	<	X
ejpam-515	207	15	µ0n	µ0n	NOUN
ejpam-515	207	16	−	−	PROPN
ejpam-515	207	17	ǫdnj	ǫdnj	NOUN
ejpam-515	207	18	x	x	SYM
ejpam-515	207	19	−1	−1	NOUN
ejpam-515	207	20	nj	nj	PROPN
ejpam-515	207	21	}	}	PUNCT
ejpam-515	207	22	≤	≤	NOUN
ejpam-515	207	23	(	(	PUNCT
ejpam-515	207	24	2π)−	2π)−	NUM
ejpam-515	207	25	1	1	NUM
ejpam-515	207	26	2	2	NUM
ejpam-515	207	27	∫	∫	NOUN
ejpam-515	207	28	(	(	PUNCT
ejpam-515	207	29	1−ǫdn	1−ǫdn	NOUN
ejpam-515	207	30	j	j	X
ejpam-515	207	31	x	x	SYM
ejpam-515	207	32	−1	−1	VERB
ejpam-515	207	33	n	n	PRON
ejpam-515	207	34	j	j	PROPN
ejpam-515	207	35	)	)	PUNCT
ejpam-515	207	36	/	/	SYM
ejpam-515	208	1	p	p	NOUN
ejpam-515	208	2	µ0n	µ0n	NOUN
ejpam-515	208	3	−∞	−∞	PUNCT
ejpam-515	208	4	e−	e−	PROPN
ejpam-515	208	5	1	1	NUM
ejpam-515	208	6	2	2	NUM
ejpam-515	208	7	t2	t2	NOUN
ejpam-515	208	8	d	d	NOUN
ejpam-515	208	9	t.	t.	NOUN
ejpam-515	208	10	(	(	PUNCT
ejpam-515	208	11	17	17	NUM
ejpam-515	208	12	)	)	PUNCT
ejpam-515	208	13	g.	g.	PROPN
ejpam-515	208	14	qian	qian	PROPN
ejpam-515	208	15	/	/	SYM
ejpam-515	208	16	eur	eur	PROPN
ejpam-515	208	17	.	.	PUNCT
ejpam-515	209	1	j.	j.	PROPN
ejpam-515	209	2	pure	pure	PROPN
ejpam-515	209	3	appl	appl	PROPN
ejpam-515	209	4	.	.	PROPN
ejpam-515	209	5	math	math	PROPN
ejpam-515	209	6	,	,	PUNCT
ejpam-515	209	7	3	3	NUM
ejpam-515	209	8	(	(	PUNCT
ejpam-515	209	9	2010	2010	NUM
ejpam-515	209	10	)	)	PUNCT
ejpam-515	209	11	,	,	PUNCT
ejpam-515	209	12	417	417	NUM
ejpam-515	209	13	-	-	SYM
ejpam-515	209	14	434	434	NUM
ejpam-515	209	15	425	425	NUM
ejpam-515	209	16	note	note	NOUN
ejpam-515	209	17	that	that	SCONJ
ejpam-515	209	18	from	from	ADP
ejpam-515	209	19	condition	condition	NOUN
ejpam-515	209	20	(	(	PUNCT
ejpam-515	209	21	c.1	c.1	NOUN
ejpam-515	209	22	)	)	PUNCT
ejpam-515	209	23	and	and	CCONJ
ejpam-515	209	24	the	the	DET
ejpam-515	209	25	inequality	inequality	NOUN
ejpam-515	209	26	λ1{in(β0	λ1{in(β0	NOUN
ejpam-515	209	27	)	)	PUNCT
ejpam-515	209	28	}	}	PUNCT
ejpam-515	209	29	≤	≤	NUM
ejpam-515	209	30	in(β0	in(β0	NOUN
ejpam-515	209	31	)	)	PUNCT
ejpam-515	209	32	(	(	PUNCT
ejpam-515	209	33	j	j	PROPN
ejpam-515	209	34	,	,	PUNCT
ejpam-515	209	35	j	j	PROPN
ejpam-515	209	36	)	)	PUNCT
ejpam-515	209	37	≤	≤	NOUN
ejpam-515	209	38	λp{in(β0	λp{in(β0	NUM
ejpam-515	209	39	)	)	PUNCT
ejpam-515	209	40	}	}	PUNCT
ejpam-515	209	41	we	we	PRON
ejpam-515	209	42	have	have	VERB
ejpam-515	209	43	µ0n	µ0n	PRON
ejpam-515	210	1	x2	x2	PRON
ejpam-515	210	2	nj	nj	PROPN
ejpam-515	210	3	≤	≤	PROPN
ejpam-515	210	4	µ0nxt	µ0nxt	VERB
ejpam-515	210	5	nxn	nxn	PROPN
ejpam-515	210	6	≤	≤	PROPN
ejpam-515	210	7	λp{in(β0)}µ0nxt	λp{in(β0)}µ0nxt	VERB
ejpam-515	210	8	n	n	PRON
ejpam-515	210	9	in(β0	in(β0	NOUN
ejpam-515	210	10	)	)	PUNCT
ejpam-515	211	1	−1xn	−1xn	ADJ
ejpam-515	211	2	≤	≤	NOUN
ejpam-515	211	3	λp{in(β0)}δ2	λp{in(β0)}δ2	ADP
ejpam-515	211	4	n	n	NOUN
ejpam-515	211	5	≤	≤	NOUN
ejpam-515	211	6	λp{in(β0	λp{in(β0	NUM
ejpam-515	211	7	)	)	PUNCT
ejpam-515	211	8	}	}	PUNCT
ejpam-515	212	1	λ1{in(β0	λ1{in(β0	NOUN
ejpam-515	212	2	)	)	PUNCT
ejpam-515	212	3	}	}	PUNCT
ejpam-515	212	4	in(β0	in(β0	NOUN
ejpam-515	212	5	)	)	PUNCT
ejpam-515	212	6	(	(	PUNCT
ejpam-515	212	7	j	j	PROPN
ejpam-515	212	8	,	,	PUNCT
ejpam-515	212	9	j	j	PROPN
ejpam-515	212	10	)	)	PUNCT
ejpam-515	212	11	log	log	VERB
ejpam-515	212	12	log	log	NOUN
ejpam-515	212	13	in(β0	in(β0	NOUN
ejpam-515	212	14	)	)	PUNCT
ejpam-515	213	1	(	(	PUNCT
ejpam-515	213	2	j	j	PROPN
ejpam-515	213	3	,	,	PUNCT
ejpam-515	213	4	j	j	PROPN
ejpam-515	213	5	)	)	PUNCT
ejpam-515	213	6	δ2	δ2	PROPN
ejpam-515	213	7	n	n	CCONJ
ejpam-515	213	8	log	log	VERB
ejpam-515	213	9	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	213	10	)	)	PUNCT
ejpam-515	213	11	}	}	PUNCT
ejpam-515	213	12	≤	≤	NOUN
ejpam-515	213	13	b0d2	b0d2	VERB
ejpam-515	213	14	njδ	njδ	NOUN
ejpam-515	213	15	2	2	NUM
ejpam-515	213	16	n	n	NOUN
ejpam-515	213	17	log	log	NOUN
ejpam-515	213	18	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	213	19	)	)	PUNCT
ejpam-515	213	20	}	}	PUNCT
ejpam-515	213	21	.	.	PUNCT
ejpam-515	214	1	(	(	PUNCT
ejpam-515	214	2	18	18	NUM
ejpam-515	214	3	)	)	PUNCT
ejpam-515	214	4	thus	thus	ADV
ejpam-515	214	5	when	when	SCONJ
ejpam-515	214	6	µ0n	µ0n	PRON
ejpam-515	214	7	≥	≥	NOUN
ejpam-515	214	8	1	1	NUM
ejpam-515	214	9	,	,	PUNCT
ejpam-515	214	10	1−	1−	NUM
ejpam-515	214	11	ǫdnj	ǫdnj	NOUN
ejpam-515	214	12	x	x	SYM
ejpam-515	214	13	−1	−1	NOUN
ejpam-515	214	14	njp	njp	NOUN
ejpam-515	214	15	µ0n	µ0n	NOUN
ejpam-515	215	1	=	=	SYM
ejpam-515	215	2	1	1	NUM
ejpam-515	215	3	p	p	NOUN
ejpam-515	215	4	µ0n	µ0n	NOUN
ejpam-515	215	5	−	−	PROPN
ejpam-515	215	6	ǫ	ǫ	NOUN
ejpam-515	215	7	√	√	NOUN
ejpam-515	215	8	√	√	NUM
ejpam-515	215	9	√	√	PROPN
ejpam-515	215	10	√	√	NUM
ejpam-515	215	11	d2	d2	PROPN
ejpam-515	215	12	nj	nj	PROPN
ejpam-515	215	13	µ0n	µ0n	NOUN
ejpam-515	216	1	x2	x2	PRON
ejpam-515	216	2	nj	nj	PROPN
ejpam-515	216	3	≤	≤	NUM
ejpam-515	217	1	1−	1−	NUM
ejpam-515	218	1	ǫ	ǫ	PRON
ejpam-515	218	2	p	p	X
ejpam-515	218	3	b0δ	b0δ	ADP
ejpam-515	218	4	2	2	NUM
ejpam-515	218	5	n	n	NOUN
ejpam-515	218	6	log	log	NOUN
ejpam-515	218	7	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	218	8	)	)	PUNCT
ejpam-515	218	9	}	}	PUNCT
ejpam-515	218	10	→	→	SYM
ejpam-515	218	11	−∞	−∞	PUNCT
ejpam-515	218	12	by	by	ADP
ejpam-515	218	13	condition	condition	NOUN
ejpam-515	218	14	(	(	PUNCT
ejpam-515	218	15	c.4	c.4	PROPN
ejpam-515	218	16	)	)	PUNCT
ejpam-515	218	17	.	.	PUNCT
ejpam-515	219	1	this	this	PRON
ejpam-515	219	2	implies	imply	VERB
ejpam-515	219	3	that	that	SCONJ
ejpam-515	219	4	1−	1−	NUM
ejpam-515	219	5	ǫdnj	ǫdnj	NOUN
ejpam-515	219	6	x	x	SYM
ejpam-515	219	7	−1	−1	NOUN
ejpam-515	219	8	njp	njp	PROPN
ejpam-515	219	9	µ0n	µ0n	NOUN
ejpam-515	219	10	≤	≤	ADV
ejpam-515	219	11	−1	−1	NOUN
ejpam-515	219	12	2	2	NUM
ejpam-515	219	13	ǫ(b0δ	ǫ(b0δ	SYM
ejpam-515	219	14	2	2	NUM
ejpam-515	219	15	n	n	NOUN
ejpam-515	219	16	log	log	VERB
ejpam-515	219	17	logλp{in(β0)})−	logλp{in(β0)})−	NUM
ejpam-515	219	18	1	1	NUM
ejpam-515	219	19	2	2	NUM
ejpam-515	219	20	when	when	SCONJ
ejpam-515	219	21	n	n	X
ejpam-515	219	22	is	be	AUX
ejpam-515	219	23	sufficiently	sufficiently	ADV
ejpam-515	219	24	large	large	ADJ
ejpam-515	219	25	.	.	PUNCT
ejpam-515	220	1	(	(	PUNCT
ejpam-515	220	2	19	19	NUM
ejpam-515	220	3	)	)	PUNCT
ejpam-515	220	4	from	from	ADP
ejpam-515	220	5	(	(	PUNCT
ejpam-515	220	6	17	17	NUM
ejpam-515	220	7	)	)	PUNCT
ejpam-515	220	8	,	,	PUNCT
ejpam-515	220	9	(	(	PUNCT
ejpam-515	220	10	19	19	NUM
ejpam-515	220	11	)	)	PUNCT
ejpam-515	220	12	and	and	CCONJ
ejpam-515	220	13	a	a	DET
ejpam-515	220	14	well	well	ADV
ejpam-515	220	15	-	-	PUNCT
ejpam-515	220	16	known	know	VERB
ejpam-515	220	17	inequality	inequality	NOUN
ejpam-515	220	18	∫	∫	PROPN
ejpam-515	220	19	∞	∞	PROPN
ejpam-515	221	1	a	a	DET
ejpam-515	221	2	e−	e−	PROPN
ejpam-515	221	3	1	1	NUM
ejpam-515	221	4	2	2	NUM
ejpam-515	221	5	t2	t2	NOUN
ejpam-515	221	6	d	d	X
ejpam-515	221	7	t	t	X
ejpam-515	221	8	<	<	X
ejpam-515	221	9	1	1	NUM
ejpam-515	221	10	a	a	DET
ejpam-515	221	11	e−	e−	PROPN
ejpam-515	221	12	1	1	NUM
ejpam-515	221	13	2	2	NUM
ejpam-515	221	14	a2	a2	NOUN
ejpam-515	221	15	for	for	ADP
ejpam-515	221	16	all	all	DET
ejpam-515	221	17	a	a	PRON
ejpam-515	221	18	>	>	X
ejpam-515	221	19	0	0	PUNCT
ejpam-515	222	1	[	[	X
ejpam-515	222	2	see	see	VERB
ejpam-515	222	3	3	3	NUM
ejpam-515	222	4	,	,	PUNCT
ejpam-515	222	5	p.49	p.49	ADP
ejpam-515	222	6	]	]	PUNCT
ejpam-515	222	7	,	,	PUNCT
ejpam-515	222	8	it	it	PRON
ejpam-515	222	9	follows	follow	VERB
ejpam-515	222	10	that	that	SCONJ
ejpam-515	222	11	when	when	SCONJ
ejpam-515	222	12	µ0n	µ0n	PRON
ejpam-515	222	13	≥	≥	AUX
ejpam-515	222	14	1	1	NUM
ejpam-515	222	15	and	and	CCONJ
ejpam-515	222	16	n	n	PRON
ejpam-515	222	17	is	be	AUX
ejpam-515	222	18	sufficiently	sufficiently	ADV
ejpam-515	222	19	large	large	ADJ
ejpam-515	222	20	,	,	PUNCT
ejpam-515	222	21	p{yn	p{yn	PROPN
ejpam-515	222	22	<	<	X
ejpam-515	222	23	µ0n	µ0n	NOUN
ejpam-515	222	24	−	−	NUM
ejpam-515	222	25	ǫdnj	ǫdnj	NOUN
ejpam-515	222	26	x	x	SYM
ejpam-515	222	27	−1	−1	NOUN
ejpam-515	222	28	nj	nj	PROPN
ejpam-515	222	29	}	}	PUNCT
ejpam-515	222	30	≤	≤	NOUN
ejpam-515	222	31	(	(	PUNCT
ejpam-515	222	32	2π)−	2π)−	NUM
ejpam-515	222	33	1	1	NUM
ejpam-515	222	34	2	2	NUM
ejpam-515	222	35	∫	∫	NOUN
ejpam-515	222	36	−	−	NUM
ejpam-515	222	37	1	1	NUM
ejpam-515	222	38	2	2	NUM
ejpam-515	222	39	ǫ(b0δ	ǫ(b0δ	SYM
ejpam-515	222	40	2	2	NUM
ejpam-515	222	41	n	n	NOUN
ejpam-515	222	42	log	log	VERB
ejpam-515	222	43	logλp{in(β0)})−	logλp{in(β0)})−	NUM
ejpam-515	222	44	1	1	NUM
ejpam-515	222	45	2	2	NUM
ejpam-515	222	46	−∞	−∞	ADP
ejpam-515	222	47	e−	e−	PROPN
ejpam-515	222	48	1	1	NUM
ejpam-515	222	49	2	2	NUM
ejpam-515	222	50	t2	t2	NOUN
ejpam-515	222	51	d	d	X
ejpam-515	222	52	t	t	X
ejpam-515	222	53	<	<	X
ejpam-515	222	54	(	(	PUNCT
ejpam-515	222	55	2π)−12ǫ−1(b0δ	2π)−12ǫ−1(b0δ	NUM
ejpam-515	222	56	2	2	NUM
ejpam-515	222	57	n	n	NOUN
ejpam-515	222	58	log	log	NOUN
ejpam-515	222	59	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	222	60	)	)	PUNCT
ejpam-515	222	61	}	}	PUNCT
ejpam-515	222	62	)	)	PUNCT
ejpam-515	222	63	1	1	NUM
ejpam-515	222	64	2	2	NUM
ejpam-515	222	65	e−	e−	PROPN
ejpam-515	222	66	1	1	NUM
ejpam-515	222	67	8	8	NUM
ejpam-515	222	68	ǫ2(b0δ	ǫ2(b0δ	NUM
ejpam-515	222	69	2	2	NUM
ejpam-515	222	70	n	n	NOUN
ejpam-515	222	71	log	log	VERB
ejpam-515	222	72	logλp{in(β0)})−1	logλp{in(β0)})−1	ADJ
ejpam-515	222	73	≤	≤	NUM
ejpam-515	222	74	(	(	PUNCT
ejpam-515	222	75	logλp{in(β0)})−	logλp{in(β0)})−	PROPN
ejpam-515	222	76	1	1	NUM
ejpam-515	222	77	2	2	NUM
ejpam-515	222	78	(	(	PUNCT
ejpam-515	222	79	λp{in(β0)})−2	λp{in(β0)})−2	PROPN
ejpam-515	222	80	(	(	PUNCT
ejpam-515	222	81	20	20	NUM
ejpam-515	222	82	)	)	PUNCT
ejpam-515	222	83	where	where	SCONJ
ejpam-515	222	84	the	the	DET
ejpam-515	222	85	last	last	ADJ
ejpam-515	222	86	inequality	inequality	NOUN
ejpam-515	222	87	follows	follow	VERB
ejpam-515	222	88	from	from	ADP
ejpam-515	222	89	condition	condition	NOUN
ejpam-515	222	90	(	(	PUNCT
ejpam-515	222	91	c.4	c.4	PROPN
ejpam-515	222	92	)	)	PUNCT
ejpam-515	222	93	.	.	PUNCT
ejpam-515	223	1	when	when	SCONJ
ejpam-515	223	2	µ0n	µ0n	PRON
ejpam-515	223	3	<	<	X
ejpam-515	223	4	1	1	NUM
ejpam-515	223	5	,	,	PUNCT
ejpam-515	223	6	it	it	PRON
ejpam-515	223	7	follows	follow	VERB
ejpam-515	223	8	from	from	ADP
ejpam-515	223	9	(	(	PUNCT
ejpam-515	223	10	18	18	NUM
ejpam-515	223	11	)	)	PUNCT
ejpam-515	223	12	and	and	CCONJ
ejpam-515	223	13	conditions	condition	NOUN
ejpam-515	223	14	(	(	PUNCT
ejpam-515	223	15	c.1	c.1	NOUN
ejpam-515	223	16	)	)	PUNCT
ejpam-515	223	17	and	and	CCONJ
ejpam-515	223	18	(	(	PUNCT
ejpam-515	223	19	c.4	c.4	PROPN
ejpam-515	223	20	)	)	PUNCT
ejpam-515	224	1	that	that	SCONJ
ejpam-515	224	2	µ0n	µ0n	NOUN
ejpam-515	225	1	−	−	NUM
ejpam-515	225	2	ǫdnj	ǫdnj	NOUN
ejpam-515	225	3	x	x	SYM
ejpam-515	225	4	−1	−1	NOUN
ejpam-515	225	5	nj	nj	PROPN
ejpam-515	225	6	=	=	PUNCT
ejpam-515	226	1	p	p	PROPN
ejpam-515	226	2	µ0n	µ0n	NOUN
ejpam-515	226	3	(	(	PUNCT
ejpam-515	226	4	p	p	NOUN
ejpam-515	226	5	µ0n−	µ0n−	X
ejpam-515	226	6	ǫ	ǫ	ADV
ejpam-515	226	7	√	√	NOUN
ejpam-515	226	8	√	√	NUM
ejpam-515	226	9	√	√	PROPN
ejpam-515	226	10	√	√	NUM
ejpam-515	226	11	d2	d2	PROPN
ejpam-515	226	12	nj	nj	PROPN
ejpam-515	226	13	µ0n	µ0n	NOUN
ejpam-515	227	1	x2	x2	PROPN
ejpam-515	227	2	nj	nj	PROPN
ejpam-515	227	3	)	)	PUNCT
ejpam-515	228	1	≤pµ0n(1−	≤pµ0n(1−	PROPN
ejpam-515	228	2	ǫ(b0δ	ǫ(b0δ	ADP
ejpam-515	228	3	2	2	NUM
ejpam-515	228	4	n	n	NOUN
ejpam-515	228	5	log	log	VERB
ejpam-515	228	6	logλp{in(β0)})−	logλp{in(β0)})−	NUM
ejpam-515	228	7	1	1	NUM
ejpam-515	228	8	2	2	NUM
ejpam-515	228	9	)	)	PUNCT
ejpam-515	228	10	<	<	X
ejpam-515	228	11	0	0	PUNCT
ejpam-515	229	1	if	if	SCONJ
ejpam-515	229	2	n	n	NOUN
ejpam-515	229	3	is	be	AUX
ejpam-515	229	4	sufficiently	sufficiently	ADV
ejpam-515	229	5	large	large	ADJ
ejpam-515	229	6	,	,	PUNCT
ejpam-515	229	7	which	which	PRON
ejpam-515	229	8	suggests	suggest	VERB
ejpam-515	229	9	p{yn	p{yn	PROPN
ejpam-515	229	10	<	<	X
ejpam-515	229	11	µ0n	µ0n	NOUN
ejpam-515	229	12	−	−	NUM
ejpam-515	229	13	ǫdnj	ǫdnj	NOUN
ejpam-515	229	14	x	x	SYM
ejpam-515	229	15	−1	−1	NOUN
ejpam-515	229	16	nj	nj	PROPN
ejpam-515	229	17	}	}	PUNCT
ejpam-515	230	1	=	=	PUNCT
ejpam-515	230	2	0	0	PUNCT
ejpam-515	231	1	if	if	SCONJ
ejpam-515	231	2	n	n	NOUN
ejpam-515	231	3	is	be	AUX
ejpam-515	231	4	sufficiently	sufficiently	ADV
ejpam-515	231	5	large	large	ADJ
ejpam-515	231	6	.	.	PUNCT
ejpam-515	232	1	therefore	therefore	ADV
ejpam-515	232	2	the	the	DET
ejpam-515	232	3	result	result	NOUN
ejpam-515	232	4	(	(	PUNCT
ejpam-515	232	5	20	20	NUM
ejpam-515	232	6	)	)	PUNCT
ejpam-515	232	7	is	be	AUX
ejpam-515	232	8	true	true	ADJ
ejpam-515	232	9	for	for	ADP
ejpam-515	232	10	any	any	DET
ejpam-515	232	11	µ0n	µ0n	NOUN
ejpam-515	232	12	>	>	X
ejpam-515	232	13	0	0	NUM
ejpam-515	232	14	,	,	PUNCT
ejpam-515	232	15	which	which	PRON
ejpam-515	232	16	implies	imply	VERB
ejpam-515	232	17	that	that	SCONJ
ejpam-515	232	18	∞	∞	PROPN
ejpam-515	232	19	∑	∑	PROPN
ejpam-515	232	20	n=1	n=1	PROPN
ejpam-515	232	21	p{yn	p{yn	PROPN
ejpam-515	232	22	<	<	X
ejpam-515	232	23	µ0n	µ0n	NOUN
ejpam-515	233	1	−	−	NUM
ejpam-515	233	2	ǫdnj	ǫdnj	NOUN
ejpam-515	233	3	x	x	SYM
ejpam-515	233	4	−1	−1	NOUN
ejpam-515	233	5	nj	nj	PROPN
ejpam-515	233	6	}	}	PUNCT
ejpam-515	233	7	<	<	X
ejpam-515	233	8	∞	∞	PROPN
ejpam-515	233	9	(	(	PUNCT
ejpam-515	233	10	21	21	NUM
ejpam-515	233	11	)	)	PUNCT
ejpam-515	233	12	g.	g.	PROPN
ejpam-515	233	13	qian	qian	PROPN
ejpam-515	233	14	/	/	SYM
ejpam-515	233	15	eur	eur	PROPN
ejpam-515	233	16	.	.	PUNCT
ejpam-515	234	1	j.	j.	PROPN
ejpam-515	234	2	pure	pure	PROPN
ejpam-515	234	3	appl	appl	PROPN
ejpam-515	234	4	.	.	PROPN
ejpam-515	234	5	math	math	PROPN
ejpam-515	234	6	,	,	PUNCT
ejpam-515	234	7	3	3	NUM
ejpam-515	234	8	(	(	PUNCT
ejpam-515	234	9	2010	2010	NUM
ejpam-515	234	10	)	)	PUNCT
ejpam-515	234	11	,	,	PUNCT
ejpam-515	234	12	417	417	NUM
ejpam-515	234	13	-	-	SYM
ejpam-515	234	14	434	434	NUM
ejpam-515	234	15	426	426	NUM
ejpam-515	234	16	if	if	SCONJ
ejpam-515	234	17	further	further	ADJ
ejpam-515	234	18	condition	condition	NOUN
ejpam-515	234	19	(	(	PUNCT
ejpam-515	234	20	c.2	c.2	NOUN
ejpam-515	234	21	)	)	PUNCT
ejpam-515	234	22	holds	hold	VERB
ejpam-515	234	23	.	.	PUNCT
ejpam-515	235	1	now	now	ADV
ejpam-515	235	2	let	let	VERB
ejpam-515	235	3	{	{	PUNCT
ejpam-515	235	4	cn	cn	NOUN
ejpam-515	235	5	:	:	PUNCT
ejpam-515	235	6	0	0	PUNCT
ejpam-515	236	1	<	<	X
ejpam-515	236	2	cn	cn	X
ejpam-515	236	3	≤	≤	ADV
ejpam-515	236	4	1	1	NUM
ejpam-515	236	5	2	2	NUM
ejpam-515	236	6	,	,	PUNCT
ejpam-515	236	7	n	n	NOUN
ejpam-515	236	8	=	=	SYM
ejpam-515	236	9	1,2	1,2	NUM
ejpam-515	236	10	,	,	PUNCT
ejpam-515	236	11	·	·	PUNCT
ejpam-515	236	12	·	·	PUNCT
ejpam-515	236	13	·	·	PUNCT
ejpam-515	236	14	}	}	PUNCT
ejpam-515	236	15	be	be	AUX
ejpam-515	236	16	a	a	DET
ejpam-515	236	17	sequence	sequence	NOUN
ejpam-515	236	18	of	of	ADP
ejpam-515	236	19	numbers	number	NOUN
ejpam-515	236	20	to	to	PART
ejpam-515	236	21	be	be	AUX
ejpam-515	236	22	determined	determine	VERB
ejpam-515	236	23	.	.	PUNCT
ejpam-515	237	1	applying	apply	VERB
ejpam-515	237	2	the	the	DET
ejpam-515	237	3	result	result	NOUN
ejpam-515	237	4	(	(	PUNCT
ejpam-515	237	5	10	10	NUM
ejpam-515	237	6	)	)	PUNCT
ejpam-515	237	7	of	of	ADP
ejpam-515	237	8	lemma	lemma	PROPN
ejpam-515	237	9	2	2	NUM
ejpam-515	237	10	and	and	CCONJ
ejpam-515	237	11	the	the	DET
ejpam-515	237	12	property	property	NOUN
ejpam-515	237	13	for	for	ADP
ejpam-515	237	14	the	the	DET
ejpam-515	237	15	gamma	gamma	NOUN
ejpam-515	237	16	function	function	NOUN
ejpam-515	237	17	[	[	X
ejpam-515	237	18	γ(θ	γ(θ	PROPN
ejpam-515	237	19	+	+	NOUN
ejpam-515	237	20	1)]−1	1)]−1	NUM
ejpam-515	237	21	∫∞	∫∞	NOUN
ejpam-515	237	22	0	0	NUM
ejpam-515	237	23	tθ	tθ	NOUN
ejpam-515	237	24	e−t	e−t	NOUN
ejpam-515	237	25	d	d	X
ejpam-515	237	26	t	t	NOUN
ejpam-515	237	27	=	=	SYM
ejpam-515	237	28	1	1	NUM
ejpam-515	237	29	,	,	PUNCT
ejpam-515	237	30	we	we	PRON
ejpam-515	237	31	have	have	VERB
ejpam-515	237	32	p{yn	p{yn	PROPN
ejpam-515	237	33	>	>	X
ejpam-515	237	34	µ0n+	µ0n+	PROPN
ejpam-515	237	35	ǫdnj	ǫdnj	PROPN
ejpam-515	237	36	x	x	SYM
ejpam-515	237	37	−1	−1	NOUN
ejpam-515	237	38	nj	nj	PROPN
ejpam-515	237	39	}	}	PUNCT
ejpam-515	237	40	≤	≤	NOUN
ejpam-515	238	1	[	[	X
ejpam-515	238	2	γ(µ0n+	γ(µ0n+	PROPN
ejpam-515	238	3	1)]−1	1)]−1	NUM
ejpam-515	238	4	∫	∫	NOUN
ejpam-515	238	5	∞	∞	PROPN
ejpam-515	238	6	µ0n+ǫdn	µ0n+ǫdn	NUM
ejpam-515	238	7	j	j	PROPN
ejpam-515	238	8	x	x	SYM
ejpam-515	238	9	−1	−1	VERB
ejpam-515	238	10	n	n	PRON
ejpam-515	238	11	j	j	NOUN
ejpam-515	238	12	tµ0n	tµ0n	PRON
ejpam-515	238	13	e−t	e−t	VERB
ejpam-515	238	14	d	d	NOUN
ejpam-515	238	15	t	t	NOUN
ejpam-515	238	16	≤	≤	NUM
ejpam-515	238	17	e	e	NOUN
ejpam-515	238	18	−cn(µ0n+ǫdn	−cn(µ0n+ǫdn	VERB
ejpam-515	238	19	j	j	PROPN
ejpam-515	238	20	x	x	SYM
ejpam-515	238	21	−1	−1	VERB
ejpam-515	238	22	n	n	PRON
ejpam-515	238	23	j	j	PROPN
ejpam-515	238	24	)	)	PUNCT
ejpam-515	239	1	[	[	X
ejpam-515	239	2	γ(µ0n+	γ(µ0n+	PROPN
ejpam-515	239	3	1)]−1	1)]−1	NUM
ejpam-515	239	4	∫	∫	NOUN
ejpam-515	239	5	∞	∞	PROPN
ejpam-515	239	6	µ0n+ǫdn	µ0n+ǫdn	NUM
ejpam-515	239	7	j	j	PROPN
ejpam-515	239	8	x	x	SYM
ejpam-515	239	9	−1	−1	VERB
ejpam-515	239	10	n	n	PRON
ejpam-515	239	11	j	j	PROPN
ejpam-515	239	12	tµ0n	tµ0n	NUM
ejpam-515	239	13	e−(1−cn)t	e−(1−cn)t	PUNCT
ejpam-515	240	1	d	d	NOUN
ejpam-515	240	2	t	t	NOUN
ejpam-515	240	3	≤	≤	NUM
ejpam-515	240	4	e	e	NOUN
ejpam-515	240	5	−cn(µ0n+ǫdn	−cn(µ0n+ǫdn	VERB
ejpam-515	240	6	j	j	PROPN
ejpam-515	240	7	x	x	SYM
ejpam-515	240	8	−1	−1	VERB
ejpam-515	240	9	n	n	PRON
ejpam-515	240	10	j	j	PROPN
ejpam-515	240	11	)	)	PUNCT
ejpam-515	240	12	(	(	PUNCT
ejpam-515	240	13	1−	1−	NUM
ejpam-515	240	14	cn	cn	NOUN
ejpam-515	240	15	)	)	PUNCT
ejpam-515	240	16	−(µ0n+1	−(µ0n+1	PROPN
ejpam-515	240	17	)	)	PUNCT
ejpam-515	240	18	=	=	SYM
ejpam-515	241	1	1	1	NUM
ejpam-515	241	2	1−	1−	NUM
ejpam-515	241	3	cn	cn	PROPN
ejpam-515	241	4	�	�	PROPN
ejpam-515	241	5	e−cn	e−cn	PROPN
ejpam-515	241	6	1−	1−	NUM
ejpam-515	241	7	cn	cn	PROPN
ejpam-515	241	8	�	�	PROPN
ejpam-515	241	9	µ0n	µ0n	ADP
ejpam-515	241	10	e	e	PROPN
ejpam-515	241	11	−ǫcndn	−ǫcndn	NOUN
ejpam-515	241	12	j	j	X
ejpam-515	241	13	x	x	SYM
ejpam-515	241	14	−1	−1	VERB
ejpam-515	241	15	n	n	PRON
ejpam-515	241	16	j	j	PROPN
ejpam-515	241	17	≤	≤	ADV
ejpam-515	241	18	2	2	NUM
ejpam-515	241	19	1−	1−	NUM
ejpam-515	241	20	cn	cn	NOUN
ejpam-515	242	1	+	+	CCONJ
ejpam-515	242	2	1	1	NUM
ejpam-515	242	3	2	2	NUM
ejpam-515	242	4	c2	c2	PROPN
ejpam-515	242	5	n	n	PROPN
ejpam-515	242	6	1−	1−	NUM
ejpam-515	242	7	cn	cn	INTJ
ejpam-515	242	8	!	!	PUNCT
ejpam-515	242	9	µ0n	µ0n	PUNCT
ejpam-515	243	1	e	e	NOUN
ejpam-515	243	2	−ǫcndn	−ǫcndn	VERB
ejpam-515	243	3	j	j	X
ejpam-515	243	4	x	x	SYM
ejpam-515	243	5	−1	−1	VERB
ejpam-515	243	6	n	n	PRON
ejpam-515	243	7	j	j	NOUN
ejpam-515	243	8	=	=	SYM
ejpam-515	243	9	2	2	NUM
ejpam-515	243	10	�	�	PROPN
ejpam-515	243	11	1	1	NUM
ejpam-515	243	12	+	+	NOUN
ejpam-515	243	13	c2	c2	PROPN
ejpam-515	243	14	n	n	CCONJ
ejpam-515	243	15	2(1−	2(1−	PROPN
ejpam-515	243	16	cn	cn	PROPN
ejpam-515	243	17	)	)	PUNCT
ejpam-515	243	18	�	�	PROPN
ejpam-515	243	19	µ0n	µ0n	NOUN
ejpam-515	243	20	e	e	PROPN
ejpam-515	243	21	−ǫcndn	−ǫcndn	NOUN
ejpam-515	243	22	j	j	X
ejpam-515	243	23	x	x	SYM
ejpam-515	243	24	−1	−1	VERB
ejpam-515	243	25	n	n	PRON
ejpam-515	243	26	j	j	PROPN
ejpam-515	243	27	.	.	PUNCT
ejpam-515	244	1	(	(	PUNCT
ejpam-515	244	2	22	22	NUM
ejpam-515	244	3	)	)	PUNCT
ejpam-515	244	4	we	we	PRON
ejpam-515	244	5	take	take	VERB
ejpam-515	244	6	cn	cn	PROPN
ejpam-515	244	7	=	=	NOUN
ejpam-515	244	8	min{1	min{1	PROPN
ejpam-515	244	9	2	2	NUM
ejpam-515	244	10	,	,	PUNCT
ejpam-515	244	11	µ	µ	NOUN
ejpam-515	244	12	−	−	NOUN
ejpam-515	244	13	1	1	NUM
ejpam-515	244	14	2	2	NUM
ejpam-515	244	15	0n	0n	NOUN
ejpam-515	244	16	}	}	PUNCT
ejpam-515	244	17	so	so	ADV
ejpam-515	244	18	c2	c2	PROPN
ejpam-515	244	19	n	n	PRON
ejpam-515	244	20	2(1−cn	2(1−cn	NUM
ejpam-515	244	21	)	)	PUNCT
ejpam-515	244	22	≤	≤	NUM
ejpam-515	244	23	min{1	min{1	PROPN
ejpam-515	244	24	4	4	NUM
ejpam-515	244	25	,	,	PUNCT
ejpam-515	244	26	µ−1	µ−1	PROPN
ejpam-515	244	27	0n	0n	NOUN
ejpam-515	244	28	}	}	PUNCT
ejpam-515	244	29	.	.	PUNCT
ejpam-515	245	1	by	by	ADP
ejpam-515	245	2	considering	consider	VERB
ejpam-515	245	3	the	the	DET
ejpam-515	245	4	two	two	NUM
ejpam-515	245	5	cases	case	NOUN
ejpam-515	245	6	µ0n	µ0n	NOUN
ejpam-515	245	7	≤	≤	NUM
ejpam-515	245	8	4	4	NUM
ejpam-515	245	9	and	and	CCONJ
ejpam-515	245	10	µ0n	µ0n	NOUN
ejpam-515	245	11	>	>	X
ejpam-515	245	12	4	4	NUM
ejpam-515	245	13	separately	separately	ADV
ejpam-515	245	14	and	and	CCONJ
ejpam-515	245	15	using	use	VERB
ejpam-515	245	16	the	the	DET
ejpam-515	245	17	property	property	NOUN
ejpam-515	245	18	that	that	PRON
ejpam-515	245	19	(	(	PUNCT
ejpam-515	245	20	1	1	NUM
ejpam-515	245	21	+	+	SYM
ejpam-515	245	22	1	1	NUM
ejpam-515	245	23	a	a	NOUN
ejpam-515	245	24	)	)	PUNCT
ejpam-515	245	25	a	a	DET
ejpam-515	245	26	↑	↑	NOUN
ejpam-515	245	27	e	e	NOUN
ejpam-515	245	28	as	as	ADP
ejpam-515	245	29	a	a	DET
ejpam-515	245	30	↑	↑	NOUN
ejpam-515	245	31	∞	∞	PROPN
ejpam-515	245	32	,	,	PUNCT
ejpam-515	245	33	it	it	PRON
ejpam-515	245	34	is	be	AUX
ejpam-515	245	35	easy	easy	ADJ
ejpam-515	245	36	to	to	PART
ejpam-515	245	37	see	see	VERB
ejpam-515	245	38	that	that	SCONJ
ejpam-515	245	39	�	�	PROPN
ejpam-515	245	40	1	1	NUM
ejpam-515	245	41	+	+	NUM
ejpam-515	245	42	c2	c2	PROPN
ejpam-515	245	43	n	n	CCONJ
ejpam-515	245	44	2(1−	2(1−	PROPN
ejpam-515	245	45	cn	cn	PROPN
ejpam-515	245	46	)	)	PUNCT
ejpam-515	245	47	�	�	PROPN
ejpam-515	245	48	µ0n	µ0n	NOUN
ejpam-515	245	49	≤max	≤max	NUM
ejpam-515	245	50	{	{	PUNCT
ejpam-515	245	51	�	�	PROPN
ejpam-515	245	52	5	5	NUM
ejpam-515	245	53	4	4	NUM
ejpam-515	245	54	�	�	NOUN
ejpam-515	245	55	4	4	NUM
ejpam-515	245	56	,	,	PUNCT
ejpam-515	245	57	e	e	NOUN
ejpam-515	245	58	}	}	PUNCT
ejpam-515	245	59	=	=	SYM
ejpam-515	245	60	e.	e.	PROPN
ejpam-515	245	61	(	(	PUNCT
ejpam-515	245	62	23	23	NUM
ejpam-515	245	63	)	)	PUNCT
ejpam-515	245	64	applying	apply	VERB
ejpam-515	245	65	(	(	PUNCT
ejpam-515	245	66	18	18	NUM
ejpam-515	245	67	)	)	PUNCT
ejpam-515	245	68	one	one	NOUN
ejpam-515	245	69	can	can	AUX
ejpam-515	245	70	show	show	VERB
ejpam-515	245	71	that	that	SCONJ
ejpam-515	245	72	when	when	SCONJ
ejpam-515	245	73	n	n	PRON
ejpam-515	245	74	is	be	AUX
ejpam-515	245	75	sufficiently	sufficiently	ADV
ejpam-515	245	76	large	large	ADJ
ejpam-515	245	77	ǫdnjµ	ǫdnjµ	NOUN
ejpam-515	245	78	−	−	NOUN
ejpam-515	245	79	1	1	NUM
ejpam-515	245	80	2	2	NUM
ejpam-515	245	81	0n	0n	NOUN
ejpam-515	245	82	x−1	x−1	PROPN
ejpam-515	245	83	nj	nj	PROPN
ejpam-515	245	84	≥	≥	NOUN
ejpam-515	245	85	ǫb	ǫb	NUM
ejpam-515	245	86	−	−	NUM
ejpam-515	245	87	1	1	NUM
ejpam-515	245	88	2	2	NUM
ejpam-515	245	89	0	0	NUM
ejpam-515	245	90	(	(	PUNCT
ejpam-515	245	91	δn	δn	PROPN
ejpam-515	245	92	p	p	PROPN
ejpam-515	245	93	log	log	NOUN
ejpam-515	245	94	logλp{in(β0)})−1	logλp{in(β0)})−1	PROPN
ejpam-515	245	95	≥	≥	NUM
ejpam-515	245	96	2	2	NUM
ejpam-515	245	97	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	245	98	)	)	PUNCT
ejpam-515	245	99	}	}	PUNCT
ejpam-515	245	100	(	(	PUNCT
ejpam-515	245	101	24	24	NUM
ejpam-515	245	102	)	)	PUNCT
ejpam-515	245	103	under	under	ADP
ejpam-515	245	104	conditions	condition	NOUN
ejpam-515	245	105	(	(	PUNCT
ejpam-515	245	106	c.1	c.1	NOUN
ejpam-515	245	107	)	)	PUNCT
ejpam-515	245	108	and	and	CCONJ
ejpam-515	245	109	(	(	PUNCT
ejpam-515	245	110	c.5	c.5	NOUN
ejpam-515	245	111	)	)	PUNCT
ejpam-515	245	112	.	.	PUNCT
ejpam-515	246	1	in	in	ADP
ejpam-515	246	2	the	the	DET
ejpam-515	246	3	same	same	ADJ
ejpam-515	246	4	way	way	NOUN
ejpam-515	246	5	as	as	ADP
ejpam-515	246	6	proving	prove	VERB
ejpam-515	246	7	(	(	PUNCT
ejpam-515	246	8	18	18	NUM
ejpam-515	246	9	)	)	PUNCT
ejpam-515	246	10	one	one	NOUN
ejpam-515	246	11	can	can	AUX
ejpam-515	246	12	show	show	VERB
ejpam-515	246	13	that	that	SCONJ
ejpam-515	246	14	under	under	ADP
ejpam-515	246	15	condition	condition	NOUN
ejpam-515	246	16	(	(	PUNCT
ejpam-515	246	17	c.1	c.1	NOUN
ejpam-515	246	18	)	)	PUNCT
ejpam-515	246	19	x2	x2	PROPN
ejpam-515	246	20	nj	nj	PROPN
ejpam-515	246	21	≤	≤	PROPN
ejpam-515	246	22	b0d2	b0d2	X
ejpam-515	246	23	njξ	njξ	PROPN
ejpam-515	246	24	2	2	NUM
ejpam-515	246	25	n	n	NUM
ejpam-515	246	26	log	log	NOUN
ejpam-515	246	27	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	246	28	)	)	PUNCT
ejpam-515	246	29	}	}	PUNCT
ejpam-515	246	30	.	.	PUNCT
ejpam-515	247	1	(	(	PUNCT
ejpam-515	247	2	25	25	NUM
ejpam-515	247	3	)	)	PUNCT
ejpam-515	247	4	by	by	ADP
ejpam-515	247	5	(	(	PUNCT
ejpam-515	247	6	25	25	NUM
ejpam-515	247	7	)	)	PUNCT
ejpam-515	247	8	and	and	CCONJ
ejpam-515	247	9	condition	condition	NOUN
ejpam-515	247	10	(	(	PUNCT
ejpam-515	247	11	c.6	c.6	X
ejpam-515	247	12	)	)	PUNCT
ejpam-515	247	13	it	it	PRON
ejpam-515	247	14	follows	follow	VERB
ejpam-515	247	15	that	that	SCONJ
ejpam-515	247	16	1	1	NUM
ejpam-515	247	17	2	2	NUM
ejpam-515	247	18	ǫdnj	ǫdnj	NOUN
ejpam-515	247	19	x	x	SYM
ejpam-515	247	20	−1	−1	PROPN
ejpam-515	247	21	nj	nj	PROPN
ejpam-515	247	22	≥	≥	NUM
ejpam-515	247	23	1	1	NUM
ejpam-515	247	24	2	2	NUM
ejpam-515	247	25	ǫb	ǫb	NUM
ejpam-515	247	26	−	−	NUM
ejpam-515	247	27	1	1	NUM
ejpam-515	247	28	2	2	NUM
ejpam-515	247	29	0	0	NUM
ejpam-515	247	30	(	(	PUNCT
ejpam-515	247	31	ξn	ξn	PROPN
ejpam-515	247	32	p	p	PROPN
ejpam-515	247	33	log	log	NOUN
ejpam-515	247	34	logλp{in(β0)})−1	logλp{in(β0)})−1	PROPN
ejpam-515	247	35	≥	≥	NUM
ejpam-515	247	36	2	2	NUM
ejpam-515	247	37	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	247	38	)	)	PUNCT
ejpam-515	247	39	}	}	PUNCT
ejpam-515	247	40	(	(	PUNCT
ejpam-515	247	41	26	26	NUM
ejpam-515	247	42	)	)	PUNCT
ejpam-515	247	43	when	when	SCONJ
ejpam-515	247	44	n	n	PRON
ejpam-515	247	45	is	be	AUX
ejpam-515	247	46	sufficiently	sufficiently	ADV
ejpam-515	247	47	large	large	ADJ
ejpam-515	247	48	.	.	PUNCT
ejpam-515	248	1	by	by	ADP
ejpam-515	248	2	(	(	PUNCT
ejpam-515	248	3	24	24	NUM
ejpam-515	248	4	)	)	PUNCT
ejpam-515	248	5	,	,	PUNCT
ejpam-515	248	6	(	(	PUNCT
ejpam-515	248	7	26	26	NUM
ejpam-515	248	8	)	)	PUNCT
ejpam-515	248	9	and	and	CCONJ
ejpam-515	248	10	the	the	DET
ejpam-515	248	11	fact	fact	NOUN
ejpam-515	248	12	that	that	SCONJ
ejpam-515	248	13	cn	cn	PROPN
ejpam-515	248	14	=	=	NOUN
ejpam-515	248	15	min{1	min{1	PROPN
ejpam-515	248	16	2	2	NUM
ejpam-515	248	17	,	,	PUNCT
ejpam-515	248	18	µ	µ	NOUN
ejpam-515	248	19	−	−	NOUN
ejpam-515	248	20	1	1	NUM
ejpam-515	248	21	2	2	NUM
ejpam-515	248	22	0n	0n	NOUN
ejpam-515	248	23	}	}	PUNCT
ejpam-515	248	24	it	it	PRON
ejpam-515	248	25	follows	follow	VERB
ejpam-515	248	26	that	that	SCONJ
ejpam-515	248	27	when	when	SCONJ
ejpam-515	248	28	n	n	PRON
ejpam-515	248	29	is	be	AUX
ejpam-515	248	30	sufficiently	sufficiently	ADV
ejpam-515	248	31	large	large	ADJ
ejpam-515	248	32	,	,	PUNCT
ejpam-515	248	33	ǫcndnj	ǫcndnj	PROPN
ejpam-515	248	34	x	x	SYM
ejpam-515	248	35	−1	−1	PROPN
ejpam-515	248	36	nj	nj	PROPN
ejpam-515	248	37	≥	≥	PROPN
ejpam-515	248	38	2	2	NUM
ejpam-515	248	39	logλp{in(β0	logλp{in(β0	NOUN
ejpam-515	248	40	)	)	PUNCT
ejpam-515	248	41	}	}	PUNCT
ejpam-515	248	42	.	.	PUNCT
ejpam-515	249	1	(	(	PUNCT
ejpam-515	249	2	27	27	NUM
ejpam-515	249	3	)	)	PUNCT
ejpam-515	249	4	now	now	ADV
ejpam-515	249	5	from	from	ADP
ejpam-515	249	6	(	(	PUNCT
ejpam-515	249	7	22	22	NUM
ejpam-515	249	8	)	)	PUNCT
ejpam-515	249	9	,	,	PUNCT
ejpam-515	249	10	(	(	PUNCT
ejpam-515	249	11	23	23	NUM
ejpam-515	249	12	)	)	PUNCT
ejpam-515	249	13	and	and	CCONJ
ejpam-515	249	14	(	(	PUNCT
ejpam-515	249	15	27	27	NUM
ejpam-515	249	16	)	)	PUNCT
ejpam-515	249	17	we	we	PRON
ejpam-515	249	18	have	have	VERB
ejpam-515	249	19	p{yn	p{yn	PROPN
ejpam-515	249	20	>	>	X
ejpam-515	249	21	µ0n+	µ0n+	PROPN
ejpam-515	249	22	ǫdnj	ǫdnj	PROPN
ejpam-515	249	23	x	x	SYM
ejpam-515	249	24	−1	−1	NOUN
ejpam-515	249	25	nj	nj	PROPN
ejpam-515	249	26	}	}	PUNCT
ejpam-515	249	27	≤	≤	ADV
ejpam-515	249	28	2eλp{in(β0)}−2	2eλp{in(β0)}−2	NUM
ejpam-515	249	29	(	(	PUNCT
ejpam-515	249	30	28	28	NUM
ejpam-515	249	31	)	)	PUNCT
ejpam-515	249	32	g.	g.	PROPN
ejpam-515	249	33	qian	qian	PROPN
ejpam-515	249	34	/	/	SYM
ejpam-515	249	35	eur	eur	PROPN
ejpam-515	249	36	.	.	PUNCT
ejpam-515	250	1	j.	j.	PROPN
ejpam-515	250	2	pure	pure	PROPN
ejpam-515	250	3	appl	appl	PROPN
ejpam-515	250	4	.	.	PROPN
ejpam-515	250	5	math	math	PROPN
ejpam-515	250	6	,	,	PUNCT
ejpam-515	250	7	3	3	NUM
ejpam-515	250	8	(	(	PUNCT
ejpam-515	250	9	2010	2010	NUM
ejpam-515	250	10	)	)	PUNCT
ejpam-515	250	11	,	,	PUNCT
ejpam-515	250	12	417	417	NUM
ejpam-515	250	13	-	-	SYM
ejpam-515	250	14	434	434	NUM
ejpam-515	250	15	427	427	NUM
ejpam-515	250	16	when	when	SCONJ
ejpam-515	250	17	n	n	X
ejpam-515	250	18	is	be	AUX
ejpam-515	250	19	sufficiently	sufficiently	ADV
ejpam-515	250	20	large	large	ADJ
ejpam-515	250	21	.	.	PUNCT
ejpam-515	251	1	from	from	ADP
ejpam-515	251	2	(	(	PUNCT
ejpam-515	251	3	28	28	NUM
ejpam-515	251	4	)	)	PUNCT
ejpam-515	251	5	and	and	CCONJ
ejpam-515	251	6	condition	condition	NOUN
ejpam-515	251	7	(	(	PUNCT
ejpam-515	251	8	c.2	c.2	PROPN
ejpam-515	251	9	)	)	PUNCT
ejpam-515	251	10	it	it	PRON
ejpam-515	251	11	follows	follow	VERB
ejpam-515	251	12	that	that	SCONJ
ejpam-515	251	13	∞	∞	PROPN
ejpam-515	251	14	∑	∑	PROPN
ejpam-515	251	15	n=1	n=1	PROPN
ejpam-515	251	16	p{yn	p{yn	PROPN
ejpam-515	251	17	>	>	X
ejpam-515	251	18	µ0n+	µ0n+	PROPN
ejpam-515	251	19	ǫdnj	ǫdnj	PROPN
ejpam-515	251	20	x	x	SYM
ejpam-515	251	21	−1	−1	NOUN
ejpam-515	251	22	nj	nj	PROPN
ejpam-515	251	23	}	}	PUNCT
ejpam-515	252	1	<	<	X
ejpam-515	252	2	∞.	∞.	PROPN
ejpam-515	252	3	(	(	PUNCT
ejpam-515	252	4	29	29	NUM
ejpam-515	252	5	)	)	PUNCT
ejpam-515	252	6	following	follow	VERB
ejpam-515	252	7	the	the	DET
ejpam-515	252	8	results	result	NOUN
ejpam-515	252	9	(	(	PUNCT
ejpam-515	252	10	16	16	NUM
ejpam-515	252	11	)	)	PUNCT
ejpam-515	252	12	,	,	PUNCT
ejpam-515	252	13	(	(	PUNCT
ejpam-515	252	14	21	21	NUM
ejpam-515	252	15	)	)	PUNCT
ejpam-515	252	16	and	and	CCONJ
ejpam-515	252	17	(	(	PUNCT
ejpam-515	252	18	29	29	NUM
ejpam-515	252	19	)	)	PUNCT
ejpam-515	252	20	we	we	PRON
ejpam-515	252	21	have	have	VERB
ejpam-515	252	22	∞	∞	PROPN
ejpam-515	252	23	∑	∑	PUNCT
ejpam-515	252	24	n=1	n=1	PROPN
ejpam-515	252	25	p{|(yn	p{|(yn	PROPN
ejpam-515	252	26	−µ0n)xnj|	−µ0n)xnj|	PROPN
ejpam-515	252	27	>	>	X
ejpam-515	252	28	ǫdnj}<∞.	ǫdnj}<∞.	PROPN
ejpam-515	252	29	hence	hence	ADV
ejpam-515	252	30	by	by	ADP
ejpam-515	252	31	the	the	DET
ejpam-515	252	32	borel	borel	PROPN
ejpam-515	252	33	-	-	PUNCT
ejpam-515	252	34	cantelli	cantelli	PROPN
ejpam-515	252	35	lemma	lemma	PROPN
ejpam-515	252	36	,	,	PUNCT
ejpam-515	252	37	p{|(yn−µ0n)xnj|	p{|(yn−µ0n)xnj|	PROPN
ejpam-515	252	38	>	>	X
ejpam-515	252	39	ǫdnj	ǫdnj	PROPN
ejpam-515	252	40	occurs	occur	VERB
ejpam-515	252	41	infinitely	infinitely	ADV
ejpam-515	252	42	often	often	ADV
ejpam-515	252	43	}	}	PUNCT
ejpam-515	252	44	=	=	SYM
ejpam-515	252	45	0	0	NUM
ejpam-515	252	46	for	for	ADP
ejpam-515	252	47	any	any	DET
ejpam-515	252	48	ǫ	ǫ	NOUN
ejpam-515	252	49	>	>	X
ejpam-515	252	50	0	0	NUM
ejpam-515	252	51	,	,	PUNCT
ejpam-515	252	52	which	which	PRON
ejpam-515	252	53	implies	imply	VERB
ejpam-515	252	54	that	that	SCONJ
ejpam-515	252	55	(	(	PUNCT
ejpam-515	252	56	15	15	NUM
ejpam-515	252	57	)	)	PUNCT
ejpam-515	252	58	is	be	AUX
ejpam-515	252	59	true	true	ADJ
ejpam-515	252	60	.	.	PUNCT
ejpam-515	253	1	since	since	SCONJ
ejpam-515	253	2	(	(	PUNCT
ejpam-515	253	3	13	13	NUM
ejpam-515	253	4	)	)	PUNCT
ejpam-515	253	5	to	to	AUX
ejpam-515	253	6	(	(	PUNCT
ejpam-515	253	7	15	15	NUM
ejpam-515	253	8	)	)	PUNCT
ejpam-515	253	9	are	be	AUX
ejpam-515	253	10	true	true	ADJ
ejpam-515	253	11	,	,	PUNCT
ejpam-515	253	12	the	the	DET
ejpam-515	253	13	result	result	NOUN
ejpam-515	253	14	(	(	PUNCT
ejpam-515	253	15	11	11	NUM
ejpam-515	253	16	)	)	PUNCT
ejpam-515	253	17	is	be	AUX
ejpam-515	253	18	followed	follow	VERB
ejpam-515	253	19	by	by	ADP
ejpam-515	253	20	applying	apply	VERB
ejpam-515	253	21	lemma	lemma	PROPN
ejpam-515	253	22	3	3	NUM
ejpam-515	253	23	for	for	ADP
ejpam-515	253	24	the	the	DET
ejpam-515	253	25	independent	independent	ADJ
ejpam-515	253	26	random	random	ADJ
ejpam-515	253	27	variables	variable	NOUN
ejpam-515	253	28	{	{	PUNCT
ejpam-515	253	29	(	(	PUNCT
ejpam-515	253	30	yk	yk	PROPN
ejpam-515	253	31	−µ0k)xk	−µ0k)xk	PROPN
ejpam-515	253	32	j	j	PROPN
ejpam-515	253	33	,	,	PUNCT
ejpam-515	253	34	k	k	PROPN
ejpam-515	253	35	=	=	SYM
ejpam-515	253	36	1,2	1,2	NUM
ejpam-515	253	37	,	,	PUNCT
ejpam-515	253	38	·	·	PUNCT
ejpam-515	253	39	·	·	PUNCT
ejpam-515	253	40	·	·	PUNCT
ejpam-515	253	41	}	}	PUNCT
ejpam-515	253	42	.	.	PUNCT
ejpam-515	254	1	proof	proof	NOUN
ejpam-515	254	2	.	.	PUNCT
ejpam-515	255	1	(	(	PUNCT
ejpam-515	255	2	theorem	theorem	NOUN
ejpam-515	255	3	1	1	NUM
ejpam-515	255	4	)	)	PUNCT
ejpam-515	255	5	clearly	clearly	ADV
ejpam-515	255	6	it	it	PRON
ejpam-515	255	7	is	be	AUX
ejpam-515	255	8	sufficient	sufficient	ADJ
ejpam-515	255	9	to	to	PART
ejpam-515	255	10	prove	prove	VERB
ejpam-515	255	11	(	(	PUNCT
ejpam-515	255	12	4	4	NUM
ejpam-515	255	13	)	)	PUNCT
ejpam-515	255	14	only	only	ADV
ejpam-515	255	15	for	for	ADP
ejpam-515	255	16	the	the	DET
ejpam-515	255	17	full	full	ADJ
ejpam-515	255	18	model	model	NOUN
ejpam-515	255	19	:	:	PUNCT
ejpam-515	255	20	||β̂	||β̂	X
ejpam-515	256	1	−β0||=	−β0||=	X
ejpam-515	257	1	o	o	INTJ
ejpam-515	257	2	(	(	PUNCT
ejpam-515	257	3	p	p	PROPN
ejpam-515	257	4	n−1	n−1	PROPN
ejpam-515	257	5	log	log	NOUN
ejpam-515	257	6	log	log	NOUN
ejpam-515	257	7	n	n	CCONJ
ejpam-515	257	8	)	)	PUNCT
ejpam-515	257	9	a.s	a.s	PROPN
ejpam-515	257	10	.	.	PROPN
ejpam-515	257	11	(	(	PUNCT
ejpam-515	257	12	30	30	NUM
ejpam-515	257	13	)	)	PUNCT
ejpam-515	257	14	applying	apply	VERB
ejpam-515	257	15	result	result	NOUN
ejpam-515	257	16	(	(	PUNCT
ejpam-515	257	17	iii	iii	NOUN
ejpam-515	257	18	)	)	PUNCT
ejpam-515	257	19	of	of	ADP
ejpam-515	257	20	lemma	lemma	PROPN
ejpam-515	257	21	1	1	NUM
ejpam-515	257	22	with	with	ADP
ejpam-515	257	23	t	t	PROPN
ejpam-515	257	24	=	=	SYM
ejpam-515	257	25	xt	xt	PROPN
ejpam-515	257	26	k	k	PROPN
ejpam-515	257	27	β	β	X
ejpam-515	257	28	,	,	PUNCT
ejpam-515	257	29	s	s	PROPN
ejpam-515	257	30	=	=	X
ejpam-515	257	31	xt	xt	PROPN
ejpam-515	257	32	k	k	PROPN
ejpam-515	257	33	β0	β0	PROPN
ejpam-515	257	34	and	and	CCONJ
ejpam-515	257	35	∆=	∆=	VERB
ejpam-515	257	36	|xt	|xt	X
ejpam-515	257	37	k	k	X
ejpam-515	257	38	β	β	X
ejpam-515	257	39	−	−	PROPN
ejpam-515	257	40	xt	xt	PROPN
ejpam-515	258	1	k	k	PROPN
ejpam-515	258	2	β0|	β0|	PROPN
ejpam-515	258	3	,	,	PUNCT
ejpam-515	258	4	it	it	PRON
ejpam-515	258	5	follows	follow	VERB
ejpam-515	258	6	that	that	SCONJ
ejpam-515	258	7	k(xt	k(xt	NOUN
ejpam-515	258	8	k	k	PROPN
ejpam-515	258	9	β	β	X
ejpam-515	258	10	,	,	PUNCT
ejpam-515	258	11	xt	xt	PROPN
ejpam-515	258	12	k	k	PROPN
ejpam-515	259	1	β0)≥	β0)≥	PRON
ejpam-515	259	2	1	1	NUM
ejpam-515	259	3	2	2	NUM
ejpam-515	259	4	e−2|xt	e−2|xt	X
ejpam-515	259	5	k	k	X
ejpam-515	259	6	(	(	PUNCT
ejpam-515	259	7	β−β0)|µ0k[x	β−β0)|µ0k[x	PROPN
ejpam-515	259	8	t	t	PROPN
ejpam-515	259	9	k	k	X
ejpam-515	259	10	(	(	PUNCT
ejpam-515	259	11	β	β	X
ejpam-515	259	12	−β0	−β0	PROPN
ejpam-515	259	13	)	)	PUNCT
ejpam-515	259	14	]	]	PUNCT
ejpam-515	259	15	2	2	X
ejpam-515	259	16	.	.	PUNCT
ejpam-515	259	17	(	(	PUNCT
ejpam-515	259	18	31	31	NUM
ejpam-515	259	19	)	)	PUNCT
ejpam-515	259	20	following	follow	VERB
ejpam-515	259	21	the	the	DET
ejpam-515	259	22	definition	definition	NOUN
ejpam-515	259	23	of	of	ADP
ejpam-515	259	24	ξn	ξn	NOUN
ejpam-515	259	25	and	and	CCONJ
ejpam-515	259	26	condition	condition	NOUN
ejpam-515	259	27	(	(	PUNCT
ejpam-515	259	28	c.2	c.2	NOUN
ejpam-515	259	29	)	)	PUNCT
ejpam-515	259	30	one	one	PRON
ejpam-515	259	31	can	can	AUX
ejpam-515	259	32	find	find	VERB
ejpam-515	259	33	that	that	SCONJ
ejpam-515	259	34	max	max	PROPN
ejpam-515	259	35	1≤k≤n	1≤k≤n	NUM
ejpam-515	259	36	||xk||2	||xk||2	PROPN
ejpam-515	259	37	≤	≤	PROPN
ejpam-515	259	38	λp{in(β0	λp{in(β0	NUM
ejpam-515	259	39	)	)	PUNCT
ejpam-515	259	40	}	}	PUNCT
ejpam-515	259	41	max	max	PROPN
ejpam-515	260	1	1≤k≤n	1≤k≤n	NUM
ejpam-515	260	2	xt	xt	PROPN
ejpam-515	260	3	k	k	PROPN
ejpam-515	260	4	in(β0	in(β0	PROPN
ejpam-515	260	5	)	)	PUNCT
ejpam-515	260	6	−1xk	−1xk	X
ejpam-515	261	1	=	=	PUNCT
ejpam-515	261	2	λp{in(β0)}ξ2	λp{in(β0)}ξ2	SYM
ejpam-515	261	3	n	n	CCONJ
ejpam-515	261	4	≤	≤	ADV
ejpam-515	261	5	b2nξ2	b2nξ2	NOUN
ejpam-515	261	6	n.	n.	NOUN
ejpam-515	261	7	(	(	PUNCT
ejpam-515	261	8	32	32	NUM
ejpam-515	261	9	)	)	PUNCT
ejpam-515	261	10	thus	thus	ADV
ejpam-515	261	11	by	by	ADP
ejpam-515	261	12	(	(	PUNCT
ejpam-515	261	13	32	32	NUM
ejpam-515	261	14	)	)	PUNCT
ejpam-515	261	15	and	and	CCONJ
ejpam-515	261	16	cauchy	cauchy	PROPN
ejpam-515	261	17	-	-	PUNCT
ejpam-515	261	18	schwarz	schwarz	PROPN
ejpam-515	261	19	inequality	inequality	NOUN
ejpam-515	261	20	,	,	PUNCT
ejpam-515	261	21	max	max	PROPN
ejpam-515	261	22	1≤k≤n	1≤k≤n	NUM
ejpam-515	261	23	|xt	|xt	X
ejpam-515	261	24	k	k	X
ejpam-515	261	25	(	(	PUNCT
ejpam-515	261	26	β−β0)|i(β∈∂an	β−β0)|i(β∈∂an	NOUN
ejpam-515	261	27	)	)	PUNCT
ejpam-515	261	28	≤	≤	NUM
ejpam-515	261	29	max	max	PROPN
ejpam-515	261	30	1≤k≤n	1≤k≤n	NUM
ejpam-515	261	31	||xk||	||xk||	NOUN
ejpam-515	261	32	·	·	PUNCT
ejpam-515	262	1	||β−β0||i(β∈∂an)≤	||β−β0||i(β∈∂an)≤	ADV
ejpam-515	262	2	p	p	NOUN
ejpam-515	262	3	b2ξnτn	b2ξnτn	PROPN
ejpam-515	262	4	p	p	PROPN
ejpam-515	262	5	log	log	NOUN
ejpam-515	262	6	log	log	NOUN
ejpam-515	262	7	n	n	PROPN
ejpam-515	262	8	(	(	PUNCT
ejpam-515	262	9	33	33	NUM
ejpam-515	262	10	)	)	PUNCT
ejpam-515	262	11	where	where	SCONJ
ejpam-515	262	12	i(β	i(β	PROPN
ejpam-515	262	13	∈	∈	PROPN
ejpam-515	262	14	∂an	∂an	PROPN
ejpam-515	262	15	)	)	PUNCT
ejpam-515	262	16	is	be	AUX
ejpam-515	262	17	an	an	DET
ejpam-515	262	18	indicator	indicator	NOUN
ejpam-515	262	19	function	function	NOUN
ejpam-515	262	20	indicating	indicate	VERB
ejpam-515	262	21	that	that	SCONJ
ejpam-515	262	22	only	only	ADV
ejpam-515	262	23	those	those	DET
ejpam-515	262	24	β	β	NOUN
ejpam-515	262	25	in	in	ADP
ejpam-515	262	26	∂an	∂an	PROPN
ejpam-515	262	27	will	will	AUX
ejpam-515	262	28	be	be	AUX
ejpam-515	262	29	under	under	ADP
ejpam-515	262	30	consideration	consideration	NOUN
ejpam-515	262	31	.	.	PUNCT
ejpam-515	263	1	(	(	PUNCT
ejpam-515	263	2	this	this	DET
ejpam-515	263	3	type	type	NOUN
ejpam-515	263	4	of	of	ADP
ejpam-515	263	5	definition	definition	NOUN
ejpam-515	263	6	for	for	ADP
ejpam-515	263	7	the	the	DET
ejpam-515	263	8	indicator	indicator	NOUN
ejpam-515	263	9	function	function	NOUN
ejpam-515	263	10	will	will	AUX
ejpam-515	263	11	be	be	AUX
ejpam-515	263	12	used	use	VERB
ejpam-515	263	13	in	in	ADP
ejpam-515	263	14	the	the	DET
ejpam-515	263	15	rest	rest	NOUN
ejpam-515	263	16	of	of	ADP
ejpam-515	263	17	the	the	DET
ejpam-515	263	18	paper	paper	NOUN
ejpam-515	263	19	.	.	PUNCT
ejpam-515	263	20	)	)	PUNCT
ejpam-515	264	1	it	it	PRON
ejpam-515	264	2	follows	follow	VERB
ejpam-515	264	3	from	from	ADP
ejpam-515	264	4	(	(	PUNCT
ejpam-515	264	5	8)	8)	NUM
ejpam-515	264	6	and	and	CCONJ
ejpam-515	264	7	(	(	PUNCT
ejpam-515	264	8	31	31	NUM
ejpam-515	264	9	)	)	PUNCT
ejpam-515	264	10	to	to	ADP
ejpam-515	264	11	(	(	PUNCT
ejpam-515	264	12	33	33	NUM
ejpam-515	264	13	)	)	PUNCT
ejpam-515	264	14	that	that	SCONJ
ejpam-515	264	15	r1(β	r1(β	NOUN
ejpam-515	264	16	,	,	PUNCT
ejpam-515	264	17	n)i(β∈∂an)≥	n)i(β∈∂an)≥	NOUN
ejpam-515	264	18	1	1	NUM
ejpam-515	264	19	2	2	NUM
ejpam-515	264	20	e−2max1≤k≤n	e−2max1≤k≤n	PROPN
ejpam-515	264	21	|xt	|xt	X
ejpam-515	264	22	k	k	X
ejpam-515	264	23	(	(	PUNCT
ejpam-515	264	24	β−β0)|	β−β0)|	NUM
ejpam-515	264	25	n	n	PROPN
ejpam-515	264	26	∑	∑	PUNCT
ejpam-515	264	27	k=1	k=1	PROPN
ejpam-515	264	28	µ0k[x	µ0k[x	PROPN
ejpam-515	264	29	t	t	PROPN
ejpam-515	264	30	k(β	k(β	PROPN
ejpam-515	264	31	−	−	PROPN
ejpam-515	264	32	β0	β0	PROPN
ejpam-515	264	33	)	)	PUNCT
ejpam-515	264	34	]	]	PUNCT
ejpam-515	265	1	2i(β∈∂an	2i(β∈∂an	NUM
ejpam-515	265	2	)	)	PUNCT
ejpam-515	265	3	≥	≥	NOUN
ejpam-515	265	4	1	1	NUM
ejpam-515	265	5	2	2	X
ejpam-515	265	6	e−2	e−2	PROPN
ejpam-515	265	7	p	p	NOUN
ejpam-515	265	8	b2ξnτn	b2ξnτn	PROPN
ejpam-515	265	9	p	p	PROPN
ejpam-515	265	10	log	log	NOUN
ejpam-515	265	11	log	log	NOUN
ejpam-515	265	12	n(β	n(β	PROPN
ejpam-515	265	13	−	−	PROPN
ejpam-515	265	14	β0	β0	PROPN
ejpam-515	265	15	)	)	PUNCT
ejpam-515	265	16	t	t	PROPN
ejpam-515	265	17	in(β0)(β	in(β0)(β	VERB
ejpam-515	265	18	−	−	PROPN
ejpam-515	265	19	β0)i(β∈∂an	β0)i(β∈∂an	NUM
ejpam-515	265	20	)	)	PUNCT
ejpam-515	265	21	≥	≥	NOUN
ejpam-515	265	22	1	1	NUM
ejpam-515	265	23	2	2	X
ejpam-515	265	24	e−2	e−2	PROPN
ejpam-515	265	25	p	p	NOUN
ejpam-515	265	26	b2ξnτn	b2ξnτn	PROPN
ejpam-515	265	27	p	p	PROPN
ejpam-515	265	28	log	log	NOUN
ejpam-515	265	29	log	log	VERB
ejpam-515	265	30	nλ1{in(β0)}||β	nλ1{in(β0)}||β	NOUN
ejpam-515	265	31	−	−	NOUN
ejpam-515	265	32	β0||2i(β∈∂an	β0||2i(β∈∂an	NUM
ejpam-515	265	33	)	)	PUNCT
ejpam-515	265	34	.	.	PUNCT
ejpam-515	266	1	(	(	PUNCT
ejpam-515	266	2	34	34	NUM
ejpam-515	266	3	)	)	PUNCT
ejpam-515	266	4	g.	g.	PROPN
ejpam-515	266	5	qian	qian	PROPN
ejpam-515	266	6	/	/	SYM
ejpam-515	266	7	eur	eur	PROPN
ejpam-515	266	8	.	.	PUNCT
ejpam-515	267	1	j.	j.	PROPN
ejpam-515	267	2	pure	pure	PROPN
ejpam-515	267	3	appl	appl	PROPN
ejpam-515	267	4	.	.	PROPN
ejpam-515	267	5	math	math	PROPN
ejpam-515	267	6	,	,	PUNCT
ejpam-515	267	7	3	3	NUM
ejpam-515	267	8	(	(	PUNCT
ejpam-515	267	9	2010	2010	NUM
ejpam-515	267	10	)	)	PUNCT
ejpam-515	267	11	,	,	PUNCT
ejpam-515	267	12	417	417	NUM
ejpam-515	267	13	-	-	SYM
ejpam-515	267	14	434	434	NUM
ejpam-515	267	15	428	428	NUM
ejpam-515	267	16	by	by	ADP
ejpam-515	267	17	conditions	condition	NOUN
ejpam-515	267	18	(	(	PUNCT
ejpam-515	267	19	c.1	c.1	NOUN
ejpam-515	267	20	)	)	PUNCT
ejpam-515	267	21	and	and	CCONJ
ejpam-515	267	22	(	(	PUNCT
ejpam-515	267	23	c.2	c.2	PROPN
ejpam-515	267	24	)	)	PUNCT
ejpam-515	267	25	and	and	CCONJ
ejpam-515	267	26	the	the	DET
ejpam-515	267	27	fact	fact	NOUN
ejpam-515	267	28	that	that	PRON
ejpam-515	267	29	τnξn	τnξn	VERB
ejpam-515	267	30	p	p	NOUN
ejpam-515	267	31	log	log	NOUN
ejpam-515	267	32	log	log	NOUN
ejpam-515	267	33	n→	n→	ADV
ejpam-515	267	34	0	0	NUM
ejpam-515	267	35	,	,	PUNCT
ejpam-515	267	36	it	it	PRON
ejpam-515	267	37	comes	come	VERB
ejpam-515	267	38	after	after	ADP
ejpam-515	267	39	(	(	PUNCT
ejpam-515	267	40	34	34	NUM
ejpam-515	267	41	)	)	PUNCT
ejpam-515	267	42	that	that	SCONJ
ejpam-515	267	43	there	there	PRON
ejpam-515	267	44	exists	exist	VERB
ejpam-515	267	45	a	a	DET
ejpam-515	267	46	constant	constant	ADJ
ejpam-515	267	47	b4	b4	NOUN
ejpam-515	267	48	>	>	X
ejpam-515	267	49	0	0	NUM
ejpam-515	268	1	such	such	ADJ
ejpam-515	268	2	that	that	DET
ejpam-515	268	3	r1(β	r1(β	PROPN
ejpam-515	268	4	,	,	PUNCT
ejpam-515	268	5	n)i(β∈∂an)≥	n)i(β∈∂an)≥	NOUN
ejpam-515	268	6	b4τ	b4τ	ADJ
ejpam-515	268	7	2	2	NUM
ejpam-515	268	8	n	n	PRON
ejpam-515	268	9	log	log	NOUN
ejpam-515	268	10	log	log	NOUN
ejpam-515	268	11	n.	n.	NOUN
ejpam-515	268	12	(	(	PUNCT
ejpam-515	268	13	35	35	NUM
ejpam-515	268	14	)	)	PUNCT
ejpam-515	268	15	on	on	ADP
ejpam-515	268	16	the	the	DET
ejpam-515	268	17	other	other	ADJ
ejpam-515	268	18	hand	hand	NOUN
ejpam-515	268	19	,	,	PUNCT
ejpam-515	268	20	by	by	ADP
ejpam-515	268	21	result	result	NOUN
ejpam-515	268	22	(	(	PUNCT
ejpam-515	268	23	12	12	NUM
ejpam-515	268	24	)	)	PUNCT
ejpam-515	268	25	of	of	ADP
ejpam-515	268	26	lemma	lemma	PROPN
ejpam-515	268	27	4	4	NUM
ejpam-515	268	28	and	and	CCONJ
ejpam-515	268	29	(	(	PUNCT
ejpam-515	268	30	8)	8)	NUM
ejpam-515	268	31	,	,	PUNCT
ejpam-515	268	32	|r2(β	|r2(β	NUM
ejpam-515	268	33	,	,	PUNCT
ejpam-515	268	34	n)|i(β∈∂an)≤	n)|i(β∈∂an)≤	PROPN
ejpam-515	268	35	||	||	PROPN
ejpam-515	269	1	n	n	PROPN
ejpam-515	269	2	∑	∑	ADV
ejpam-515	269	3	k=1	k=1	PROPN
ejpam-515	269	4	(	(	PUNCT
ejpam-515	269	5	yk	yk	PROPN
ejpam-515	269	6	−µ0k)xk||	−µ0k)xk||	PROPN
ejpam-515	269	7	·	·	PUNCT
ejpam-515	269	8	||β	||β	NOUN
ejpam-515	269	9	−	−	NOUN
ejpam-515	269	10	β0||i(β∈∂an	β0||i(β∈∂an	NOUN
ejpam-515	269	11	)	)	PUNCT
ejpam-515	270	1	=	=	SYM
ejpam-515	270	2	o	o	NOUN
ejpam-515	270	3	(	(	PUNCT
ejpam-515	270	4	p	p	NOUN
ejpam-515	270	5	n	n	NUM
ejpam-515	270	6	log	log	VERB
ejpam-515	270	7	log	log	NOUN
ejpam-515	271	1	n)τn	n)τn	PROPN
ejpam-515	272	1	p	p	PRON
ejpam-515	272	2	n−1	n−1	PROPN
ejpam-515	272	3	log	log	NOUN
ejpam-515	272	4	log	log	NOUN
ejpam-515	272	5	n=	n=	ADJ
ejpam-515	272	6	o(1)τn	o(1)τn	ADJ
ejpam-515	272	7	log	log	NOUN
ejpam-515	272	8	log	log	NOUN
ejpam-515	272	9	n	n	PRON
ejpam-515	272	10	a.s	a.s	PROPN
ejpam-515	272	11	..	..	PUNCT
ejpam-515	272	12	(	(	PUNCT
ejpam-515	272	13	36	36	NUM
ejpam-515	272	14	)	)	PUNCT
ejpam-515	272	15	knowing	know	VERB
ejpam-515	272	16	(	(	PUNCT
ejpam-515	272	17	35	35	NUM
ejpam-515	272	18	)	)	PUNCT
ejpam-515	272	19	and	and	CCONJ
ejpam-515	272	20	(	(	PUNCT
ejpam-515	272	21	36	36	NUM
ejpam-515	272	22	)	)	PUNCT
ejpam-515	272	23	we	we	PRON
ejpam-515	272	24	can	can	AUX
ejpam-515	272	25	find	find	VERB
ejpam-515	272	26	a	a	DET
ejpam-515	272	27	constant	constant	ADJ
ejpam-515	272	28	b5	b5	PROPN
ejpam-515	272	29	>	>	X
ejpam-515	272	30	0	0	PUNCT
ejpam-515	273	1	so	so	SCONJ
ejpam-515	273	2	that	that	SCONJ
ejpam-515	273	3	h(β	h(β	PROPN
ejpam-515	273	4	,	,	PUNCT
ejpam-515	273	5	n)i(β∈∂an	n)i(β∈∂an	PROPN
ejpam-515	273	6	)	)	PUNCT
ejpam-515	273	7	=	=	SYM
ejpam-515	273	8	{	{	PUNCT
ejpam-515	273	9	r1(β	r1(β	PROPN
ejpam-515	273	10	,	,	PUNCT
ejpam-515	273	11	n	n	CCONJ
ejpam-515	273	12	)	)	PUNCT
ejpam-515	273	13	+	+	CCONJ
ejpam-515	273	14	r2(β	r2(β	PROPN
ejpam-515	273	15	,	,	PUNCT
ejpam-515	273	16	n)}i(β∈∂an)≥	n)}i(β∈∂an)≥	NOUN
ejpam-515	273	17	b5τ	b5τ	ADP
ejpam-515	273	18	2	2	NUM
ejpam-515	273	19	n	n	NOUN
ejpam-515	273	20	log	log	NOUN
ejpam-515	273	21	log	log	NOUN
ejpam-515	273	22	n	n	PRON
ejpam-515	273	23	a.s	a.s	PROPN
ejpam-515	273	24	..	..	PUNCT
ejpam-515	273	25	(	(	PUNCT
ejpam-515	273	26	37	37	NUM
ejpam-515	273	27	)	)	PUNCT
ejpam-515	273	28	it	it	PRON
ejpam-515	273	29	is	be	AUX
ejpam-515	273	30	easy	easy	ADJ
ejpam-515	273	31	to	to	PART
ejpam-515	273	32	see	see	VERB
ejpam-515	273	33	that	that	SCONJ
ejpam-515	273	34	h(β	h(β	PROPN
ejpam-515	273	35	,	,	PUNCT
ejpam-515	273	36	n	n	CCONJ
ejpam-515	273	37	)	)	PUNCT
ejpam-515	273	38	is	be	AUX
ejpam-515	273	39	convex	convex	ADJ
ejpam-515	273	40	by	by	ADP
ejpam-515	273	41	lemma	lemma	PROPN
ejpam-515	273	42	1	1	NUM
ejpam-515	273	43	and	and	CCONJ
ejpam-515	273	44	that	that	DET
ejpam-515	273	45	h(β0	h(β0	NOUN
ejpam-515	273	46	,	,	PUNCT
ejpam-515	273	47	n	n	CCONJ
ejpam-515	273	48	)	)	PUNCT
ejpam-515	273	49	=	=	SYM
ejpam-515	274	1	0	0	X
ejpam-515	274	2	.	.	PUNCT
ejpam-515	275	1	this	this	PRON
ejpam-515	275	2	and	and	CCONJ
ejpam-515	275	3	(	(	PUNCT
ejpam-515	275	4	37	37	NUM
ejpam-515	275	5	)	)	PUNCT
ejpam-515	275	6	suggests	suggest	VERB
ejpam-515	275	7	that	that	SCONJ
ejpam-515	275	8	the	the	DET
ejpam-515	275	9	mle	mle	NOUN
ejpam-515	275	10	β̂	β̂	ADP
ejpam-515	275	11	which	which	PRON
ejpam-515	275	12	also	also	ADV
ejpam-515	275	13	minimizes	minimize	VERB
ejpam-515	275	14	h(β	h(β	PROPN
ejpam-515	275	15	,	,	PUNCT
ejpam-515	275	16	n	n	CCONJ
ejpam-515	275	17	)	)	PUNCT
ejpam-515	275	18	must	must	AUX
ejpam-515	275	19	be	be	AUX
ejpam-515	275	20	inside	inside	ADP
ejpam-515	275	21	the	the	DET
ejpam-515	275	22	subset	subset	NOUN
ejpam-515	275	23	an	an	DET
ejpam-515	275	24	almost	almost	ADV
ejpam-515	275	25	surely	surely	ADV
ejpam-515	275	26	;	;	PUNCT
ejpam-515	275	27	namely	namely	ADV
ejpam-515	275	28	||β̂	||β̂	PUNCT
ejpam-515	276	1	−	−	NOUN
ejpam-515	276	2	β0||	β0||	PROPN
ejpam-515	276	3	≤	≤	ADV
ejpam-515	276	4	τn	τn	ADP
ejpam-515	276	5	p	p	PROPN
ejpam-515	276	6	n−1	n−1	PROPN
ejpam-515	276	7	log	log	NOUN
ejpam-515	276	8	log	log	NOUN
ejpam-515	276	9	n	n	PRON
ejpam-515	276	10	a.s	a.s	PROPN
ejpam-515	276	11	..	..	PUNCT
ejpam-515	276	12	(	(	PUNCT
ejpam-515	276	13	38	38	NUM
ejpam-515	276	14	)	)	PUNCT
ejpam-515	276	15	equation	equation	NOUN
ejpam-515	276	16	(	(	PUNCT
ejpam-515	276	17	38	38	NUM
ejpam-515	276	18	)	)	PUNCT
ejpam-515	276	19	implies	imply	VERB
ejpam-515	276	20	(	(	PUNCT
ejpam-515	276	21	30	30	NUM
ejpam-515	276	22	)	)	PUNCT
ejpam-515	276	23	because	because	SCONJ
ejpam-515	276	24	the	the	DET
ejpam-515	276	25	sequence	sequence	NOUN
ejpam-515	276	26	{	{	PUNCT
ejpam-515	276	27	τn	τn	NOUN
ejpam-515	276	28	}	}	PUNCT
ejpam-515	276	29	can	can	AUX
ejpam-515	276	30	be	be	AUX
ejpam-515	276	31	chosen	choose	VERB
ejpam-515	276	32	to	to	PART
ejpam-515	276	33	diverge	diverge	VERB
ejpam-515	276	34	as	as	ADV
ejpam-515	276	35	slowly	slowly	ADV
ejpam-515	276	36	as	as	ADP
ejpam-515	276	37	possible	possible	ADJ
ejpam-515	276	38	.	.	PUNCT
ejpam-515	277	1	we	we	PRON
ejpam-515	277	2	now	now	ADV
ejpam-515	277	3	proceed	proceed	VERB
ejpam-515	277	4	to	to	PART
ejpam-515	277	5	prove	prove	VERB
ejpam-515	277	6	(	(	PUNCT
ejpam-515	277	7	5	5	NUM
ejpam-515	277	8	)	)	PUNCT
ejpam-515	277	9	for	for	ADP
ejpam-515	277	10	the	the	DET
ejpam-515	277	11	full	full	ADJ
ejpam-515	277	12	model	model	NOUN
ejpam-515	277	13	.	.	PUNCT
ejpam-515	278	1	suppose	suppose	VERB
ejpam-515	278	2	(	(	PUNCT
ejpam-515	278	3	5	5	NUM
ejpam-515	278	4	)	)	PUNCT
ejpam-515	278	5	does	do	AUX
ejpam-515	278	6	not	not	PART
ejpam-515	278	7	hold	hold	VERB
ejpam-515	278	8	for	for	ADP
ejpam-515	278	9	the	the	DET
ejpam-515	278	10	full	full	ADJ
ejpam-515	278	11	model	model	NOUN
ejpam-515	278	12	.	.	PUNCT
ejpam-515	279	1	this	this	PRON
ejpam-515	279	2	implies	imply	VERB
ejpam-515	279	3	that	that	PRON
ejpam-515	279	4	||β̂	||β̂	PUNCT
ejpam-515	280	1	−	−	NOUN
ejpam-515	280	2	β0||	β0||	PROPN
ejpam-515	280	3	=	=	SYM
ejpam-515	280	4	o	o	PROPN
ejpam-515	280	5	(	(	PUNCT
ejpam-515	280	6	p	p	PROPN
ejpam-515	280	7	n−1	n−1	PROPN
ejpam-515	280	8	log	log	NOUN
ejpam-515	280	9	log	log	NOUN
ejpam-515	280	10	n	n	CCONJ
ejpam-515	280	11	)	)	PUNCT
ejpam-515	280	12	a.s	a.s	PROPN
ejpam-515	280	13	.	.	PROPN
ejpam-515	280	14	,	,	PUNCT
ejpam-515	280	15	(	(	PUNCT
ejpam-515	280	16	39	39	NUM
ejpam-515	280	17	)	)	PUNCT
ejpam-515	280	18	knowing	know	VERB
ejpam-515	280	19	that	that	SCONJ
ejpam-515	280	20	(	(	PUNCT
ejpam-515	280	21	30	30	NUM
ejpam-515	280	22	)	)	PUNCT
ejpam-515	280	23	is	be	AUX
ejpam-515	280	24	true	true	ADJ
ejpam-515	280	25	.	.	PUNCT
ejpam-515	281	1	by	by	ADP
ejpam-515	281	2	applying	apply	VERB
ejpam-515	281	3	result	result	NOUN
ejpam-515	281	4	(	(	PUNCT
ejpam-515	281	5	iii	iii	NOUN
ejpam-515	281	6	)	)	PUNCT
ejpam-515	281	7	of	of	ADP
ejpam-515	281	8	lemma	lemma	PROPN
ejpam-515	281	9	1	1	NUM
ejpam-515	281	10	,	,	PUNCT
ejpam-515	281	11	(	(	PUNCT
ejpam-515	281	12	8)	8)	NUM
ejpam-515	281	13	and	and	CCONJ
ejpam-515	281	14	(	(	PUNCT
ejpam-515	281	15	32	32	NUM
ejpam-515	281	16	)	)	PUNCT
ejpam-515	281	17	it	it	PRON
ejpam-515	281	18	follows	follow	VERB
ejpam-515	281	19	that	that	SCONJ
ejpam-515	281	20	r1(β̂	r1(β̂	NUM
ejpam-515	281	21	,	,	PUNCT
ejpam-515	281	22	n)≤	n)≤	NOUN
ejpam-515	281	23	1	1	NUM
ejpam-515	281	24	2	2	NUM
ejpam-515	281	25	e2max1≤k≤n	e2max1≤k≤n	NUM
ejpam-515	281	26	|xt	|xt	X
ejpam-515	281	27	k	k	X
ejpam-515	281	28	(	(	PUNCT
ejpam-515	281	29	β̂−β0)|	β̂−β0)|	X
ejpam-515	281	30	n	n	CCONJ
ejpam-515	281	31	∑	∑	PUNCT
ejpam-515	282	1	k=1	k=1	PROPN
ejpam-515	282	2	µ0k[x	µ0k[x	PROPN
ejpam-515	282	3	t	t	PROPN
ejpam-515	282	4	k(β̂	k(β̂	PROPN
ejpam-515	282	5	−	−	PROPN
ejpam-515	282	6	β0	β0	PROPN
ejpam-515	282	7	)	)	PUNCT
ejpam-515	282	8	]	]	PUNCT
ejpam-515	282	9	2	2	NUM
ejpam-515	282	10	≤	≤	NUM
ejpam-515	282	11	1	1	NUM
ejpam-515	282	12	2	2	NUM
ejpam-515	282	13	e2	e2	NOUN
ejpam-515	282	14	p	p	NOUN
ejpam-515	282	15	b2nξn||β̂−β0||(β̂−β0	b2nξn||β̂−β0||(β̂−β0	PROPN
ejpam-515	282	16	)	)	PUNCT
ejpam-515	282	17	t	t	NOUN
ejpam-515	282	18	in(β0)(β̂−β0)≤	in(β0)(β̂−β0)≤	NOUN
ejpam-515	282	19	1	1	NUM
ejpam-515	282	20	2	2	NUM
ejpam-515	282	21	e2	e2	NOUN
ejpam-515	282	22	p	p	NOUN
ejpam-515	282	23	b2nξn||β̂−β0||λp{in(β0)}||β̂−β0||2	b2nξn||β̂−β0||λp{in(β0)}||β̂−β0||2	NOUN
ejpam-515	282	24	.	.	PUNCT
ejpam-515	283	1	(	(	PUNCT
ejpam-515	283	2	40	40	NUM
ejpam-515	283	3	)	)	PUNCT
ejpam-515	283	4	thus	thus	ADV
ejpam-515	283	5	by	by	ADP
ejpam-515	283	6	(	(	PUNCT
ejpam-515	283	7	39	39	NUM
ejpam-515	283	8	)	)	PUNCT
ejpam-515	283	9	and	and	CCONJ
ejpam-515	283	10	conditions	condition	NOUN
ejpam-515	283	11	(	(	PUNCT
ejpam-515	283	12	c.2	c.2	NOUN
ejpam-515	283	13	)	)	PUNCT
ejpam-515	283	14	and	and	CCONJ
ejpam-515	283	15	(	(	PUNCT
ejpam-515	283	16	c.3	c.3	X
ejpam-515	283	17	)	)	PUNCT
ejpam-515	283	18	we	we	PRON
ejpam-515	283	19	have	have	VERB
ejpam-515	283	20	r1(β̂	r1(β̂	NUM
ejpam-515	283	21	,	,	PUNCT
ejpam-515	283	22	n	n	CCONJ
ejpam-515	283	23	)	)	PUNCT
ejpam-515	283	24	=	=	SYM
ejpam-515	283	25	o(1	o(1	PROPN
ejpam-515	283	26	)	)	PUNCT
ejpam-515	283	27	log	log	NOUN
ejpam-515	283	28	log	log	NOUN
ejpam-515	283	29	n	n	PRON
ejpam-515	283	30	a.s	a.s	PROPN
ejpam-515	283	31	..	..	PROPN
ejpam-515	283	32	corresponding	correspond	VERB
ejpam-515	283	33	to	to	ADP
ejpam-515	283	34	(	(	PUNCT
ejpam-515	283	35	36	36	NUM
ejpam-515	283	36	)	)	PUNCT
ejpam-515	283	37	it	it	PRON
ejpam-515	283	38	can	can	AUX
ejpam-515	283	39	be	be	AUX
ejpam-515	283	40	seen	see	VERB
ejpam-515	283	41	that	that	SCONJ
ejpam-515	283	42	r2(β̂	r2(β̂	NOUN
ejpam-515	283	43	,	,	PUNCT
ejpam-515	283	44	n	n	CCONJ
ejpam-515	283	45	)	)	PUNCT
ejpam-515	283	46	=	=	SYM
ejpam-515	283	47	o(1	o(1	PROPN
ejpam-515	283	48	)	)	PUNCT
ejpam-515	283	49	log	log	NOUN
ejpam-515	283	50	log	log	NOUN
ejpam-515	283	51	n	n	PRON
ejpam-515	283	52	a.s	a.s	PROPN
ejpam-515	283	53	.	.	PROPN
ejpam-515	283	54	under	under	ADP
ejpam-515	283	55	the	the	DET
ejpam-515	283	56	assumption	assumption	NOUN
ejpam-515	283	57	of	of	ADP
ejpam-515	283	58	(	(	PUNCT
ejpam-515	283	59	39	39	NUM
ejpam-515	283	60	)	)	PUNCT
ejpam-515	283	61	.	.	PUNCT
ejpam-515	284	1	hence	hence	ADV
ejpam-515	284	2	we	we	PRON
ejpam-515	284	3	have	have	VERB
ejpam-515	284	4	that	that	PRON
ejpam-515	284	5	under	under	ADP
ejpam-515	284	6	assumption	assumption	NOUN
ejpam-515	284	7	(	(	PUNCT
ejpam-515	284	8	39	39	NUM
ejpam-515	284	9	)	)	PUNCT
ejpam-515	284	10	h(β̂	h(β̂	NOUN
ejpam-515	284	11	,	,	PUNCT
ejpam-515	284	12	n	n	CCONJ
ejpam-515	284	13	)	)	PUNCT
ejpam-515	284	14	=	=	SYM
ejpam-515	284	15	r1(β̂	r1(β̂	NUM
ejpam-515	284	16	,	,	PUNCT
ejpam-515	284	17	n	n	CCONJ
ejpam-515	284	18	)	)	PUNCT
ejpam-515	285	1	+	+	ADJ
ejpam-515	285	2	r2(β̂	r2(β̂	NOUN
ejpam-515	285	3	,	,	PUNCT
ejpam-515	285	4	n	n	CCONJ
ejpam-515	285	5	)	)	PUNCT
ejpam-515	285	6	=	=	SYM
ejpam-515	285	7	o(1	o(1	PROPN
ejpam-515	285	8	)	)	PUNCT
ejpam-515	285	9	log	log	NOUN
ejpam-515	285	10	log	log	NOUN
ejpam-515	285	11	n	n	PRON
ejpam-515	285	12	a.s	a.s	PROPN
ejpam-515	285	13	..	..	PUNCT
ejpam-515	285	14	(	(	PUNCT
ejpam-515	285	15	41	41	NUM
ejpam-515	285	16	)	)	PUNCT
ejpam-515	285	17	on	on	ADP
ejpam-515	285	18	the	the	DET
ejpam-515	285	19	other	other	ADJ
ejpam-515	285	20	hand	hand	NOUN
ejpam-515	285	21	,	,	PUNCT
ejpam-515	285	22	from	from	ADP
ejpam-515	285	23	(	(	PUNCT
ejpam-515	285	24	11	11	NUM
ejpam-515	285	25	)	)	PUNCT
ejpam-515	285	26	of	of	ADP
ejpam-515	285	27	lemma	lemma	PROPN
ejpam-515	285	28	4	4	NUM
ejpam-515	285	29	we	we	PRON
ejpam-515	285	30	know	know	VERB
ejpam-515	285	31	there	there	PRON
ejpam-515	285	32	exists	exist	VERB
ejpam-515	285	33	a	a	DET
ejpam-515	285	34	sequence	sequence	NOUN
ejpam-515	285	35	of	of	ADP
ejpam-515	285	36	positive	positive	ADJ
ejpam-515	285	37	integers	integer	NOUN
ejpam-515	285	38	{	{	PUNCT
ejpam-515	285	39	ni	ni	PROPN
ejpam-515	285	40	↑	↑	PROPN
ejpam-515	285	41	∞	∞	PROPN
ejpam-515	285	42	}	}	PUNCT
ejpam-515	285	43	such	such	ADJ
ejpam-515	285	44	that	that	SCONJ
ejpam-515	285	45	lim	lim	PROPN
ejpam-515	285	46	i→∞	i→∞	VERB
ejpam-515	285	47	∑ni	∑ni	NOUN
ejpam-515	286	1	k=1	k=1	PROPN
ejpam-515	287	1	(	(	PUNCT
ejpam-515	287	2	yk	yk	PROPN
ejpam-515	287	3	−µ0k)xk1	−µ0k)xk1	NOUN
ejpam-515	287	4	p	p	PROPN
ejpam-515	287	5	2ini	2ini	PROPN
ejpam-515	287	6	(	(	PUNCT
ejpam-515	287	7	β0)(1,1	β0)(1,1	NOUN
ejpam-515	287	8	)	)	PUNCT
ejpam-515	287	9	log	log	PROPN
ejpam-515	287	10	log	log	PROPN
ejpam-515	287	11	ini	ini	PROPN
ejpam-515	287	12	(	(	PUNCT
ejpam-515	287	13	β0)(1,1	β0)(1,1	NOUN
ejpam-515	287	14	)	)	PUNCT
ejpam-515	287	15	=	=	SYM
ejpam-515	287	16	1	1	NUM
ejpam-515	287	17	a.s	a.s	PROPN
ejpam-515	287	18	..	..	PROPN
ejpam-515	287	19	g.	g.	PROPN
ejpam-515	287	20	qian	qian	PROPN
ejpam-515	287	21	/	/	SYM
ejpam-515	287	22	eur	eur	PROPN
ejpam-515	287	23	.	.	PUNCT
ejpam-515	288	1	j.	j.	PROPN
ejpam-515	288	2	pure	pure	PROPN
ejpam-515	288	3	appl	appl	PROPN
ejpam-515	288	4	.	.	PROPN
ejpam-515	288	5	math	math	PROPN
ejpam-515	288	6	,	,	PUNCT
ejpam-515	288	7	3	3	NUM
ejpam-515	288	8	(	(	PUNCT
ejpam-515	288	9	2010	2010	NUM
ejpam-515	288	10	)	)	PUNCT
ejpam-515	288	11	,	,	PUNCT
ejpam-515	288	12	417	417	NUM
ejpam-515	288	13	-	-	SYM
ejpam-515	288	14	434	434	NUM
ejpam-515	288	15	429	429	NUM
ejpam-515	288	16	thus	thus	ADV
ejpam-515	288	17	when	when	SCONJ
ejpam-515	288	18	ni	ni	PROPN
ejpam-515	288	19	is	be	AUX
ejpam-515	288	20	sufficiently	sufficiently	ADV
ejpam-515	288	21	large	large	ADJ
ejpam-515	288	22	,	,	PUNCT
ejpam-515	288	23	∑ni	∑ni	PUNCT
ejpam-515	288	24	k=1	k=1	PUNCT
ejpam-515	288	25	(	(	PUNCT
ejpam-515	288	26	yk	yk	PROPN
ejpam-515	288	27	−µ0k)xk1	−µ0k)xk1	NOUN
ejpam-515	288	28	p	p	PROPN
ejpam-515	288	29	2ini	2ini	PROPN
ejpam-515	288	30	(	(	PUNCT
ejpam-515	288	31	β0)(1,1	β0)(1,1	NOUN
ejpam-515	288	32	)	)	PUNCT
ejpam-515	288	33	log	log	PROPN
ejpam-515	288	34	log	log	PROPN
ejpam-515	288	35	ini	ini	PROPN
ejpam-515	288	36	(	(	PUNCT
ejpam-515	288	37	β0)(1,1	β0)(1,1	NOUN
ejpam-515	288	38	)	)	PUNCT
ejpam-515	288	39	≥	≥	NOUN
ejpam-515	288	40	1	1	NUM
ejpam-515	288	41	2	2	NUM
ejpam-515	288	42	a.s	a.s	X
ejpam-515	288	43	..	..	PUNCT
ejpam-515	288	44	(	(	PUNCT
ejpam-515	288	45	42	42	NUM
ejpam-515	288	46	)	)	PUNCT
ejpam-515	288	47	now	now	ADV
ejpam-515	288	48	define	define	VERB
ejpam-515	288	49	a	a	DET
ejpam-515	288	50	p×	p×	NOUN
ejpam-515	288	51	1	1	NUM
ejpam-515	288	52	vector	vector	NOUN
ejpam-515	288	53	β̃ni	β̃ni	PUNCT
ejpam-515	288	54	with	with	ADP
ejpam-515	288	55	β̃ni	β̃ni	PRON
ejpam-515	288	56	(	(	PUNCT
ejpam-515	288	57	j	j	NOUN
ejpam-515	288	58	)	)	PUNCT
ejpam-515	288	59	=	=	PUNCT
ejpam-515	289	1	β0	β0	PROPN
ejpam-515	289	2	j	j	PROPN
ejpam-515	289	3	for	for	ADP
ejpam-515	289	4	j	j	PROPN
ejpam-515	289	5	=	=	SYM
ejpam-515	289	6	2	2	NUM
ejpam-515	289	7	,	,	PUNCT
ejpam-515	289	8	·	·	PUNCT
ejpam-515	289	9	·	·	PUNCT
ejpam-515	289	10	·	·	PUNCT
ejpam-515	289	11	,	,	PUNCT
ejpam-515	289	12	p	p	NOUN
ejpam-515	289	13	and	and	CCONJ
ejpam-515	289	14	β̃ni	β̃ni	PROPN
ejpam-515	289	15	(	(	PUNCT
ejpam-515	289	16	1	1	X
ejpam-515	289	17	)	)	PUNCT
ejpam-515	289	18	=	=	NOUN
ejpam-515	289	19	b1	b1	NOUN
ejpam-515	289	20	4b0	4b0	NUM
ejpam-515	289	21	b2	b2	PROPN
ejpam-515	289	22	è	è	PROPN
ejpam-515	289	23	2	2	NUM
ejpam-515	289	24	log	log	NOUN
ejpam-515	289	25	log	log	NOUN
ejpam-515	289	26	ini	ini	PROPN
ejpam-515	289	27	(	(	PUNCT
ejpam-515	289	28	β0)(1,1	β0)(1,1	NOUN
ejpam-515	289	29	)	)	PUNCT
ejpam-515	289	30	ini	ini	PROPN
ejpam-515	289	31	(	(	PUNCT
ejpam-515	289	32	β0)(1,1	β0)(1,1	NOUN
ejpam-515	289	33	)	)	PUNCT
ejpam-515	289	34	+	+	CCONJ
ejpam-515	289	35	β01	β01	NOUN
ejpam-515	289	36	.	.	PUNCT
ejpam-515	290	1	then	then	ADV
ejpam-515	290	2	from	from	ADP
ejpam-515	290	3	(	(	PUNCT
ejpam-515	290	4	42	42	NUM
ejpam-515	290	5	)	)	PUNCT
ejpam-515	290	6	it	it	PRON
ejpam-515	290	7	follows	follow	VERB
ejpam-515	290	8	that	that	SCONJ
ejpam-515	290	9	,	,	PUNCT
ejpam-515	290	10	when	when	SCONJ
ejpam-515	290	11	ni	ni	PROPN
ejpam-515	290	12	is	be	AUX
ejpam-515	290	13	sufficiently	sufficiently	ADV
ejpam-515	290	14	large	large	ADJ
ejpam-515	290	15	r2(β̃ni	r2(β̃ni	PROPN
ejpam-515	290	16	,	,	PUNCT
ejpam-515	290	17	ni	ni	PROPN
ejpam-515	290	18	)	)	PUNCT
ejpam-515	290	19	=	=	SYM
ejpam-515	290	20	ni	ni	PROPN
ejpam-515	290	21	∑	∑	PROPN
ejpam-515	290	22	k=1	k=1	PROPN
ejpam-515	290	23	(	(	PUNCT
ejpam-515	290	24	µ0k−	µ0k−	ADV
ejpam-515	290	25	yk)x	yk)x	PROPN
ejpam-515	290	26	t	t	PROPN
ejpam-515	291	1	k	k	PROPN
ejpam-515	291	2	(	(	PUNCT
ejpam-515	291	3	β̃ni	β̃ni	ADP
ejpam-515	291	4	−β0	−β0	PROPN
ejpam-515	291	5	)	)	PUNCT
ejpam-515	291	6	=	=	SYM
ejpam-515	291	7	ni	ni	PROPN
ejpam-515	291	8	∑	∑	PROPN
ejpam-515	291	9	k=1	k=1	PROPN
ejpam-515	291	10	(	(	PUNCT
ejpam-515	291	11	µ0k−	µ0k−	ADV
ejpam-515	291	12	yk)xk1(β̃ni	yk)xk1(β̃ni	X
ejpam-515	292	1	(	(	PUNCT
ejpam-515	292	2	1)−	1)−	NUM
ejpam-515	292	3	β01	β01	NOUN
ejpam-515	292	4	)	)	PUNCT
ejpam-515	292	5	≤	≤	NUM
ejpam-515	292	6	−1	−1	NOUN
ejpam-515	292	7	2	2	NUM
ejpam-515	292	8	p	p	NOUN
ejpam-515	292	9	2ini	2ini	PROPN
ejpam-515	292	10	(	(	PUNCT
ejpam-515	292	11	β0)(1,1	β0)(1,1	NOUN
ejpam-515	292	12	)	)	PUNCT
ejpam-515	292	13	log	log	PROPN
ejpam-515	292	14	log	log	PROPN
ejpam-515	292	15	ini	ini	PROPN
ejpam-515	292	16	(	(	PUNCT
ejpam-515	292	17	β0)(1,1	β0)(1,1	NOUN
ejpam-515	292	18	)	)	PUNCT
ejpam-515	292	19	·	·	PUNCT
ejpam-515	292	20	b1	b1	NOUN
ejpam-515	292	21	4b0	4b0	NUM
ejpam-515	292	22	b2	b2	PROPN
ejpam-515	292	23	è	è	PROPN
ejpam-515	292	24	2	2	NUM
ejpam-515	292	25	log	log	NOUN
ejpam-515	292	26	log	log	NOUN
ejpam-515	292	27	ini	ini	PROPN
ejpam-515	292	28	(	(	PUNCT
ejpam-515	292	29	β0)(1,1	β0)(1,1	NOUN
ejpam-515	292	30	)	)	PUNCT
ejpam-515	292	31	ini	ini	PROPN
ejpam-515	292	32	(	(	PUNCT
ejpam-515	292	33	β0)(1,1	β0)(1,1	NOUN
ejpam-515	292	34	)	)	PUNCT
ejpam-515	292	35	=	=	SYM
ejpam-515	293	1	−	−	PROPN
ejpam-515	293	2	b1	b1	NOUN
ejpam-515	293	3	4b0	4b0	NUM
ejpam-515	293	4	b2	b2	NOUN
ejpam-515	293	5	log	log	NOUN
ejpam-515	293	6	log	log	PROPN
ejpam-515	293	7	ini	ini	PROPN
ejpam-515	293	8	(	(	PUNCT
ejpam-515	293	9	β0)(1,1	β0)(1,1	NOUN
ejpam-515	293	10	)	)	PUNCT
ejpam-515	294	1	a.s	a.s	PROPN
ejpam-515	294	2	..	..	PUNCT
ejpam-515	294	3	(	(	PUNCT
ejpam-515	294	4	43	43	X
ejpam-515	294	5	)	)	PUNCT
ejpam-515	294	6	note	note	VERB
ejpam-515	294	7	that	that	SCONJ
ejpam-515	294	8	by	by	ADP
ejpam-515	294	9	conditions	condition	NOUN
ejpam-515	294	10	(	(	PUNCT
ejpam-515	294	11	c.1	c.1	NOUN
ejpam-515	294	12	)	)	PUNCT
ejpam-515	294	13	and	and	CCONJ
ejpam-515	294	14	(	(	PUNCT
ejpam-515	294	15	c.2	c.2	NOUN
ejpam-515	294	16	)	)	PUNCT
ejpam-515	294	17	b2n≥	b2n≥	ADP
ejpam-515	294	18	λp{in(β0	λp{in(β0	NUM
ejpam-515	294	19	)	)	PUNCT
ejpam-515	294	20	}	}	PUNCT
ejpam-515	294	21	≥	≥	PRON
ejpam-515	294	22	in(β0)(1,1)≥	in(β0)(1,1)≥	VERB
ejpam-515	294	23	λ1{in(β0	λ1{in(β0	NOUN
ejpam-515	294	24	)	)	PUNCT
ejpam-515	294	25	}	}	PUNCT
ejpam-515	294	26	≥	≥	NOUN
ejpam-515	294	27	b1	b1	PROPN
ejpam-515	294	28	b0	b0	PROPN
ejpam-515	294	29	n	n	PROPN
ejpam-515	294	30	(	(	PUNCT
ejpam-515	294	31	44	44	NUM
ejpam-515	294	32	)	)	PUNCT
ejpam-515	294	33	and	and	CCONJ
ejpam-515	294	34	accordingly	accordingly	ADV
ejpam-515	294	35	2	2	NUM
ejpam-515	294	36	log	log	NOUN
ejpam-515	294	37	log	log	NOUN
ejpam-515	294	38	n≥	n≥	NOUN
ejpam-515	294	39	log	log	NOUN
ejpam-515	294	40	log	log	NOUN
ejpam-515	294	41	in(β0)(1,1)≥	in(β0)(1,1)≥	NOUN
ejpam-515	294	42	1	1	NUM
ejpam-515	294	43	2	2	NUM
ejpam-515	294	44	log	log	NOUN
ejpam-515	294	45	log	log	NOUN
ejpam-515	294	46	n	n	CCONJ
ejpam-515	294	47	when	when	SCONJ
ejpam-515	294	48	n	n	X
ejpam-515	294	49	is	be	AUX
ejpam-515	294	50	sufficiently	sufficiently	ADV
ejpam-515	294	51	large	large	ADJ
ejpam-515	294	52	.	.	PUNCT
ejpam-515	295	1	(	(	PUNCT
ejpam-515	295	2	45	45	NUM
ejpam-515	295	3	)	)	PUNCT
ejpam-515	295	4	it	it	PRON
ejpam-515	295	5	follows	follow	VERB
ejpam-515	295	6	that	that	SCONJ
ejpam-515	295	7	p	p	PROPN
ejpam-515	295	8	b1	b1	NOUN
ejpam-515	295	9	2	2	NUM
ejpam-515	295	10	p	p	NOUN
ejpam-515	295	11	b0	b0	NOUN
ejpam-515	295	12	b2	b2	NOUN
ejpam-515	296	1	æ	æ	PROPN
ejpam-515	296	2	n−1	n−1	PROPN
ejpam-515	296	3	i	i	PRON
ejpam-515	296	4	log	log	VERB
ejpam-515	296	5	log	log	PROPN
ejpam-515	296	6	ni	ni	PROPN
ejpam-515	296	7	≥	≥	PROPN
ejpam-515	296	8	β̃ni	β̃ni	PRON
ejpam-515	296	9	(	(	PUNCT
ejpam-515	296	10	1)−	1)−	PROPN
ejpam-515	296	11	β01	β01	PROPN
ejpam-515	296	12	≥	≥	NOUN
ejpam-515	296	13	b1	b1	NOUN
ejpam-515	296	14	4b0	4b0	NUM
ejpam-515	296	15	b2	b2	PROPN
ejpam-515	296	16	p	p	NOUN
ejpam-515	296	17	b2	b2	NOUN
ejpam-515	297	1	æ	æ	PROPN
ejpam-515	297	2	n−1	n−1	PROPN
ejpam-515	297	3	i	i	PRON
ejpam-515	297	4	log	log	VERB
ejpam-515	297	5	log	log	PROPN
ejpam-515	297	6	ni	ni	PROPN
ejpam-515	297	7	(	(	PUNCT
ejpam-515	297	8	46	46	NUM
ejpam-515	297	9	)	)	PUNCT
ejpam-515	297	10	when	when	SCONJ
ejpam-515	297	11	ni	ni	PROPN
ejpam-515	297	12	is	be	AUX
ejpam-515	297	13	sufficiently	sufficiently	ADV
ejpam-515	297	14	large	large	ADJ
ejpam-515	297	15	.	.	PUNCT
ejpam-515	298	1	using	use	VERB
ejpam-515	298	2	(	(	PUNCT
ejpam-515	298	3	46	46	NUM
ejpam-515	298	4	)	)	PUNCT
ejpam-515	298	5	,	,	PUNCT
ejpam-515	298	6	(	(	PUNCT
ejpam-515	298	7	32	32	NUM
ejpam-515	298	8	)	)	PUNCT
ejpam-515	298	9	and	and	CCONJ
ejpam-515	298	10	the	the	DET
ejpam-515	298	11	fact	fact	NOUN
ejpam-515	298	12	that	that	SCONJ
ejpam-515	298	13	ξn	ξn	PROPN
ejpam-515	298	14	p	p	NOUN
ejpam-515	298	15	log	log	NOUN
ejpam-515	298	16	log	log	NOUN
ejpam-515	298	17	n	n	NOUN
ejpam-515	298	18	→	→	SYM
ejpam-515	298	19	0	0	NUM
ejpam-515	298	20	one	one	NUM
ejpam-515	298	21	can	can	AUX
ejpam-515	298	22	show	show	VERB
ejpam-515	298	23	that	that	SCONJ
ejpam-515	298	24	max	max	PROPN
ejpam-515	299	1	1≤k≤ni	1≤k≤ni	NUM
ejpam-515	299	2	|xt	|xt	PROPN
ejpam-515	299	3	k	k	X
ejpam-515	299	4	(	(	PUNCT
ejpam-515	299	5	β̃ni−β0)|	β̃ni−β0)|	PROPN
ejpam-515	299	6	≤	≤	PROPN
ejpam-515	299	7	max	max	PROPN
ejpam-515	299	8	1≤k≤ni	1≤k≤ni	NUM
ejpam-515	299	9	||xk||(β̃ni	||xk||(β̃ni	PROPN
ejpam-515	299	10	(	(	PUNCT
ejpam-515	299	11	1)−β01	1)−β01	NUM
ejpam-515	299	12	)	)	PUNCT
ejpam-515	299	13	≤	≤	NUM
ejpam-515	299	14	1	1	NUM
ejpam-515	299	15	2	2	NUM
ejpam-515	299	16	è	è	PROPN
ejpam-515	299	17	b1	b1	PROPN
ejpam-515	299	18	b0	b0	PROPN
ejpam-515	299	19	b2	b2	PROPN
ejpam-515	299	20	ξni	ξni	NOUN
ejpam-515	299	21	p	p	PROPN
ejpam-515	299	22	log	log	NOUN
ejpam-515	299	23	log	log	NOUN
ejpam-515	299	24	ni	ni	PROPN
ejpam-515	299	25	≤	≤	PROPN
ejpam-515	299	26	1	1	NUM
ejpam-515	299	27	2	2	NUM
ejpam-515	299	28	log2	log2	NOUN
ejpam-515	299	29	(	(	PUNCT
ejpam-515	299	30	47	47	NUM
ejpam-515	299	31	)	)	PUNCT
ejpam-515	299	32	when	when	SCONJ
ejpam-515	299	33	ni	ni	PROPN
ejpam-515	299	34	is	be	AUX
ejpam-515	299	35	sufficiently	sufficiently	ADV
ejpam-515	299	36	large	large	ADJ
ejpam-515	299	37	.	.	PUNCT
ejpam-515	299	38	similar	similar	ADJ
ejpam-515	299	39	to	to	ADP
ejpam-515	299	40	proving	prove	VERB
ejpam-515	299	41	(	(	PUNCT
ejpam-515	299	42	40	40	NUM
ejpam-515	299	43	)	)	PUNCT
ejpam-515	299	44	,	,	PUNCT
ejpam-515	299	45	by	by	ADP
ejpam-515	299	46	(	(	PUNCT
ejpam-515	299	47	44	44	NUM
ejpam-515	299	48	)	)	PUNCT
ejpam-515	299	49	and	and	CCONJ
ejpam-515	299	50	(	(	PUNCT
ejpam-515	299	51	47	47	NUM
ejpam-515	299	52	)	)	PUNCT
ejpam-515	299	53	one	one	PRON
ejpam-515	299	54	can	can	AUX
ejpam-515	299	55	see	see	VERB
ejpam-515	299	56	that	that	DET
ejpam-515	299	57	r1(β̃ni	r1(β̃ni	ADJ
ejpam-515	299	58	,	,	PUNCT
ejpam-515	299	59	ni)≤	ni)≤	X
ejpam-515	299	60	λp{ini	λp{ini	X
ejpam-515	299	61	(	(	PUNCT
ejpam-515	299	62	β0)}||β̃ni	β0)}||β̃ni	ADJ
ejpam-515	299	63	−	−	PROPN
ejpam-515	299	64	β0||2	β0||2	ADJ
ejpam-515	299	65	≤	≤	NOUN
ejpam-515	299	66	b2ni	b2ni	PUNCT
ejpam-515	299	67	b2	b2	NOUN
ejpam-515	299	68	1	1	NUM
ejpam-515	299	69	16b2	16b2	NUM
ejpam-515	299	70	0	0	NUM
ejpam-515	299	71	b2	b2	NOUN
ejpam-515	299	72	2	2	NUM
ejpam-515	299	73	2	2	NUM
ejpam-515	299	74	log	log	NOUN
ejpam-515	299	75	log	log	NOUN
ejpam-515	299	76	ini	ini	PROPN
ejpam-515	299	77	(	(	PUNCT
ejpam-515	299	78	β0)(1,1	β0)(1,1	NOUN
ejpam-515	299	79	)	)	PUNCT
ejpam-515	299	80	ini	ini	PROPN
ejpam-515	299	81	(	(	PUNCT
ejpam-515	299	82	β0)(1,1	β0)(1,1	NOUN
ejpam-515	299	83	)	)	PUNCT
ejpam-515	299	84	g.	g.	PROPN
ejpam-515	299	85	qian	qian	PROPN
ejpam-515	299	86	/	/	SYM
ejpam-515	299	87	eur	eur	PROPN
ejpam-515	299	88	.	.	PUNCT
ejpam-515	300	1	j.	j.	PROPN
ejpam-515	300	2	pure	pure	PROPN
ejpam-515	300	3	appl	appl	PROPN
ejpam-515	300	4	.	.	PROPN
ejpam-515	300	5	math	math	PROPN
ejpam-515	300	6	,	,	PUNCT
ejpam-515	300	7	3	3	NUM
ejpam-515	300	8	(	(	PUNCT
ejpam-515	300	9	2010	2010	NUM
ejpam-515	300	10	)	)	PUNCT
ejpam-515	300	11	,	,	PUNCT
ejpam-515	300	12	417	417	NUM
ejpam-515	300	13	-	-	SYM
ejpam-515	300	14	434	434	NUM
ejpam-515	300	15	430	430	NUM
ejpam-515	300	16	≤	≤	NUM
ejpam-515	300	17	b1	b1	NOUN
ejpam-515	300	18	8b0	8b0	NUM
ejpam-515	300	19	b2	b2	NOUN
ejpam-515	300	20	log	log	NOUN
ejpam-515	300	21	log	log	PROPN
ejpam-515	300	22	ini	ini	PROPN
ejpam-515	300	23	(	(	PUNCT
ejpam-515	300	24	β0)(1,1	β0)(1,1	NOUN
ejpam-515	300	25	)	)	PUNCT
ejpam-515	300	26	when	when	SCONJ
ejpam-515	300	27	ni	ni	PROPN
ejpam-515	300	28	is	be	AUX
ejpam-515	300	29	sufficiently	sufficiently	ADV
ejpam-515	300	30	large	large	ADJ
ejpam-515	300	31	.	.	PUNCT
ejpam-515	301	1	(	(	PUNCT
ejpam-515	301	2	48	48	NUM
ejpam-515	301	3	)	)	PUNCT
ejpam-515	301	4	thus	thus	ADV
ejpam-515	301	5	,	,	PUNCT
ejpam-515	301	6	by	by	ADP
ejpam-515	301	7	(	(	PUNCT
ejpam-515	301	8	43	43	NUM
ejpam-515	301	9	)	)	PUNCT
ejpam-515	301	10	,	,	PUNCT
ejpam-515	301	11	(	(	PUNCT
ejpam-515	301	12	48	48	NUM
ejpam-515	301	13	)	)	PUNCT
ejpam-515	301	14	and	and	CCONJ
ejpam-515	301	15	(	(	PUNCT
ejpam-515	301	16	45	45	NUM
ejpam-515	301	17	)	)	PUNCT
ejpam-515	301	18	it	it	PRON
ejpam-515	301	19	follows	follow	VERB
ejpam-515	301	20	that	that	SCONJ
ejpam-515	301	21	when	when	SCONJ
ejpam-515	301	22	ni	ni	PROPN
ejpam-515	301	23	is	be	AUX
ejpam-515	301	24	sufficiently	sufficiently	ADV
ejpam-515	301	25	large	large	ADJ
ejpam-515	301	26	,	,	PUNCT
ejpam-515	301	27	h(β̃ni	h(β̃ni	NOUN
ejpam-515	301	28	,	,	PUNCT
ejpam-515	301	29	ni	ni	PROPN
ejpam-515	301	30	)	)	PUNCT
ejpam-515	301	31	=	=	PUNCT
ejpam-515	302	1	r1(β̃ni	r1(β̃ni	ADJ
ejpam-515	302	2	,	,	PUNCT
ejpam-515	302	3	ni	ni	PROPN
ejpam-515	302	4	)	)	PUNCT
ejpam-515	302	5	+	+	CCONJ
ejpam-515	302	6	r2(β̃ni	r2(β̃ni	INTJ
ejpam-515	302	7	,	,	PUNCT
ejpam-515	302	8	ni	ni	PROPN
ejpam-515	302	9	)	)	PUNCT
ejpam-515	302	10	≤	≤	NOUN
ejpam-515	302	11	−	−	ADP
ejpam-515	302	12	b1	b1	NOUN
ejpam-515	302	13	8b0b2	8b0b2	NUM
ejpam-515	302	14	log	log	PROPN
ejpam-515	302	15	log	log	PROPN
ejpam-515	302	16	ini	ini	PROPN
ejpam-515	302	17	(	(	PUNCT
ejpam-515	302	18	β0)(1,1)≤	β0)(1,1)≤	PUNCT
ejpam-515	302	19	−	−	PROPN
ejpam-515	302	20	b1	b1	PROPN
ejpam-515	302	21	16b0b2	16b0b2	PROPN
ejpam-515	302	22	log	log	NOUN
ejpam-515	302	23	log	log	NOUN
ejpam-515	302	24	ni	ni	PROPN
ejpam-515	302	25	a.s	a.s	PROPN
ejpam-515	302	26	..	..	PUNCT
ejpam-515	302	27	(	(	PUNCT
ejpam-515	302	28	49	49	NUM
ejpam-515	302	29	)	)	PUNCT
ejpam-515	302	30	since	since	SCONJ
ejpam-515	302	31	β̂	β̂	NUM
ejpam-515	302	32	≡	≡	PROPN
ejpam-515	302	33	β̂n	β̂n	PROPN
ejpam-515	302	34	is	be	AUX
ejpam-515	302	35	the	the	DET
ejpam-515	302	36	mle	mle	NOUN
ejpam-515	302	37	that	that	PRON
ejpam-515	302	38	minimizes	minimize	VERB
ejpam-515	302	39	h(β	h(β	PROPN
ejpam-515	302	40	,	,	PUNCT
ejpam-515	302	41	n	n	CCONJ
ejpam-515	302	42	)	)	PUNCT
ejpam-515	302	43	,	,	PUNCT
ejpam-515	302	44	inferring	infer	VERB
ejpam-515	302	45	from	from	ADP
ejpam-515	302	46	(	(	PUNCT
ejpam-515	302	47	49	49	NUM
ejpam-515	302	48	)	)	PUNCT
ejpam-515	302	49	we	we	PRON
ejpam-515	302	50	have	have	AUX
ejpam-515	302	51	h(β̂ni	h(β̂ni	PROPN
ejpam-515	302	52	,	,	PUNCT
ejpam-515	302	53	ni)≤	ni)≤	X
ejpam-515	302	54	h(β̃ni	h(β̃ni	NOUN
ejpam-515	302	55	,	,	PUNCT
ejpam-515	302	56	ni)≤	ni)≤	X
ejpam-515	302	57	−	−	NOUN
ejpam-515	302	58	b1	b1	PROPN
ejpam-515	302	59	16b0b2	16b0b2	PROPN
ejpam-515	302	60	log	log	NOUN
ejpam-515	302	61	log	log	NOUN
ejpam-515	302	62	ni	ni	PROPN
ejpam-515	302	63	a.s	a.s	PROPN
ejpam-515	302	64	.	.	PROPN
ejpam-515	303	1	(	(	PUNCT
ejpam-515	303	2	50	50	NUM
ejpam-515	303	3	)	)	PUNCT
ejpam-515	303	4	when	when	SCONJ
ejpam-515	303	5	ni	ni	PROPN
ejpam-515	303	6	is	be	AUX
ejpam-515	303	7	sufficiently	sufficiently	ADV
ejpam-515	303	8	large	large	ADJ
ejpam-515	303	9	.	.	PUNCT
ejpam-515	304	1	clearly	clearly	ADV
ejpam-515	304	2	,	,	PUNCT
ejpam-515	304	3	(	(	PUNCT
ejpam-515	304	4	50	50	NUM
ejpam-515	304	5	)	)	PUNCT
ejpam-515	304	6	is	be	AUX
ejpam-515	304	7	contradictory	contradictory	ADJ
ejpam-515	304	8	to	to	ADP
ejpam-515	304	9	(	(	PUNCT
ejpam-515	304	10	41	41	NUM
ejpam-515	304	11	)	)	PUNCT
ejpam-515	304	12	which	which	PRON
ejpam-515	304	13	suggests	suggest	VERB
ejpam-515	304	14	that	that	SCONJ
ejpam-515	304	15	(	(	PUNCT
ejpam-515	304	16	39	39	NUM
ejpam-515	304	17	)	)	PUNCT
ejpam-515	304	18	is	be	AUX
ejpam-515	304	19	wrong	wrong	ADJ
ejpam-515	304	20	.	.	PUNCT
ejpam-515	305	1	therefore	therefore	ADV
ejpam-515	305	2	(	(	PUNCT
ejpam-515	305	3	5	5	X
ejpam-515	305	4	)	)	PUNCT
ejpam-515	305	5	is	be	AUX
ejpam-515	305	6	true	true	ADJ
ejpam-515	305	7	for	for	ADP
ejpam-515	305	8	the	the	DET
ejpam-515	305	9	full	full	ADJ
ejpam-515	305	10	model	model	NOUN
ejpam-515	305	11	and	and	CCONJ
ejpam-515	305	12	consequently	consequently	ADV
ejpam-515	305	13	so	so	ADV
ejpam-515	305	14	for	for	SCONJ
ejpam-515	305	15	the	the	DET
ejpam-515	305	16	other	other	ADJ
ejpam-515	305	17	correct	correct	ADJ
ejpam-515	305	18	models	model	NOUN
ejpam-515	305	19	inac	inac	ADJ
ejpam-515	305	20	.	.	PUNCT
ejpam-515	306	1	proof	proof	NOUN
ejpam-515	306	2	.	.	PUNCT
ejpam-515	307	1	(	(	PUNCT
ejpam-515	307	2	theorem	theorem	NOUN
ejpam-515	307	3	2	2	NUM
ejpam-515	307	4	)	)	PUNCT
ejpam-515	307	5	as	as	ADP
ejpam-515	307	6	in	in	ADP
ejpam-515	307	7	proving	prove	VERB
ejpam-515	307	8	theorem	theorem	ADJ
ejpam-515	307	9	1	1	NUM
ejpam-515	307	10	,	,	PUNCT
ejpam-515	307	11	we	we	PRON
ejpam-515	307	12	only	only	ADV
ejpam-515	307	13	need	need	VERB
ejpam-515	307	14	to	to	PART
ejpam-515	307	15	prove	prove	VERB
ejpam-515	307	16	(	(	PUNCT
ejpam-515	307	17	6	6	NUM
ejpam-515	307	18	)	)	PUNCT
ejpam-515	307	19	for	for	ADP
ejpam-515	307	20	the	the	DET
ejpam-515	307	21	full	full	ADJ
ejpam-515	307	22	model	model	NOUN
ejpam-515	307	23	which	which	PRON
ejpam-515	307	24	is	be	AUX
ejpam-515	307	25	equivalent	equivalent	ADJ
ejpam-515	307	26	to	to	ADP
ejpam-515	307	27	0≥	0≥	PROPN
ejpam-515	307	28	h(β̂	h(β̂	PROPN
ejpam-515	307	29	,	,	PUNCT
ejpam-515	307	30	n	n	CCONJ
ejpam-515	307	31	)	)	PUNCT
ejpam-515	307	32	=	=	SYM
ejpam-515	307	33	r1(β̂	r1(β̂	NUM
ejpam-515	307	34	,	,	PUNCT
ejpam-515	307	35	n	n	CCONJ
ejpam-515	307	36	)	)	PUNCT
ejpam-515	308	1	+	+	CCONJ
ejpam-515	308	2	r2(β̂	r2(β̂	NOUN
ejpam-515	308	3	,	,	PUNCT
ejpam-515	308	4	n	n	CCONJ
ejpam-515	308	5	)	)	PUNCT
ejpam-515	309	1	=	=	NOUN
ejpam-515	309	2	o(log	o(log	PROPN
ejpam-515	309	3	log	log	PROPN
ejpam-515	309	4	n	n	CCONJ
ejpam-515	309	5	)	)	PUNCT
ejpam-515	309	6	a.s	a.s	PROPN
ejpam-515	309	7	.	.	PROPN
ejpam-515	309	8	(	(	PUNCT
ejpam-515	309	9	51	51	NUM
ejpam-515	309	10	)	)	PUNCT
ejpam-515	309	11	by	by	ADP
ejpam-515	309	12	the	the	DET
ejpam-515	309	13	definition	definition	NOUN
ejpam-515	309	14	of	of	ADP
ejpam-515	309	15	h(β	h(β	PROPN
ejpam-515	309	16	,	,	PUNCT
ejpam-515	309	17	n	n	CCONJ
ejpam-515	309	18	)	)	PUNCT
ejpam-515	309	19	.	.	PUNCT
ejpam-515	310	1	the	the	DET
ejpam-515	310	2	inequality	inequality	NOUN
ejpam-515	310	3	part	part	NOUN
ejpam-515	310	4	of	of	ADP
ejpam-515	310	5	(	(	PUNCT
ejpam-515	310	6	51	51	NUM
ejpam-515	310	7	)	)	PUNCT
ejpam-515	310	8	is	be	AUX
ejpam-515	310	9	obvious	obvious	ADJ
ejpam-515	310	10	because	because	SCONJ
ejpam-515	310	11	β̂	β̂	ADP
ejpam-515	310	12	is	be	AUX
ejpam-515	310	13	the	the	DET
ejpam-515	310	14	mle	mle	NOUN
ejpam-515	310	15	of	of	ADP
ejpam-515	310	16	β	β	PROPN
ejpam-515	310	17	.	.	PUNCT
ejpam-515	311	1	note	note	VERB
ejpam-515	311	2	that	that	DET
ejpam-515	311	3	result	result	NOUN
ejpam-515	311	4	(	(	PUNCT
ejpam-515	311	5	40	40	NUM
ejpam-515	311	6	)	)	PUNCT
ejpam-515	311	7	is	be	AUX
ejpam-515	311	8	also	also	ADV
ejpam-515	311	9	valid	valid	ADJ
ejpam-515	311	10	here	here	ADV
ejpam-515	311	11	,	,	PUNCT
ejpam-515	311	12	so	so	ADV
ejpam-515	311	13	by	by	ADP
ejpam-515	311	14	theorem	theorem	NOUN
ejpam-515	311	15	1	1	NUM
ejpam-515	311	16	and	and	CCONJ
ejpam-515	311	17	conditions	condition	NOUN
ejpam-515	311	18	(	(	PUNCT
ejpam-515	311	19	c.2	c.2	NOUN
ejpam-515	311	20	)	)	PUNCT
ejpam-515	311	21	and	and	CCONJ
ejpam-515	311	22	(	(	PUNCT
ejpam-515	311	23	c.3	c.3	NOUN
ejpam-515	311	24	)	)	PUNCT
ejpam-515	311	25	0≤	0≤	NUM
ejpam-515	311	26	r1(β̂	r1(β̂	NUM
ejpam-515	311	27	,	,	PUNCT
ejpam-515	311	28	n)≤	n)≤	NOUN
ejpam-515	311	29	1	1	NUM
ejpam-515	311	30	2	2	NUM
ejpam-515	311	31	e2	e2	NOUN
ejpam-515	311	32	p	p	NOUN
ejpam-515	311	33	b2nξn||β̂−β0||λp{in(β0)}||β̂−β0||2	b2nξn||β̂−β0||λp{in(β0)}||β̂−β0||2	NOUN
ejpam-515	311	34	=	=	PUNCT
ejpam-515	311	35	o(log	o(log	PROPN
ejpam-515	311	36	log	log	PROPN
ejpam-515	311	37	n	n	CCONJ
ejpam-515	311	38	)	)	PUNCT
ejpam-515	311	39	a.s	a.s	PROPN
ejpam-515	311	40	..	..	PUNCT
ejpam-515	311	41	(	(	PUNCT
ejpam-515	311	42	52	52	NUM
ejpam-515	311	43	)	)	PUNCT
ejpam-515	311	44	on	on	ADP
ejpam-515	311	45	the	the	DET
ejpam-515	311	46	other	other	ADJ
ejpam-515	311	47	hand	hand	NOUN
ejpam-515	311	48	,	,	PUNCT
ejpam-515	311	49	by	by	ADP
ejpam-515	311	50	result	result	NOUN
ejpam-515	311	51	(	(	PUNCT
ejpam-515	311	52	12	12	NUM
ejpam-515	311	53	)	)	PUNCT
ejpam-515	311	54	of	of	ADP
ejpam-515	311	55	lemma	lemma	PROPN
ejpam-515	311	56	4	4	NUM
ejpam-515	311	57	and	and	CCONJ
ejpam-515	311	58	(	(	PUNCT
ejpam-515	311	59	4	4	NUM
ejpam-515	311	60	)	)	PUNCT
ejpam-515	311	61	of	of	ADP
ejpam-515	311	62	theorem	theorem	NOUN
ejpam-515	311	63	1	1	NUM
ejpam-515	311	64	we	we	PRON
ejpam-515	311	65	have	have	VERB
ejpam-515	311	66	|r2(β̂	|r2(β̂	NOUN
ejpam-515	311	67	,	,	PUNCT
ejpam-515	311	68	n)|	n)|	ADJ
ejpam-515	311	69	≤	≤	PROPN
ejpam-515	311	70	||	||	PUNCT
ejpam-515	312	1	n	n	CCONJ
ejpam-515	312	2	∑	∑	PUNCT
ejpam-515	312	3	k=1	k=1	PROPN
ejpam-515	312	4	(	(	PUNCT
ejpam-515	312	5	yk	yk	PROPN
ejpam-515	312	6	−µ0k)xk||	−µ0k)xk||	PROPN
ejpam-515	312	7	·	·	PUNCT
ejpam-515	312	8	||β̂	||β̂	X
ejpam-515	313	1	−	−	NOUN
ejpam-515	314	1	β0||	β0||	PROPN
ejpam-515	314	2	=	=	PROPN
ejpam-515	314	3	o(log	o(log	PROPN
ejpam-515	314	4	log	log	PROPN
ejpam-515	314	5	n	n	CCONJ
ejpam-515	314	6	)	)	PUNCT
ejpam-515	314	7	a.s	a.s	PROPN
ejpam-515	314	8	..	..	PUNCT
ejpam-515	314	9	(	(	PUNCT
ejpam-515	314	10	53	53	NUM
ejpam-515	314	11	)	)	PUNCT
ejpam-515	314	12	by	by	ADP
ejpam-515	314	13	(	(	PUNCT
ejpam-515	314	14	52	52	NUM
ejpam-515	314	15	)	)	PUNCT
ejpam-515	314	16	and	and	CCONJ
ejpam-515	314	17	(	(	PUNCT
ejpam-515	314	18	53	53	NUM
ejpam-515	314	19	)	)	PUNCT
ejpam-515	314	20	it	it	PRON
ejpam-515	314	21	follows	follow	VERB
ejpam-515	314	22	that	that	SCONJ
ejpam-515	314	23	|h(β̂	|h(β̂	PROPN
ejpam-515	314	24	,	,	PUNCT
ejpam-515	314	25	n)|	n)|	NOUN
ejpam-515	314	26	=	=	SYM
ejpam-515	314	27	o(log	o(log	PROPN
ejpam-515	314	28	log	log	PROPN
ejpam-515	314	29	n	n	CCONJ
ejpam-515	314	30	)	)	PUNCT
ejpam-515	314	31	a.s	a.s	PROPN
ejpam-515	314	32	.	.	PROPN
ejpam-515	314	33	which	which	PRON
ejpam-515	314	34	suffices	suffice	VERB
ejpam-515	314	35	the	the	DET
ejpam-515	314	36	proof	proof	NOUN
ejpam-515	314	37	of	of	ADP
ejpam-515	314	38	the	the	DET
ejpam-515	314	39	theorem	theorem	NOUN
ejpam-515	314	40	.	.	PUNCT
ejpam-515	315	1	proof	proof	NOUN
ejpam-515	315	2	.	.	PUNCT
ejpam-515	316	1	(	(	PUNCT
ejpam-515	316	2	theorem	theorem	NOUN
ejpam-515	316	3	3	3	NUM
ejpam-515	316	4	)	)	PUNCT
ejpam-515	316	5	first	first	ADV
ejpam-515	316	6	we	we	PRON
ejpam-515	316	7	extend	extend	VERB
ejpam-515	316	8	the	the	DET
ejpam-515	316	9	pα×1	pα×1	PROPN
ejpam-515	316	10	vector	vector	NOUN
ejpam-515	316	11	β̂(α	β̂(α	PUNCT
ejpam-515	316	12	)	)	PUNCT
ejpam-515	316	13	,	,	PUNCT
ejpam-515	316	14	the	the	DET
ejpam-515	316	15	mle	mle	NOUN
ejpam-515	316	16	of	of	ADP
ejpam-515	316	17	βα	βα	PROPN
ejpam-515	316	18	,	,	PUNCT
ejpam-515	316	19	to	to	ADP
ejpam-515	316	20	a	a	DET
ejpam-515	316	21	p×1	p×1	PROPN
ejpam-515	316	22	vector	vector	NOUN
ejpam-515	316	23	β̂∗(α	β̂∗(α	PROPN
ejpam-515	316	24	)	)	PUNCT
ejpam-515	316	25	by	by	ADP
ejpam-515	316	26	inserting	insert	VERB
ejpam-515	316	27	p−	p−	NOUN
ejpam-515	316	28	pα	pα	NOUN
ejpam-515	316	29	0	0	NUM
ejpam-515	316	30	’s	’s	NOUN
ejpam-515	316	31	into	into	ADP
ejpam-515	316	32	β̂(α	β̂(α	PRON
ejpam-515	316	33	)	)	PUNCT
ejpam-515	316	34	in	in	ADP
ejpam-515	316	35	such	such	DET
ejpam-515	316	36	a	a	DET
ejpam-515	316	37	way	way	NOUN
ejpam-515	316	38	that	that	PRON
ejpam-515	316	39	the	the	DET
ejpam-515	316	40	sub	sub	NOUN
ejpam-515	316	41	-	-	NOUN
ejpam-515	316	42	vector	vector	NOUN
ejpam-515	316	43	of	of	ADP
ejpam-515	316	44	β̂∗(α	β̂∗(α	NOUN
ejpam-515	316	45	)	)	PUNCT
ejpam-515	316	46	indexed	index	VERB
ejpam-515	316	47	by	by	ADP
ejpam-515	316	48	α	α	PROPN
ejpam-515	316	49	is	be	AUX
ejpam-515	316	50	equal	equal	ADJ
ejpam-515	316	51	to	to	ADP
ejpam-515	316	52	β̂(α	β̂(α	PUNCT
ejpam-515	316	53	)	)	PUNCT
ejpam-515	316	54	.	.	PUNCT
ejpam-515	317	1	then	then	ADV
ejpam-515	317	2	it	it	PRON
ejpam-515	317	3	is	be	AUX
ejpam-515	317	4	easy	easy	ADJ
ejpam-515	317	5	to	to	PART
ejpam-515	317	6	see	see	VERB
ejpam-515	317	7	that	that	DET
ejpam-515	317	8	proving	prove	VERB
ejpam-515	317	9	(	(	PUNCT
ejpam-515	317	10	7	7	NUM
ejpam-515	317	11	)	)	PUNCT
ejpam-515	317	12	is	be	AUX
ejpam-515	317	13	equivalent	equivalent	ADJ
ejpam-515	317	14	to	to	ADP
ejpam-515	317	15	proving	prove	VERB
ejpam-515	317	16	lim	lim	PROPN
ejpam-515	317	17	inf	inf	PROPN
ejpam-515	317	18	n→∞	n→∞	X
ejpam-515	317	19	n−1h(β̂∗(α	n−1h(β̂∗(α	NOUN
ejpam-515	317	20	)	)	PUNCT
ejpam-515	317	21	,	,	PUNCT
ejpam-515	317	22	n	n	CCONJ
ejpam-515	317	23	)	)	PUNCT
ejpam-515	317	24	>	>	X
ejpam-515	317	25	0	0	PUNCT
ejpam-515	318	1	a.s	a.s	PROPN
ejpam-515	318	2	.	.	PROPN
ejpam-515	318	3	for	for	ADP
ejpam-515	318	4	any	any	DET
ejpam-515	318	5	incorrect	incorrect	ADJ
ejpam-515	318	6	model	model	NOUN
ejpam-515	318	7	α	α	NOUN
ejpam-515	318	8	∈aw	∈aw	NOUN
ejpam-515	318	9	.	.	PUNCT
ejpam-515	319	1	(	(	PUNCT
ejpam-515	319	2	54	54	NUM
ejpam-515	319	3	)	)	PUNCT
ejpam-515	319	4	define	define	VERB
ejpam-515	319	5	a0	a0	NOUN
ejpam-515	319	6	=	=	SYM
ejpam-515	319	7	{	{	PUNCT
ejpam-515	319	8	β	β	X
ejpam-515	319	9	:	:	PUNCT
ejpam-515	319	10	||β	||β	NOUN
ejpam-515	319	11	−	−	NOUN
ejpam-515	319	12	β0||	β0||	PROPN
ejpam-515	319	13	≤	≤	NOUN
ejpam-515	319	14	1	1	NUM
ejpam-515	319	15	2	2	NUM
ejpam-515	319	16	min	min	NOUN
ejpam-515	319	17	1≤i≤pα0	1≤i≤pα0	NUM
ejpam-515	319	18	|β0(α0)i|=	|β0(α0)i|=	NOUN
ejpam-515	319	19	b	b	NOUN
ejpam-515	319	20	}	}	PUNCT
ejpam-515	319	21	.	.	PUNCT
ejpam-515	320	1	g.	g.	PROPN
ejpam-515	320	2	qian	qian	PROPN
ejpam-515	320	3	/	/	SYM
ejpam-515	320	4	eur	eur	PROPN
ejpam-515	320	5	.	.	PUNCT
ejpam-515	321	1	j.	j.	PROPN
ejpam-515	321	2	pure	pure	PROPN
ejpam-515	321	3	appl	appl	PROPN
ejpam-515	321	4	.	.	PROPN
ejpam-515	321	5	math	math	PROPN
ejpam-515	321	6	,	,	PUNCT
ejpam-515	321	7	3	3	NUM
ejpam-515	321	8	(	(	PUNCT
ejpam-515	321	9	2010	2010	NUM
ejpam-515	321	10	)	)	PUNCT
ejpam-515	321	11	,	,	PUNCT
ejpam-515	321	12	417	417	NUM
ejpam-515	321	13	-	-	SYM
ejpam-515	321	14	434	434	NUM
ejpam-515	321	15	431	431	NUM
ejpam-515	321	16	clearly	clearly	ADV
ejpam-515	321	17	a0	a0	PROPN
ejpam-515	321	18	is	be	AUX
ejpam-515	321	19	a	a	DET
ejpam-515	321	20	compact	compact	ADJ
ejpam-515	321	21	set	set	NOUN
ejpam-515	321	22	;	;	PUNCT
ejpam-515	321	23	and	and	CCONJ
ejpam-515	321	24	for	for	ADP
ejpam-515	321	25	any	any	DET
ejpam-515	321	26	incorrect	incorrect	ADJ
ejpam-515	321	27	model	model	NOUN
ejpam-515	321	28	α	α	X
ejpam-515	321	29	∈	∈	PROPN
ejpam-515	322	1	aw	aw	INTJ
ejpam-515	322	2	we	we	PRON
ejpam-515	322	3	have	have	VERB
ejpam-515	322	4	β̂∗(α	β̂∗(α	NOUN
ejpam-515	322	5	)	)	PUNCT
ejpam-515	323	1	6∈	6∈	PROPN
ejpam-515	323	2	a0	a0	PROPN
ejpam-515	323	3	because	because	SCONJ
ejpam-515	323	4	||β̂∗(α)−β0||	||β̂∗(α)−β0||	PROPN
ejpam-515	323	5	≥	≥	NOUN
ejpam-515	323	6	2b	2b	NOUN
ejpam-515	323	7	.	.	PUNCT
ejpam-515	324	1	moreover	moreover	ADV
ejpam-515	324	2	,	,	PUNCT
ejpam-515	324	3	by	by	ADP
ejpam-515	324	4	theorem	theorem	NOUN
ejpam-515	324	5	1	1	NUM
ejpam-515	324	6	,	,	PUNCT
ejpam-515	324	7	the	the	DET
ejpam-515	324	8	mle	mle	NOUN
ejpam-515	324	9	β̂	β̂	X
ejpam-515	324	10	for	for	ADP
ejpam-515	324	11	the	the	DET
ejpam-515	324	12	full	full	ADJ
ejpam-515	324	13	model	model	NOUN
ejpam-515	324	14	is	be	AUX
ejpam-515	324	15	an	an	DET
ejpam-515	324	16	interior	interior	ADJ
ejpam-515	324	17	point	point	NOUN
ejpam-515	324	18	of	of	ADP
ejpam-515	324	19	a0	a0	NOUN
ejpam-515	324	20	almost	almost	ADV
ejpam-515	324	21	surely	surely	ADV
ejpam-515	324	22	when	when	SCONJ
ejpam-515	324	23	n	n	PRON
ejpam-515	324	24	is	be	AUX
ejpam-515	324	25	sufficiently	sufficiently	ADV
ejpam-515	324	26	large	large	ADJ
ejpam-515	324	27	.	.	PUNCT
ejpam-515	325	1	since	since	SCONJ
ejpam-515	325	2	h(β	h(β	PROPN
ejpam-515	325	3	,	,	PUNCT
ejpam-515	325	4	n	n	CCONJ
ejpam-515	325	5	)	)	PUNCT
ejpam-515	325	6	is	be	AUX
ejpam-515	325	7	convex	convex	ADJ
ejpam-515	325	8	with	with	ADP
ejpam-515	325	9	respect	respect	NOUN
ejpam-515	325	10	to	to	ADP
ejpam-515	325	11	β	β	PRON
ejpam-515	325	12	,	,	PUNCT
ejpam-515	325	13	it	it	PRON
ejpam-515	325	14	follows	follow	VERB
ejpam-515	325	15	that	that	SCONJ
ejpam-515	325	16	h(β̂∗(α	h(β̂∗(α	PROPN
ejpam-515	325	17	)	)	PUNCT
ejpam-515	325	18	,	,	PUNCT
ejpam-515	325	19	n)≥	n)≥	PROPN
ejpam-515	325	20	inf	inf	PROPN
ejpam-515	325	21	β∈∂	β∈∂	CCONJ
ejpam-515	325	22	a0	a0	PROPN
ejpam-515	325	23	h(β	h(β	PROPN
ejpam-515	325	24	,	,	PUNCT
ejpam-515	325	25	n	n	CCONJ
ejpam-515	325	26	)	)	PUNCT
ejpam-515	325	27	where	where	SCONJ
ejpam-515	325	28	∂a0	∂a0	NOUN
ejpam-515	325	29	is	be	AUX
ejpam-515	325	30	the	the	DET
ejpam-515	325	31	boundary	boundary	NOUN
ejpam-515	325	32	of	of	ADP
ejpam-515	325	33	a0	a0	PROPN
ejpam-515	325	34	.	.	PUNCT
ejpam-515	326	1	now	now	ADV
ejpam-515	326	2	it	it	PRON
ejpam-515	326	3	is	be	AUX
ejpam-515	326	4	sufficient	sufficient	ADJ
ejpam-515	326	5	to	to	PART
ejpam-515	326	6	prove	prove	VERB
ejpam-515	326	7	lim	lim	PROPN
ejpam-515	326	8	inf	inf	PROPN
ejpam-515	326	9	n→∞	n→∞	X
ejpam-515	326	10	inf	inf	PROPN
ejpam-515	326	11	β∈∂	β∈∂	CCONJ
ejpam-515	326	12	a0	a0	PROPN
ejpam-515	326	13	n−1h(β	n−1h(β	PROPN
ejpam-515	326	14	,	,	PUNCT
ejpam-515	326	15	n	n	CCONJ
ejpam-515	326	16	)	)	PUNCT
ejpam-515	326	17	>	>	X
ejpam-515	326	18	0	0	PUNCT
ejpam-515	327	1	a.s	a.s	PROPN
ejpam-515	327	2	.	.	PROPN
ejpam-515	327	3	(	(	PUNCT
ejpam-515	327	4	55	55	NUM
ejpam-515	327	5	)	)	PUNCT
ejpam-515	327	6	in	in	ADP
ejpam-515	327	7	order	order	NOUN
ejpam-515	327	8	to	to	PART
ejpam-515	327	9	prove	prove	VERB
ejpam-515	327	10	(	(	PUNCT
ejpam-515	327	11	54	54	NUM
ejpam-515	327	12	)	)	PUNCT
ejpam-515	327	13	.	.	PUNCT
ejpam-515	328	1	using	use	VERB
ejpam-515	328	2	result	result	NOUN
ejpam-515	328	3	(	(	PUNCT
ejpam-515	328	4	iii	iii	NOUN
ejpam-515	328	5	)	)	PUNCT
ejpam-515	328	6	of	of	ADP
ejpam-515	328	7	lemma	lemma	PROPN
ejpam-515	328	8	1	1	NUM
ejpam-515	328	9	,	,	PUNCT
ejpam-515	328	10	(	(	PUNCT
ejpam-515	328	11	8)	8)	NUM
ejpam-515	328	12	,	,	PUNCT
ejpam-515	328	13	condition	condition	NOUN
ejpam-515	328	14	(	(	PUNCT
ejpam-515	328	15	c.7	c.7	NOUN
ejpam-515	328	16	)	)	PUNCT
ejpam-515	328	17	and	and	CCONJ
ejpam-515	328	18	cauchy	cauchy	PROPN
ejpam-515	328	19	-	-	PUNCT
ejpam-515	328	20	schwarz	schwarz	PROPN
ejpam-515	328	21	inequality	inequality	NOUN
ejpam-515	328	22	one	one	PRON
ejpam-515	328	23	can	can	AUX
ejpam-515	328	24	show	show	VERB
ejpam-515	328	25	that	that	SCONJ
ejpam-515	328	26	r1(β	r1(β	PROPN
ejpam-515	328	27	,	,	PUNCT
ejpam-515	328	28	n)i(β∈∂a0)≥	n)i(β∈∂a0)≥	PROPN
ejpam-515	328	29	1	1	NUM
ejpam-515	328	30	2	2	NUM
ejpam-515	328	31	n	n	NOUN
ejpam-515	328	32	∑	∑	PUNCT
ejpam-515	328	33	k=1	k=1	PROPN
ejpam-515	328	34	e−2||xk	e−2||xk	PROPN
ejpam-515	328	35	||·||β−β0||µ0k[x	||·||β−β0||µ0k[x	PROPN
ejpam-515	328	36	t	t	PROPN
ejpam-515	328	37	k	k	PROPN
ejpam-515	328	38	(	(	PUNCT
ejpam-515	328	39	β	β	X
ejpam-515	328	40	−	−	PROPN
ejpam-515	328	41	β0	β0	NOUN
ejpam-515	328	42	)	)	PUNCT
ejpam-515	328	43	]	]	PUNCT
ejpam-515	328	44	2i(β∈∂a0	2i(β∈∂a0	X
ejpam-515	328	45	)	)	PUNCT
ejpam-515	328	46	=	=	SYM
ejpam-515	328	47	1	1	NUM
ejpam-515	328	48	2	2	NUM
ejpam-515	328	49	(	(	PUNCT
ejpam-515	328	50	β	β	NOUN
ejpam-515	328	51	−	−	PROPN
ejpam-515	328	52	β0	β0	PROPN
ejpam-515	328	53	)	)	PUNCT
ejpam-515	328	54	tx	tx	PROPN
ejpam-515	328	55	t	t	PROPN
ejpam-515	328	56	nmnxn(β	nmnxn(β	NOUN
ejpam-515	328	57	−	−	PROPN
ejpam-515	328	58	β0)i(β∈∂a0	β0)i(β∈∂a0	NUM
ejpam-515	328	59	)	)	PUNCT
ejpam-515	328	60	≥	≥	NOUN
ejpam-515	328	61	1	1	NUM
ejpam-515	328	62	2	2	NUM
ejpam-515	328	63	λ1{x	λ1{x	NOUN
ejpam-515	328	64	t	t	PROPN
ejpam-515	328	65	nmnxn}||β	nmnxn}||β	PROPN
ejpam-515	328	66	−	−	PROPN
ejpam-515	328	67	β0||2i(β∈∂a0)≥	β0||2i(β∈∂a0)≥	NOUN
ejpam-515	328	68	1	1	NUM
ejpam-515	328	69	2	2	NUM
ejpam-515	328	70	b3	b3	NOUN
ejpam-515	328	71	b2n	b2n	NOUN
ejpam-515	328	72	,	,	PUNCT
ejpam-515	328	73	which	which	PRON
ejpam-515	328	74	suggests	suggest	VERB
ejpam-515	328	75	inf	inf	PROPN
ejpam-515	328	76	β∈∂	β∈∂	CCONJ
ejpam-515	328	77	a0	a0	PROPN
ejpam-515	328	78	r1(β	r1(β	PROPN
ejpam-515	328	79	,	,	PUNCT
ejpam-515	328	80	n)≥	n)≥	PROPN
ejpam-515	328	81	1	1	NUM
ejpam-515	328	82	2	2	NUM
ejpam-515	328	83	b3	b3	PROPN
ejpam-515	328	84	b2n	b2n	NOUN
ejpam-515	328	85	.	.	PUNCT
ejpam-515	329	1	(	(	PUNCT
ejpam-515	329	2	56	56	NUM
ejpam-515	329	3	)	)	PUNCT
ejpam-515	329	4	following	follow	VERB
ejpam-515	329	5	the	the	DET
ejpam-515	329	6	same	same	ADJ
ejpam-515	329	7	line	line	NOUN
ejpam-515	329	8	as	as	ADP
ejpam-515	329	9	proving	prove	VERB
ejpam-515	329	10	(	(	PUNCT
ejpam-515	329	11	36	36	NUM
ejpam-515	329	12	)	)	PUNCT
ejpam-515	329	13	one	one	NOUN
ejpam-515	329	14	can	can	AUX
ejpam-515	329	15	show	show	VERB
ejpam-515	329	16	that	that	DET
ejpam-515	329	17	sup	sup	NOUN
ejpam-515	329	18	β∈∂	β∈∂	CCONJ
ejpam-515	329	19	a0	a0	NOUN
ejpam-515	329	20	|r2(β	|r2(β	PROPN
ejpam-515	329	21	,	,	PUNCT
ejpam-515	329	22	n)|	n)|	PROPN
ejpam-515	329	23	=	=	SYM
ejpam-515	329	24	o	o	PROPN
ejpam-515	329	25	(	(	PUNCT
ejpam-515	329	26	p	p	NOUN
ejpam-515	329	27	n	n	CCONJ
ejpam-515	329	28	log	log	VERB
ejpam-515	329	29	log	log	NOUN
ejpam-515	329	30	n	n	CCONJ
ejpam-515	329	31	)	)	PUNCT
ejpam-515	329	32	a.s	a.s	PROPN
ejpam-515	329	33	..	..	PUNCT
ejpam-515	329	34	(	(	PUNCT
ejpam-515	329	35	57	57	NUM
ejpam-515	329	36	)	)	PUNCT
ejpam-515	329	37	since	since	SCONJ
ejpam-515	329	38	infβ∈∂	infβ∈∂	PROPN
ejpam-515	329	39	a0	a0	PROPN
ejpam-515	329	40	h(β	h(β	PROPN
ejpam-515	329	41	,	,	PUNCT
ejpam-515	329	42	n	n	CCONJ
ejpam-515	329	43	)	)	PUNCT
ejpam-515	329	44	≥	≥	NOUN
ejpam-515	329	45	infβ∈∂	infβ∈∂	PROPN
ejpam-515	329	46	a0	a0	PROPN
ejpam-515	329	47	r1(β	r1(β	PROPN
ejpam-515	329	48	,	,	PUNCT
ejpam-515	329	49	n)−	n)−	PROPN
ejpam-515	329	50	supβ∈∂	supβ∈∂	PROPN
ejpam-515	329	51	a0	a0	NOUN
ejpam-515	329	52	|r2(β	|r2(β	PROPN
ejpam-515	329	53	,	,	PUNCT
ejpam-515	329	54	n)|	n)|	PROPN
ejpam-515	329	55	by	by	ADP
ejpam-515	329	56	(	(	PUNCT
ejpam-515	329	57	8)	8)	NUM
ejpam-515	329	58	,	,	PUNCT
ejpam-515	329	59	it	it	PRON
ejpam-515	329	60	follows	follow	VERB
ejpam-515	329	61	from	from	ADP
ejpam-515	329	62	(	(	PUNCT
ejpam-515	329	63	56	56	NUM
ejpam-515	329	64	)	)	PUNCT
ejpam-515	329	65	and	and	CCONJ
ejpam-515	329	66	(	(	PUNCT
ejpam-515	329	67	57	57	NUM
ejpam-515	329	68	)	)	PUNCT
ejpam-515	329	69	that	that	SCONJ
ejpam-515	329	70	(	(	PUNCT
ejpam-515	329	71	55	55	NUM
ejpam-515	329	72	)	)	PUNCT
ejpam-515	329	73	is	be	AUX
ejpam-515	329	74	true	true	ADJ
ejpam-515	329	75	and	and	CCONJ
ejpam-515	329	76	consequently	consequently	ADV
ejpam-515	329	77	(	(	PUNCT
ejpam-515	329	78	54	54	NUM
ejpam-515	329	79	)	)	PUNCT
ejpam-515	329	80	is	be	AUX
ejpam-515	329	81	true	true	ADJ
ejpam-515	329	82	.	.	PUNCT
ejpam-515	330	1	5	5	X
ejpam-515	330	2	.	.	X
ejpam-515	330	3	discussion	discussion	NOUN
ejpam-515	330	4	the	the	DET
ejpam-515	330	5	asymptotic	asymptotic	ADJ
ejpam-515	330	6	results	result	NOUN
ejpam-515	330	7	obtained	obtain	VERB
ejpam-515	330	8	in	in	ADP
ejpam-515	330	9	this	this	DET
ejpam-515	330	10	paper	paper	NOUN
ejpam-515	330	11	are	be	AUX
ejpam-515	330	12	based	base	VERB
ejpam-515	330	13	on	on	ADP
ejpam-515	330	14	the	the	DET
ejpam-515	330	15	assumption	assumption	NOUN
ejpam-515	330	16	that	that	SCONJ
ejpam-515	330	17	the	the	DET
ejpam-515	330	18	response	response	NOUN
ejpam-515	330	19	variable	variable	NOUN
ejpam-515	330	20	follows	follow	VERB
ejpam-515	330	21	a	a	DET
ejpam-515	330	22	poisson	poisson	NOUN
ejpam-515	330	23	distribution	distribution	NOUN
ejpam-515	330	24	and	and	CCONJ
ejpam-515	330	25	the	the	DET
ejpam-515	330	26	candidate	candidate	NOUN
ejpam-515	330	27	models	model	NOUN
ejpam-515	330	28	under	under	ADP
ejpam-515	330	29	consideration	consideration	NOUN
ejpam-515	330	30	for	for	ADP
ejpam-515	330	31	selection	selection	NOUN
ejpam-515	330	32	are	be	AUX
ejpam-515	330	33	based	base	VERB
ejpam-515	330	34	on	on	ADP
ejpam-515	330	35	those	those	DET
ejpam-515	330	36	available	available	ADJ
ejpam-515	330	37	explanatory	explanatory	ADJ
ejpam-515	330	38	variables	variable	NOUN
ejpam-515	330	39	.	.	PUNCT
ejpam-515	331	1	in	in	ADP
ejpam-515	331	2	practice	practice	NOUN
ejpam-515	331	3	,	,	PUNCT
ejpam-515	331	4	there	there	PRON
ejpam-515	331	5	may	may	AUX
ejpam-515	331	6	exist	exist	VERB
ejpam-515	331	7	some	some	DET
ejpam-515	331	8	lurking	lurk	VERB
ejpam-515	331	9	variables	variable	NOUN
ejpam-515	331	10	which	which	PRON
ejpam-515	331	11	also	also	ADV
ejpam-515	331	12	affect	affect	VERB
ejpam-515	331	13	the	the	DET
ejpam-515	331	14	response	response	NOUN
ejpam-515	331	15	variable	variable	NOUN
ejpam-515	331	16	.	.	PUNCT
ejpam-515	332	1	in	in	ADP
ejpam-515	332	2	this	this	DET
ejpam-515	332	3	situation	situation	NOUN
ejpam-515	332	4	,	,	PUNCT
ejpam-515	332	5	a	a	DET
ejpam-515	332	6	mixture	mixture	NOUN
ejpam-515	332	7	poisson	poisson	NOUN
ejpam-515	332	8	distribution	distribution	NOUN
ejpam-515	332	9	may	may	AUX
ejpam-515	332	10	be	be	AUX
ejpam-515	332	11	introduced	introduce	VERB
ejpam-515	332	12	for	for	ADP
ejpam-515	332	13	modeling	model	VERB
ejpam-515	332	14	the	the	DET
ejpam-515	332	15	response	response	NOUN
ejpam-515	332	16	variable	variable	NOUN
ejpam-515	332	17	,	,	PUNCT
ejpam-515	332	18	in	in	ADP
ejpam-515	332	19	which	which	PRON
ejpam-515	332	20	an	an	DET
ejpam-515	332	21	overdispersion	overdispersion	NOUN
ejpam-515	332	22	parameter	parameter	NOUN
ejpam-515	332	23	is	be	AUX
ejpam-515	332	24	used	use	VERB
ejpam-515	332	25	to	to	PART
ejpam-515	332	26	account	account	VERB
ejpam-515	332	27	for	for	ADP
ejpam-515	332	28	the	the	DET
ejpam-515	332	29	effects	effect	NOUN
ejpam-515	332	30	of	of	ADP
ejpam-515	332	31	the	the	DET
ejpam-515	332	32	lurking	lurking	NOUN
ejpam-515	332	33	variables	variable	NOUN
ejpam-515	332	34	.	.	PUNCT
ejpam-515	333	1	we	we	PRON
ejpam-515	333	2	refer	refer	VERB
ejpam-515	333	3	to	to	ADP
ejpam-515	333	4	[	[	X
ejpam-515	333	5	8	8	NUM
ejpam-515	333	6	,	,	PUNCT
ejpam-515	333	7	section	section	NOUN
ejpam-515	333	8	6.2.3	6.2.3	NUM
ejpam-515	333	9	]	]	PUNCT
ejpam-515	333	10	for	for	ADP
ejpam-515	333	11	details	detail	NOUN
ejpam-515	333	12	of	of	ADP
ejpam-515	333	13	the	the	DET
ejpam-515	333	14	over	over	ADP
ejpam-515	333	15	-	-	PUNCT
ejpam-515	333	16	dispersion	dispersion	NOUN
ejpam-515	333	17	log	log	NOUN
ejpam-515	333	18	-	-	PUNCT
ejpam-515	333	19	linear	linear	NOUN
ejpam-515	333	20	models	model	NOUN
ejpam-515	333	21	.	.	PUNCT
ejpam-515	334	1	then	then	ADV
ejpam-515	334	2	the	the	DET
ejpam-515	334	3	model	model	NOUN
ejpam-515	334	4	selection	selection	NOUN
ejpam-515	334	5	procedure	procedure	NOUN
ejpam-515	334	6	can	can	AUX
ejpam-515	334	7	still	still	ADV
ejpam-515	334	8	focus	focus	VERB
ejpam-515	334	9	on	on	ADP
ejpam-515	334	10	those	those	DET
ejpam-515	334	11	available	available	ADJ
ejpam-515	334	12	explanatory	explanatory	ADJ
ejpam-515	334	13	variables	variable	NOUN
ejpam-515	334	14	.	.	PUNCT
ejpam-515	335	1	provided	provide	VERB
ejpam-515	335	2	that	that	SCONJ
ejpam-515	335	3	a	a	DET
ejpam-515	335	4	result	result	NOUN
ejpam-515	335	5	similar	similar	ADJ
ejpam-515	335	6	to	to	ADP
ejpam-515	335	7	that	that	PRON
ejpam-515	335	8	in	in	ADP
ejpam-515	335	9	lemma	lemma	PROPN
ejpam-515	335	10	2	2	NUM
ejpam-515	335	11	can	can	AUX
ejpam-515	335	12	be	be	AUX
ejpam-515	335	13	obtained	obtain	VERB
ejpam-515	335	14	for	for	ADP
ejpam-515	335	15	the	the	DET
ejpam-515	335	16	mixture	mixture	NOUN
ejpam-515	335	17	poisson	poisson	NOUN
ejpam-515	335	18	probability	probability	NOUN
ejpam-515	335	19	(	(	PUNCT
ejpam-515	335	20	which	which	PRON
ejpam-515	335	21	may	may	AUX
ejpam-515	335	22	be	be	AUX
ejpam-515	335	23	shown	show	VERB
ejpam-515	335	24	by	by	ADP
ejpam-515	335	25	imitating	imitate	VERB
ejpam-515	335	26	the	the	DET
ejpam-515	335	27	proof	proof	NOUN
ejpam-515	335	28	in	in	ADP
ejpam-515	335	29	[	[	X
ejpam-515	335	30	2	2	NUM
ejpam-515	335	31	]	]	NUM
ejpam-515	335	32	)	)	PUNCT
ejpam-515	335	33	,	,	PUNCT
ejpam-515	335	34	it	it	PRON
ejpam-515	335	35	seems	seem	VERB
ejpam-515	335	36	the	the	DET
ejpam-515	335	37	same	same	ADJ
ejpam-515	335	38	results	result	NOUN
ejpam-515	335	39	as	as	ADP
ejpam-515	335	40	of	of	ADP
ejpam-515	335	41	theorems	theorem	NOUN
ejpam-515	335	42	1	1	NUM
ejpam-515	335	43	to	to	PART
ejpam-515	335	44	4	4	NUM
ejpam-515	335	45	can	can	AUX
ejpam-515	335	46	also	also	ADV
ejpam-515	335	47	references	reference	NOUN
ejpam-515	335	48	432	432	NUM
ejpam-515	335	49	be	be	AUX
ejpam-515	335	50	established	establish	VERB
ejpam-515	335	51	for	for	ADP
ejpam-515	335	52	the	the	DET
ejpam-515	335	53	over	over	ADP
ejpam-515	335	54	-	-	PUNCT
ejpam-515	335	55	dispersion	dispersion	NOUN
ejpam-515	335	56	log	log	NOUN
ejpam-515	335	57	-	-	PUNCT
ejpam-515	335	58	linear	linear	NOUN
ejpam-515	335	59	models	model	NOUN
ejpam-515	335	60	using	use	VERB
ejpam-515	335	61	the	the	DET
ejpam-515	335	62	same	same	ADJ
ejpam-515	335	63	methods	method	NOUN
ejpam-515	335	64	employed	employ	VERB
ejpam-515	335	65	in	in	ADP
ejpam-515	335	66	this	this	DET
ejpam-515	335	67	paper	paper	NOUN
ejpam-515	335	68	.	.	PUNCT
ejpam-515	336	1	it	it	PRON
ejpam-515	336	2	is	be	AUX
ejpam-515	336	3	also	also	ADV
ejpam-515	336	4	possible	possible	ADJ
ejpam-515	336	5	to	to	PART
ejpam-515	336	6	extend	extend	VERB
ejpam-515	336	7	our	our	PRON
ejpam-515	336	8	asymptotic	asymptotic	ADJ
ejpam-515	336	9	results	result	NOUN
ejpam-515	336	10	to	to	ADP
ejpam-515	336	11	the	the	DET
ejpam-515	336	12	poisson	poisson	NOUN
ejpam-515	336	13	regression	regression	NOUN
ejpam-515	336	14	models	model	NOUN
ejpam-515	336	15	with	with	ADP
ejpam-515	336	16	the	the	DET
ejpam-515	336	17	link	link	NOUN
ejpam-515	336	18	functions	function	NOUN
ejpam-515	336	19	other	other	ADJ
ejpam-515	336	20	than	than	ADP
ejpam-515	336	21	the	the	DET
ejpam-515	336	22	log	log	NOUN
ejpam-515	336	23	-	-	PUNCT
ejpam-515	336	24	link	link	NOUN
ejpam-515	336	25	,	,	PUNCT
ejpam-515	336	26	which	which	PRON
ejpam-515	336	27	is	be	AUX
ejpam-515	336	28	still	still	ADV
ejpam-515	336	29	under	under	ADP
ejpam-515	336	30	our	our	PRON
ejpam-515	336	31	investigation	investigation	NOUN
ejpam-515	336	32	but	but	CCONJ
ejpam-515	336	33	shall	shall	AUX
ejpam-515	336	34	be	be	AUX
ejpam-515	336	35	presented	present	VERB
ejpam-515	336	36	somewhere	somewhere	ADV
ejpam-515	336	37	else	else	ADV
ejpam-515	336	38	.	.	PUNCT
ejpam-515	337	1	in	in	ADP
ejpam-515	337	2	addition	addition	NOUN
ejpam-515	337	3	to	to	ADP
ejpam-515	337	4	the	the	DET
ejpam-515	337	5	model	model	NOUN
ejpam-515	337	6	selection	selection	NOUN
ejpam-515	337	7	criteria	criterion	NOUN
ejpam-515	337	8	studied	study	VERB
ejpam-515	337	9	here	here	ADV
ejpam-515	337	10	,	,	PUNCT
ejpam-515	337	11	there	there	PRON
ejpam-515	337	12	are	be	VERB
ejpam-515	337	13	many	many	ADJ
ejpam-515	337	14	other	other	ADJ
ejpam-515	337	15	approaches	approach	NOUN
ejpam-515	337	16	for	for	ADP
ejpam-515	337	17	model	model	NOUN
ejpam-515	337	18	selection	selection	NOUN
ejpam-515	337	19	,	,	PUNCT
ejpam-515	337	20	e.g.	e.g.	ADV
ejpam-515	337	21	the	the	DET
ejpam-515	337	22	hierarchical	hierarchical	ADJ
ejpam-515	337	23	bayesian	bayesian	NOUN
ejpam-515	337	24	approach	approach	NOUN
ejpam-515	337	25	[	[	X
ejpam-515	337	26	5	5	NUM
ejpam-515	337	27	]	]	PUNCT
ejpam-515	337	28	and	and	CCONJ
ejpam-515	337	29	lasso	lasso	NOUN
ejpam-515	337	30	[	[	X
ejpam-515	337	31	19	19	NUM
ejpam-515	337	32	]	]	PUNCT
ejpam-515	337	33	etc	etc	X
ejpam-515	337	34	.	.	X
ejpam-515	337	35	,	,	PUNCT
ejpam-515	337	36	for	for	ADP
ejpam-515	337	37	which	which	PRON
ejpam-515	337	38	our	our	PRON
ejpam-515	337	39	results	result	NOUN
ejpam-515	337	40	may	may	AUX
ejpam-515	337	41	not	not	PART
ejpam-515	337	42	be	be	AUX
ejpam-515	337	43	applicable	applicable	ADJ
ejpam-515	337	44	.	.	PUNCT
ejpam-515	338	1	finally	finally	ADV
ejpam-515	338	2	,	,	PUNCT
ejpam-515	338	3	other	other	ADJ
ejpam-515	338	4	than	than	ADP
ejpam-515	338	5	determining	determine	VERB
ejpam-515	338	6	a	a	DET
ejpam-515	338	7	model	model	NOUN
ejpam-515	338	8	selection	selection	NOUN
ejpam-515	338	9	criterion	criterion	NOUN
ejpam-515	338	10	and	and	CCONJ
ejpam-515	338	11	assessing	assess	VERB
ejpam-515	338	12	its	its	PRON
ejpam-515	338	13	asymptotic	asymptotic	ADJ
ejpam-515	338	14	performance	performance	NOUN
ejpam-515	338	15	,	,	PUNCT
ejpam-515	338	16	there	there	PRON
ejpam-515	338	17	is	be	VERB
ejpam-515	338	18	a	a	DET
ejpam-515	338	19	computational	computational	ADJ
ejpam-515	338	20	issue	issue	NOUN
ejpam-515	338	21	on	on	ADP
ejpam-515	338	22	how	how	SCONJ
ejpam-515	338	23	to	to	PART
ejpam-515	338	24	execute	execute	VERB
ejpam-515	338	25	a	a	DET
ejpam-515	338	26	model	model	NOUN
ejpam-515	338	27	selection	selection	NOUN
ejpam-515	338	28	procedure	procedure	NOUN
ejpam-515	338	29	from	from	ADP
ejpam-515	338	30	many	many	ADJ
ejpam-515	338	31	possible	possible	ADJ
ejpam-515	338	32	candidate	candidate	NOUN
ejpam-515	338	33	models	model	NOUN
ejpam-515	338	34	.	.	PUNCT
ejpam-515	339	1	this	this	PRON
ejpam-515	339	2	becomes	become	VERB
ejpam-515	339	3	especially	especially	ADV
ejpam-515	339	4	important	important	ADJ
ejpam-515	339	5	when	when	SCONJ
ejpam-515	339	6	the	the	DET
ejpam-515	339	7	number	number	NOUN
ejpam-515	339	8	of	of	ADP
ejpam-515	339	9	candidate	candidate	NOUN
ejpam-515	339	10	models	model	NOUN
ejpam-515	339	11	,	,	PUNCT
ejpam-515	339	12	often	often	ADV
ejpam-515	339	13	of	of	ADP
ejpam-515	339	14	the	the	DET
ejpam-515	339	15	magnitude	magnitude	NOUN
ejpam-515	339	16	2p	2p	NOUN
ejpam-515	339	17	,	,	PUNCT
ejpam-515	339	18	is	be	AUX
ejpam-515	339	19	enormous	enormous	ADJ
ejpam-515	339	20	.	.	PUNCT
ejpam-515	340	1	however	however	ADV
ejpam-515	340	2	,	,	PUNCT
ejpam-515	340	3	a	a	DET
ejpam-515	340	4	thorough	thorough	ADJ
ejpam-515	340	5	investigation	investigation	NOUN
ejpam-515	340	6	of	of	ADP
ejpam-515	340	7	this	this	DET
ejpam-515	340	8	issue	issue	NOUN
ejpam-515	340	9	is	be	AUX
ejpam-515	340	10	beyond	beyond	ADP
ejpam-515	340	11	the	the	DET
ejpam-515	340	12	scope	scope	NOUN
ejpam-515	340	13	of	of	ADP
ejpam-515	340	14	this	this	DET
ejpam-515	340	15	paper	paper	NOUN
ejpam-515	340	16	.	.	PUNCT
ejpam-515	341	1	we	we	PRON
ejpam-515	341	2	refer	refer	VERB
ejpam-515	341	3	to	to	ADP
ejpam-515	341	4	[	[	X
ejpam-515	341	5	10	10	NUM
ejpam-515	341	6	]	]	PUNCT
ejpam-515	341	7	and	and	CCONJ
ejpam-515	341	8	[	[	X
ejpam-515	341	9	13	13	NUM
ejpam-515	341	10	]	]	PUNCT
ejpam-515	341	11	for	for	ADP
ejpam-515	341	12	some	some	DET
ejpam-515	341	13	results	result	NOUN
ejpam-515	341	14	in	in	ADP
ejpam-515	341	15	this	this	DET
ejpam-515	341	16	area	area	NOUN
ejpam-515	341	17	.	.	PUNCT
ejpam-515	342	1	references	reference	NOUN
ejpam-515	342	2	[	[	X
ejpam-515	342	3	1	1	NUM
ejpam-515	342	4	]	]	PUNCT
ejpam-515	342	5	h	h	NOUN
ejpam-515	342	6	akaike	akaike	ADJ
ejpam-515	342	7	.	.	PUNCT
ejpam-515	343	1	information	information	NOUN
ejpam-515	343	2	theory	theory	NOUN
ejpam-515	343	3	and	and	CCONJ
ejpam-515	343	4	an	an	DET
ejpam-515	343	5	extension	extension	NOUN
ejpam-515	343	6	of	of	ADP
ejpam-515	343	7	the	the	DET
ejpam-515	343	8	maximum	maximum	ADJ
ejpam-515	343	9	likelihood	likelihood	NOUN
ejpam-515	343	10	principle	principle	NOUN
ejpam-515	343	11	.	.	PUNCT
ejpam-515	344	1	in	in	ADP
ejpam-515	344	2	b.n	b.n	PROPN
ejpam-515	344	3	.	.	PROPN
ejpam-515	344	4	petrov	petrov	PROPN
ejpam-515	344	5	and	and	CCONJ
ejpam-515	344	6	f.	f.	PROPN
ejpam-515	344	7	csáki	csáki	PROPN
ejpam-515	344	8	,	,	PUNCT
ejpam-515	344	9	editors	editor	NOUN
ejpam-515	344	10	,	,	PUNCT
ejpam-515	344	11	proceedings	proceeding	NOUN
ejpam-515	344	12	of	of	ADP
ejpam-515	344	13	the	the	DET
ejpam-515	344	14	second	second	ADJ
ejpam-515	344	15	international	international	ADJ
ejpam-515	344	16	symposium	symposium	NOUN
ejpam-515	344	17	on	on	ADP
ejpam-515	344	18	information	information	NOUN
ejpam-515	344	19	theory	theory	NOUN
ejpam-515	344	20	,	,	PUNCT
ejpam-515	344	21	pages	page	NOUN
ejpam-515	344	22	267–281	267–281	NUM
ejpam-515	344	23	,	,	PUNCT
ejpam-515	344	24	budapest	budapest	NOUN
ejpam-515	344	25	,	,	PUNCT
ejpam-515	344	26	1973	1973	NUM
ejpam-515	344	27	.	.	PUNCT
ejpam-515	345	1	akadémia	akadémia	NOUN
ejpam-515	345	2	kiadó	kiadó	PROPN
ejpam-515	345	3	.	.	PUNCT
ejpam-515	346	1	[	[	X
ejpam-515	346	2	2	2	NUM
ejpam-515	346	3	]	]	PUNCT
ejpam-515	346	4	h	h	NOUN
ejpam-515	346	5	bohman	bohman	NOUN
ejpam-515	346	6	.	.	PUNCT
ejpam-515	347	1	two	two	NUM
ejpam-515	347	2	inequalities	inequality	NOUN
ejpam-515	347	3	for	for	ADP
ejpam-515	347	4	poisson	poisson	NOUN
ejpam-515	347	5	distributions	distribution	NOUN
ejpam-515	347	6	.	.	PUNCT
ejpam-515	348	1	skandinavisk	skandinavisk	NOUN
ejpam-515	348	2	aktuarietidskrift	aktuarietidskrift	PROPN
ejpam-515	348	3	,	,	PUNCT
ejpam-515	348	4	46:47–52	46:47–52	PROPN
ejpam-515	348	5	,	,	PUNCT
ejpam-515	348	6	1963	1963	NUM
ejpam-515	348	7	.	.	PUNCT
ejpam-515	349	1	[	[	X
ejpam-515	349	2	3	3	X
ejpam-515	349	3	]	]	X
ejpam-515	349	4	y	y	PROPN
ejpam-515	349	5	chow	chow	PROPN
ejpam-515	349	6	and	and	CCONJ
ejpam-515	349	7	h	h	NOUN
ejpam-515	349	8	teicher	teicher	ADJ
ejpam-515	349	9	.	.	PUNCT
ejpam-515	350	1	probability	probability	NOUN
ejpam-515	350	2	theory	theory	NOUN
ejpam-515	350	3	:	:	PUNCT
ejpam-515	350	4	independence	independence	NOUN
ejpam-515	350	5	,	,	PUNCT
ejpam-515	350	6	interchangeability	interchangeability	NOUN
ejpam-515	350	7	,	,	PUNCT
ejpam-515	350	8	martingales	martingale	NOUN
ejpam-515	350	9	.	.	PUNCT
ejpam-515	350	10	springer	springer	NOUN
ejpam-515	350	11	,	,	PUNCT
ejpam-515	350	12	new	new	PROPN
ejpam-515	350	13	york	york	PROPN
ejpam-515	350	14	,	,	PUNCT
ejpam-515	350	15	3	3	NUM
ejpam-515	350	16	edition	edition	NOUN
ejpam-515	350	17	,	,	PUNCT
ejpam-515	350	18	1997	1997	NUM
ejpam-515	350	19	.	.	PUNCT
ejpam-515	351	1	[	[	X
ejpam-515	351	2	4	4	NUM
ejpam-515	351	3	]	]	PUNCT
ejpam-515	351	4	e	e	X
ejpam-515	351	5	george	george	PROPN
ejpam-515	351	6	.	.	PUNCT
ejpam-515	352	1	statistics	statistic	NOUN
ejpam-515	352	2	in	in	ADP
ejpam-515	352	3	the	the	DET
ejpam-515	352	4	21st	21st	ADJ
ejpam-515	352	5	century	century	NOUN
ejpam-515	352	6	,	,	PUNCT
ejpam-515	352	7	chapter	chapter	NOUN
ejpam-515	352	8	the	the	DET
ejpam-515	352	9	variable	variable	ADJ
ejpam-515	352	10	selection	selection	NOUN
ejpam-515	352	11	problem	problem	NOUN
ejpam-515	352	12	,	,	PUNCT
ejpam-515	352	13	pages	page	NOUN
ejpam-515	352	14	350–358	350–358	NUM
ejpam-515	352	15	.	.	PUNCT
ejpam-515	353	1	chaptman	chaptman	PROPN
ejpam-515	353	2	&	&	CCONJ
ejpam-515	353	3	hall	hall	PROPN
ejpam-515	353	4	/	/	SYM
ejpam-515	353	5	crc	crc	PROPN
ejpam-515	353	6	,	,	PUNCT
ejpam-515	353	7	2002	2002	NUM
ejpam-515	353	8	.	.	PUNCT
ejpam-515	354	1	[	[	X
ejpam-515	354	2	5	5	NUM
ejpam-515	354	3	]	]	PUNCT
ejpam-515	354	4	e	e	X
ejpam-515	354	5	george	george	NOUN
ejpam-515	354	6	and	and	CCONJ
ejpam-515	354	7	r	r	NOUN
ejpam-515	354	8	mccullock	mccullock	NOUN
ejpam-515	354	9	.	.	PUNCT
ejpam-515	355	1	approaches	approach	NOUN
ejpam-515	355	2	for	for	ADP
ejpam-515	355	3	bayesian	bayesian	ADJ
ejpam-515	355	4	variable	variable	ADJ
ejpam-515	355	5	selection	selection	NOUN
ejpam-515	355	6	.	.	PUNCT
ejpam-515	356	1	statistica	statistica	PROPN
ejpam-515	356	2	sinica	sinica	PROPN
ejpam-515	356	3	,	,	PUNCT
ejpam-515	356	4	7:339–373	7:339–373	NOUN
ejpam-515	356	5	,	,	PUNCT
ejpam-515	356	6	1997	1997	NUM
ejpam-515	356	7	.	.	PUNCT
ejpam-515	357	1	[	[	X
ejpam-515	357	2	6	6	NUM
ejpam-515	357	3	]	]	PUNCT
ejpam-515	357	4	n	n	PRON
ejpam-515	357	5	johnson	johnson	PROPN
ejpam-515	357	6	and	and	CCONJ
ejpam-515	357	7	s	s	PROPN
ejpam-515	357	8	kotz	kotz	PROPN
ejpam-515	357	9	.	.	PUNCT
ejpam-515	358	1	discrete	discrete	ADJ
ejpam-515	358	2	distributions	distribution	NOUN
ejpam-515	358	3	.	.	PUNCT
ejpam-515	359	1	houghton	houghton	PROPN
ejpam-515	359	2	mifflin	mifflin	PROPN
ejpam-515	359	3	company	company	PROPN
ejpam-515	359	4	,	,	PUNCT
ejpam-515	359	5	boston	boston	PROPN
ejpam-515	359	6	,	,	PUNCT
ejpam-515	359	7	massachusetts	massachusetts	PROPN
ejpam-515	359	8	,	,	PUNCT
ejpam-515	359	9	1969	1969	NUM
ejpam-515	359	10	.	.	PUNCT
ejpam-515	360	1	[	[	X
ejpam-515	360	2	7	7	X
ejpam-515	360	3	]	]	SYM
ejpam-515	360	4	c	c	NOUN
ejpam-515	360	5	mallows	mallow	NOUN
ejpam-515	360	6	.	.	PUNCT
ejpam-515	361	1	some	some	DET
ejpam-515	361	2	comments	comment	NOUN
ejpam-515	361	3	on	on	ADP
ejpam-515	361	4	cp	cp	PROPN
ejpam-515	361	5	.	.	PROPN
ejpam-515	361	6	technometrics	technometrics	PROPN
ejpam-515	361	7	,	,	PUNCT
ejpam-515	361	8	15:661–675	15:661–675	PROPN
ejpam-515	361	9	,	,	PUNCT
ejpam-515	361	10	1973	1973	NUM
ejpam-515	361	11	.	.	PUNCT
ejpam-515	362	1	[	[	X
ejpam-515	362	2	8	8	X
ejpam-515	362	3	]	]	X
ejpam-515	362	4	p	p	NOUN
ejpam-515	362	5	mccullagh	mccullagh	NOUN
ejpam-515	362	6	and	and	CCONJ
ejpam-515	362	7	j	j	PROPN
ejpam-515	362	8	nelder	nelder	PROPN
ejpam-515	362	9	.	.	PUNCT
ejpam-515	363	1	generalized	generalized	ADJ
ejpam-515	363	2	linear	linear	ADJ
ejpam-515	363	3	models	model	NOUN
ejpam-515	363	4	.	.	PUNCT
ejpam-515	364	1	chapman	chapman	PROPN
ejpam-515	364	2	&	&	CCONJ
ejpam-515	364	3	hall	hall	PROPN
ejpam-515	364	4	,	,	PUNCT
ejpam-515	364	5	london	london	PROPN
ejpam-515	364	6	,	,	PUNCT
ejpam-515	364	7	2	2	NUM
ejpam-515	364	8	edition	edition	NOUN
ejpam-515	364	9	,	,	PUNCT
ejpam-515	364	10	1989	1989	NUM
ejpam-515	364	11	.	.	PUNCT
ejpam-515	365	1	[	[	X
ejpam-515	365	2	9	9	NUM
ejpam-515	365	3	]	]	SYM
ejpam-515	365	4	v	v	NOUN
ejpam-515	365	5	petrov	petrov	PROPN
ejpam-515	365	6	.	.	PUNCT
ejpam-515	366	1	limit	limit	PROPN
ejpam-515	366	2	theorems	theorem	NOUN
ejpam-515	366	3	of	of	ADP
ejpam-515	366	4	probability	probability	NOUN
ejpam-515	366	5	theory	theory	NOUN
ejpam-515	366	6	:	:	PUNCT
ejpam-515	366	7	sequences	sequence	NOUN
ejpam-515	366	8	of	of	ADP
ejpam-515	366	9	independent	independent	ADJ
ejpam-515	366	10	random	random	ADJ
ejpam-515	366	11	variables	variable	NOUN
ejpam-515	366	12	.	.	PUNCT
ejpam-515	367	1	oxford	oxford	PROPN
ejpam-515	367	2	university	university	PROPN
ejpam-515	367	3	press	press	NOUN
ejpam-515	367	4	,	,	PUNCT
ejpam-515	367	5	1995	1995	NUM
ejpam-515	367	6	.	.	PUNCT
ejpam-515	368	1	[	[	X
ejpam-515	368	2	10	10	NUM
ejpam-515	368	3	]	]	X
ejpam-515	368	4	g	g	PROPN
ejpam-515	368	5	qian	qian	PROPN
ejpam-515	368	6	.	.	PUNCT
ejpam-515	369	1	computations	computation	NOUN
ejpam-515	369	2	and	and	CCONJ
ejpam-515	369	3	analysis	analysis	NOUN
ejpam-515	369	4	in	in	ADP
ejpam-515	369	5	robust	robust	ADJ
ejpam-515	369	6	regression	regression	NOUN
ejpam-515	369	7	model	model	NOUN
ejpam-515	369	8	selection	selection	NOUN
ejpam-515	369	9	using	use	VERB
ejpam-515	369	10	stochastic	stochastic	ADJ
ejpam-515	369	11	complexity	complexity	NOUN
ejpam-515	369	12	.	.	PUNCT
ejpam-515	370	1	computational	computational	ADJ
ejpam-515	370	2	statistics	statistic	NOUN
ejpam-515	370	3	,	,	PUNCT
ejpam-515	370	4	14:293–314	14:293–314	PROPN
ejpam-515	370	5	,	,	PUNCT
ejpam-515	370	6	1999	1999	NUM
ejpam-515	370	7	.	.	PUNCT
ejpam-515	371	1	references	reference	NOUN
ejpam-515	371	2	433	433	NUM
ejpam-515	372	1	[	[	X
ejpam-515	372	2	11	11	NUM
ejpam-515	372	3	]	]	PUNCT
ejpam-515	372	4	g	g	PROPN
ejpam-515	372	5	qian	qian	PROPN
ejpam-515	372	6	and	and	CCONJ
ejpam-515	372	7	h	h	PROPN
ejpam-515	372	8	künsch	künsch	NOUN
ejpam-515	372	9	.	.	PUNCT
ejpam-515	373	1	some	some	DET
ejpam-515	373	2	notes	note	NOUN
ejpam-515	373	3	on	on	ADP
ejpam-515	373	4	rissanen	rissanen	PROPN
ejpam-515	373	5	’s	’s	PART
ejpam-515	373	6	stochastic	stochastic	ADJ
ejpam-515	373	7	complexity	complexity	NOUN
ejpam-515	373	8	.	.	PUNCT
ejpam-515	374	1	ieee	ieee	NOUN
ejpam-515	374	2	transactions	transaction	NOUN
ejpam-515	374	3	on	on	ADP
ejpam-515	374	4	information	information	NOUN
ejpam-515	374	5	theory	theory	NOUN
ejpam-515	374	6	,	,	PUNCT
ejpam-515	374	7	44:782–786	44:782–786	PROPN
ejpam-515	374	8	,	,	PUNCT
ejpam-515	374	9	1998	1998	NUM
ejpam-515	374	10	.	.	PUNCT
ejpam-515	375	1	[	[	X
ejpam-515	375	2	12	12	NUM
ejpam-515	375	3	]	]	X
ejpam-515	375	4	g	g	PROPN
ejpam-515	375	5	qian	qian	PROPN
ejpam-515	375	6	and	and	CCONJ
ejpam-515	375	7	y	y	PROPN
ejpam-515	375	8	wu	wu	PROPN
ejpam-515	375	9	.	.	PUNCT
ejpam-515	376	1	strong	strong	ADJ
ejpam-515	376	2	limit	limit	NOUN
ejpam-515	376	3	theorems	theorem	NOUN
ejpam-515	376	4	on	on	ADP
ejpam-515	376	5	model	model	NOUN
ejpam-515	376	6	selection	selection	NOUN
ejpam-515	376	7	in	in	ADP
ejpam-515	376	8	generalized	generalized	ADJ
ejpam-515	376	9	linear	linear	ADJ
ejpam-515	376	10	regression	regression	NOUN
ejpam-515	376	11	with	with	ADP
ejpam-515	376	12	binomial	binomial	ADJ
ejpam-515	376	13	responses	response	NOUN
ejpam-515	376	14	.	.	PUNCT
ejpam-515	377	1	statisticsa	statisticsa	PROPN
ejpam-515	377	2	sinica	sinica	PROPN
ejpam-515	377	3	,	,	PUNCT
ejpam-515	377	4	16:1335–1365	16:1335–1365	NUM
ejpam-515	377	5	,	,	PUNCT
ejpam-515	377	6	2006	2006	NUM
ejpam-515	377	7	.	.	PUNCT
ejpam-515	378	1	[	[	X
ejpam-515	378	2	13	13	NUM
ejpam-515	378	3	]	]	X
ejpam-515	378	4	g	g	PROPN
ejpam-515	378	5	qian	qian	PROPN
ejpam-515	378	6	and	and	CCONJ
ejpam-515	378	7	x	x	PROPN
ejpam-515	378	8	zhao	zhao	PROPN
ejpam-515	378	9	.	.	PUNCT
ejpam-515	379	1	on	on	ADP
ejpam-515	379	2	time	time	PROPN
ejpam-515	379	3	series	series	PROPN
ejpam-515	379	4	model	model	PROPN
ejpam-515	379	5	selection	selection	NOUN
ejpam-515	379	6	involving	involve	VERB
ejpam-515	379	7	many	many	ADJ
ejpam-515	379	8	candidate	candidate	NOUN
ejpam-515	379	9	arma	arma	NOUN
ejpam-515	379	10	models	model	NOUN
ejpam-515	379	11	.	.	PUNCT
ejpam-515	380	1	computational	computational	ADJ
ejpam-515	380	2	statistics	statistic	NOUN
ejpam-515	380	3	&	&	CCONJ
ejpam-515	380	4	data	datum	NOUN
ejpam-515	380	5	analysis	analysis	NOUN
ejpam-515	380	6	,	,	PUNCT
ejpam-515	380	7	51:6180–6196	51:6180–6196	NUM
ejpam-515	380	8	,	,	PUNCT
ejpam-515	380	9	2007	2007	NUM
ejpam-515	380	10	.	.	PUNCT
ejpam-515	381	1	[	[	X
ejpam-515	381	2	14	14	NUM
ejpam-515	381	3	]	]	X
ejpam-515	381	4	c	c	PROPN
ejpam-515	381	5	rao	rao	PROPN
ejpam-515	381	6	and	and	CCONJ
ejpam-515	381	7	y	y	PROPN
ejpam-515	381	8	wo	will	AUX
ejpam-515	381	9	.	.	PUNCT
ejpam-515	382	1	model	model	PROPN
ejpam-515	382	2	selection	selection	NOUN
ejpam-515	382	3	,	,	PUNCT
ejpam-515	382	4	volume	volume	NOUN
ejpam-515	382	5	38	38	NUM
ejpam-515	382	6	of	of	ADP
ejpam-515	382	7	ims	ims	PROPN
ejpam-515	382	8	lecture	lecture	NOUN
ejpam-515	382	9	notes	note	NOUN
ejpam-515	382	10	monograph	monograph	PROPN
ejpam-515	382	11	series	series	PROPN
ejpam-515	382	12	,	,	PUNCT
ejpam-515	382	13	chapter	chapter	NOUN
ejpam-515	382	14	on	on	ADP
ejpam-515	382	15	model	model	NOUN
ejpam-515	382	16	selection	selection	NOUN
ejpam-515	382	17	(	(	PUNCT
ejpam-515	382	18	with	with	ADP
ejpam-515	382	19	discussion	discussion	NOUN
ejpam-515	382	20	)	)	PUNCT
ejpam-515	383	1	,	,	PUNCT
ejpam-515	383	2	pages	page	NOUN
ejpam-515	383	3	1–64	1–64	PROPN
ejpam-515	383	4	.	.	PUNCT
ejpam-515	383	5	institute	institute	PROPN
ejpam-515	383	6	of	of	ADP
ejpam-515	383	7	mathematical	mathematical	ADJ
ejpam-515	383	8	statistics	statistic	NOUN
ejpam-515	383	9	,	,	PUNCT
ejpam-515	383	10	beachwood	beachwood	PROPN
ejpam-515	383	11	,	,	PUNCT
ejpam-515	383	12	ohio	ohio	PROPN
ejpam-515	383	13	,	,	PUNCT
ejpam-515	383	14	2001	2001	NUM
ejpam-515	383	15	.	.	PUNCT
ejpam-515	384	1	[	[	X
ejpam-515	384	2	15	15	NUM
ejpam-515	384	3	]	]	X
ejpam-515	384	4	c	c	NOUN
ejpam-515	384	5	rao	rao	NOUN
ejpam-515	384	6	and	and	CCONJ
ejpam-515	384	7	l	l	NOUN
ejpam-515	384	8	zhao	zhao	PROPN
ejpam-515	384	9	.	.	PUNCT
ejpam-515	385	1	linear	linear	ADJ
ejpam-515	385	2	representation	representation	NOUN
ejpam-515	385	3	of	of	ADP
ejpam-515	385	4	m	m	NOUN
ejpam-515	385	5	-	-	NOUN
ejpam-515	385	6	estimates	estimate	NOUN
ejpam-515	385	7	in	in	ADP
ejpam-515	385	8	linear	linear	PROPN
ejpam-515	385	9	models	model	NOUN
ejpam-515	385	10	.	.	PUNCT
ejpam-515	386	1	the	the	DET
ejpam-515	386	2	canadian	canadian	ADJ
ejpam-515	386	3	journal	journal	PROPN
ejpam-515	386	4	of	of	ADP
ejpam-515	386	5	statistics	statistic	NOUN
ejpam-515	386	6	,	,	PUNCT
ejpam-515	386	7	20:359–368	20:359–368	PROPN
ejpam-515	386	8	,	,	PUNCT
ejpam-515	386	9	1992	1992	NUM
ejpam-515	386	10	.	.	PUNCT
ejpam-515	387	1	[	[	X
ejpam-515	387	2	16	16	NUM
ejpam-515	387	3	]	]	X
ejpam-515	387	4	j	j	PROPN
ejpam-515	387	5	rissanen	rissanen	PROPN
ejpam-515	387	6	.	.	PUNCT
ejpam-515	388	1	stochastic	stochastic	ADJ
ejpam-515	388	2	complexity	complexity	NOUN
ejpam-515	388	3	in	in	ADP
ejpam-515	388	4	statistical	statistical	ADJ
ejpam-515	388	5	inquiry	inquiry	NOUN
ejpam-515	388	6	.	.	PUNCT
ejpam-515	389	1	world	world	NOUN
ejpam-515	389	2	scientific	scientific	PROPN
ejpam-515	389	3	publishing	publishing	PROPN
ejpam-515	389	4	co.	co.	PROPN
ejpam-515	389	5	pte	pte	PROPN
ejpam-515	389	6	.	.	PROPN
ejpam-515	389	7	ltd	ltd	PROPN
ejpam-515	389	8	.	.	PROPN
ejpam-515	389	9	,	,	PUNCT
ejpam-515	389	10	singapore	singapore	PROPN
ejpam-515	389	11	,	,	PUNCT
ejpam-515	389	12	1989	1989	NUM
ejpam-515	389	13	.	.	PUNCT
ejpam-515	390	1	[	[	X
ejpam-515	390	2	17	17	NUM
ejpam-515	390	3	]	]	X
ejpam-515	390	4	j	j	PROPN
ejpam-515	390	5	rissanen	rissanen	PROPN
ejpam-515	390	6	.	.	PUNCT
ejpam-515	391	1	fisher	fisher	PROPN
ejpam-515	391	2	information	information	NOUN
ejpam-515	391	3	and	and	CCONJ
ejpam-515	391	4	stochastic	stochastic	ADJ
ejpam-515	391	5	complexity	complexity	NOUN
ejpam-515	391	6	.	.	PUNCT
ejpam-515	392	1	ieee	ieee	NOUN
ejpam-515	392	2	transactions	transaction	NOUN
ejpam-515	392	3	information	information	NOUN
ejpam-515	392	4	theory	theory	NOUN
ejpam-515	392	5	,	,	PUNCT
ejpam-515	392	6	42:40–47	42:40–47	NOUN
ejpam-515	392	7	,	,	PUNCT
ejpam-515	392	8	1996	1996	NUM
ejpam-515	392	9	.	.	PUNCT
ejpam-515	393	1	[	[	X
ejpam-515	393	2	18	18	NUM
ejpam-515	393	3	]	]	X
ejpam-515	393	4	g	g	PROPN
ejpam-515	393	5	schwarz	schwarz	PROPN
ejpam-515	393	6	.	.	PUNCT
ejpam-515	393	7	estimating	estimate	VERB
ejpam-515	393	8	the	the	DET
ejpam-515	393	9	dimension	dimension	NOUN
ejpam-515	393	10	of	of	ADP
ejpam-515	393	11	a	a	DET
ejpam-515	393	12	model	model	NOUN
ejpam-515	393	13	.	.	PUNCT
ejpam-515	394	1	annals	annal	NOUN
ejpam-515	394	2	of	of	ADP
ejpam-515	394	3	statistics	statistic	NOUN
ejpam-515	394	4	,	,	PUNCT
ejpam-515	394	5	6:461–464	6:461–464	PROPN
ejpam-515	394	6	,	,	PUNCT
ejpam-515	394	7	1978	1978	NUM
ejpam-515	394	8	.	.	PUNCT
ejpam-515	395	1	[	[	X
ejpam-515	395	2	19	19	NUM
ejpam-515	395	3	]	]	X
ejpam-515	395	4	r	r	NOUN
ejpam-515	395	5	tibshirani	tibshirani	NOUN
ejpam-515	395	6	.	.	PUNCT
ejpam-515	396	1	regression	regression	NOUN
ejpam-515	396	2	shrinkage	shrinkage	NOUN
ejpam-515	396	3	and	and	CCONJ
ejpam-515	396	4	selection	selection	NOUN
ejpam-515	396	5	via	via	ADP
ejpam-515	396	6	the	the	DET
ejpam-515	396	7	lasso	lasso	NOUN
ejpam-515	396	8	.	.	PUNCT
ejpam-515	397	1	journal	journal	NOUN
ejpam-515	397	2	of	of	ADP
ejpam-515	397	3	the	the	DET
ejpam-515	397	4	royal	royal	ADJ
ejpam-515	397	5	statistical	statistical	ADJ
ejpam-515	397	6	society	society	NOUN
ejpam-515	397	7	.	.	PUNCT
ejpam-515	398	1	[	[	X
ejpam-515	398	2	20	20	NUM
ejpam-515	398	3	]	]	X
ejpam-515	398	4	y	y	PROPN
ejpam-515	398	5	wu	wu	PROPN
ejpam-515	398	6	and	and	CCONJ
ejpam-515	398	7	m	m	PROPN
ejpam-515	398	8	zen	zen	PROPN
ejpam-515	398	9	.	.	PUNCT
ejpam-515	399	1	a	a	DET
ejpam-515	399	2	strongly	strongly	ADV
ejpam-515	399	3	consistent	consistent	ADJ
ejpam-515	399	4	linear	linear	ADJ
ejpam-515	399	5	model	model	NOUN
ejpam-515	399	6	selection	selection	NOUN
ejpam-515	399	7	procedure	procedure	NOUN
ejpam-515	399	8	based	base	VERB
ejpam-515	399	9	on	on	ADP
ejpam-515	399	10	m	m	NOUN
ejpam-515	399	11	-	-	NOUN
ejpam-515	399	12	estimation	estimation	NOUN
ejpam-515	399	13	.	.	PUNCT
ejpam-515	400	1	probability	probability	NOUN
ejpam-515	400	2	theory	theory	NOUN
ejpam-515	400	3	and	and	CCONJ
ejpam-515	400	4	related	related	ADJ
ejpam-515	400	5	fields	field	NOUN
ejpam-515	400	6	,	,	PUNCT
ejpam-515	400	7	113:599–625	113:599–625	NUM
ejpam-515	400	8	,	,	PUNCT
ejpam-515	400	9	1999	1999	NUM
ejpam-515	400	10	.	.	PUNCT
ejpam-515	401	1	appendix	appendix	ADJ
ejpam-515	401	2	proof	proof	NOUN
ejpam-515	401	3	.	.	PUNCT
ejpam-515	402	1	(	(	PUNCT
ejpam-515	402	2	lemma	lemma	PROPN
ejpam-515	402	3	1	1	NUM
ejpam-515	402	4	)	)	PUNCT
ejpam-515	402	5	since	since	SCONJ
ejpam-515	402	6	k(t	k(t	PROPN
ejpam-515	402	7	,	,	PUNCT
ejpam-515	402	8	s	s	PART
ejpam-515	402	9	)	)	PUNCT
ejpam-515	402	10	=	=	SYM
ejpam-515	402	11	et	et	NOUN
ejpam-515	402	12	−	−	NOUN
ejpam-515	402	13	es	es	INTJ
ejpam-515	402	14	−	−	NOUN
ejpam-515	402	15	es(t	es(t	PUNCT
ejpam-515	402	16	−	−	PROPN
ejpam-515	402	17	s	s	PART
ejpam-515	402	18	)	)	PUNCT
ejpam-515	402	19	,	,	PUNCT
ejpam-515	402	20	it	it	PRON
ejpam-515	402	21	is	be	AUX
ejpam-515	402	22	easy	easy	ADJ
ejpam-515	402	23	to	to	PART
ejpam-515	402	24	see	see	VERB
ejpam-515	402	25	that	that	SCONJ
ejpam-515	402	26	k(s	k(s	PROPN
ejpam-515	402	27	,	,	PUNCT
ejpam-515	402	28	s	s	PART
ejpam-515	402	29	)	)	PUNCT
ejpam-515	402	30	=	=	SYM
ejpam-515	402	31	0	0	NUM
ejpam-515	402	32	,	,	PUNCT
ejpam-515	402	33	k	k	PROPN
ejpam-515	402	34	′t(t	′t(t	PROPN
ejpam-515	402	35	,	,	PUNCT
ejpam-515	402	36	s	s	PART
ejpam-515	402	37	)	)	PUNCT
ejpam-515	402	38	=	=	SYM
ejpam-515	402	39	et	et	NOUN
ejpam-515	402	40	−	−	NOUN
ejpam-515	402	41	es	es	NOUN
ejpam-515	402	42	,	,	PUNCT
ejpam-515	402	43	k	k	PROPN
ejpam-515	402	44	′t(s	′t(s	PROPN
ejpam-515	402	45	,	,	PUNCT
ejpam-515	402	46	s	s	NOUN
ejpam-515	402	47	)	)	PUNCT
ejpam-515	402	48	=	=	SYM
ejpam-515	402	49	0	0	NUM
ejpam-515	402	50	and	and	CCONJ
ejpam-515	402	51	k	k	PROPN
ejpam-515	402	52	′′t	′′t	X
ejpam-515	402	53	(	(	PUNCT
ejpam-515	402	54	t	t	PROPN
ejpam-515	402	55	,	,	PUNCT
ejpam-515	402	56	s	s	PART
ejpam-515	402	57	)	)	PUNCT
ejpam-515	402	58	=	=	SYM
ejpam-515	402	59	et	et	X
ejpam-515	402	60	>	>	X
ejpam-515	402	61	0	0	X
ejpam-515	402	62	.	.	PUNCT
ejpam-515	403	1	hence	hence	ADV
ejpam-515	403	2	,	,	PUNCT
ejpam-515	403	3	k(t	k(t	PROPN
ejpam-515	403	4	,	,	PUNCT
ejpam-515	403	5	s	s	PART
ejpam-515	403	6	)	)	PUNCT
ejpam-515	403	7	is	be	AUX
ejpam-515	403	8	strictly	strictly	ADV
ejpam-515	403	9	convex	convex	ADJ
ejpam-515	403	10	with	with	ADP
ejpam-515	403	11	respect	respect	NOUN
ejpam-515	403	12	to	to	ADP
ejpam-515	403	13	t	t	PROPN
ejpam-515	403	14	,	,	PUNCT
ejpam-515	403	15	and	and	CCONJ
ejpam-515	403	16	k(t	k(t	PROPN
ejpam-515	403	17	,	,	PUNCT
ejpam-515	403	18	s	s	X
ejpam-515	403	19	)	)	PUNCT
ejpam-515	403	20	≥	≥	NOUN
ejpam-515	403	21	0	0	NUM
ejpam-515	403	22	with	with	ADP
ejpam-515	403	23	k(t	k(t	PROPN
ejpam-515	403	24	,	,	PUNCT
ejpam-515	403	25	s	s	PART
ejpam-515	403	26	)	)	PUNCT
ejpam-515	403	27	=	=	SYM
ejpam-515	403	28	0	0	PUNCT
ejpam-515	404	1	only	only	ADV
ejpam-515	404	2	if	if	SCONJ
ejpam-515	404	3	t	t	PROPN
ejpam-515	404	4	=	=	PUNCT
ejpam-515	404	5	s.	s.	PROPN
ejpam-515	404	6	now	now	ADV
ejpam-515	404	7	let	let	VERB
ejpam-515	404	8	f(t	f(t	PROPN
ejpam-515	404	9	,	,	PUNCT
ejpam-515	404	10	s	s	NOUN
ejpam-515	404	11	)	)	PUNCT
ejpam-515	404	12	=	=	SYM
ejpam-515	405	1	k(t	k(t	PROPN
ejpam-515	405	2	,	,	PUNCT
ejpam-515	405	3	s)−	s)−	PROPN
ejpam-515	405	4	1	1	NUM
ejpam-515	405	5	2	2	NUM
ejpam-515	405	6	es−2∆(t	es−2∆(t	NOUN
ejpam-515	405	7	−	−	PROPN
ejpam-515	405	8	s)2	s)2	PROPN
ejpam-515	405	9	g(t	g(t	PROPN
ejpam-515	405	10	,	,	PUNCT
ejpam-515	405	11	s	s	PART
ejpam-515	405	12	)	)	PUNCT
ejpam-515	405	13	=	=	PUNCT
ejpam-515	405	14	k(t	k(t	PROPN
ejpam-515	405	15	,	,	PUNCT
ejpam-515	405	16	s)−	s)−	PROPN
ejpam-515	405	17	1	1	NUM
ejpam-515	405	18	2	2	NUM
ejpam-515	405	19	es+2∆(t	es+2∆(t	NOUN
ejpam-515	405	20	−	−	NOUN
ejpam-515	405	21	s)2	s)2	NOUN
ejpam-515	405	22	references	reference	NOUN
ejpam-515	405	23	434	434	NUM
ejpam-515	405	24	for	for	ADP
ejpam-515	405	25	any	any	DET
ejpam-515	405	26	real	real	ADJ
ejpam-515	405	27	numbers	number	NOUN
ejpam-515	405	28	s	s	PROPN
ejpam-515	405	29	,	,	PUNCT
ejpam-515	405	30	t	t	PROPN
ejpam-515	405	31	and	and	CCONJ
ejpam-515	405	32	∆	∆	PROPN
ejpam-515	405	33	>	>	X
ejpam-515	405	34	0	0	PROPN
ejpam-515	405	35	,	,	PUNCT
ejpam-515	405	36	suppose	suppose	VERB
ejpam-515	405	37	|t	|t	PROPN
ejpam-515	405	38	−	−	PROPN
ejpam-515	405	39	s|	s|	VERB
ejpam-515	405	40	≤∆.	≤∆.	NOUN
ejpam-515	405	41	then	then	ADV
ejpam-515	405	42	we	we	PRON
ejpam-515	405	43	have	have	VERB
ejpam-515	405	44	s−	s−	NOUN
ejpam-515	405	45	2∆	2∆	NUM
ejpam-515	405	46	<	<	SYM
ejpam-515	405	47	s−∆	s−∆	PROPN
ejpam-515	405	48	≤	≤	X
ejpam-515	405	49	t	t	NOUN
ejpam-515	405	50	≤	≤	NUM
ejpam-515	405	51	s+∆	s+∆	X
ejpam-515	405	52	<	<	X
ejpam-515	405	53	s+	s+	NUM
ejpam-515	405	54	2∆.	2∆.	NUM
ejpam-515	405	55	it	it	PRON
ejpam-515	405	56	is	be	AUX
ejpam-515	405	57	clear	clear	ADJ
ejpam-515	405	58	that	that	SCONJ
ejpam-515	405	59	f(s	f(s	ADV
ejpam-515	405	60	,	,	PUNCT
ejpam-515	405	61	s	s	X
ejpam-515	405	62	)	)	PUNCT
ejpam-515	405	63	=	=	SYM
ejpam-515	405	64	0	0	NUM
ejpam-515	405	65	,	,	PUNCT
ejpam-515	405	66	f	f	PROPN
ejpam-515	405	67	′t(t	′t(t	PROPN
ejpam-515	405	68	,	,	PUNCT
ejpam-515	405	69	s	s	PART
ejpam-515	405	70	)	)	PUNCT
ejpam-515	405	71	=	=	SYM
ejpam-515	405	72	et	et	NOUN
ejpam-515	405	73	−	−	NOUN
ejpam-515	405	74	es	es	X
ejpam-515	405	75	−	−	PROPN
ejpam-515	405	76	es−2∆(t	es−2∆(t	NOUN
ejpam-515	405	77	−	−	PROPN
ejpam-515	405	78	s	s	NOUN
ejpam-515	405	79	)	)	PUNCT
ejpam-515	405	80	,	,	PUNCT
ejpam-515	405	81	f	f	PROPN
ejpam-515	405	82	′t(s	′t(s	PROPN
ejpam-515	405	83	,	,	PUNCT
ejpam-515	405	84	s	s	NOUN
ejpam-515	405	85	)	)	PUNCT
ejpam-515	405	86	=	=	SYM
ejpam-515	405	87	0	0	NUM
ejpam-515	405	88	and	and	CCONJ
ejpam-515	405	89	f	f	PROPN
ejpam-515	405	90	′′t	′′t	X
ejpam-515	405	91	(	(	PUNCT
ejpam-515	405	92	t	t	PROPN
ejpam-515	405	93	,	,	PUNCT
ejpam-515	405	94	s	s	PART
ejpam-515	405	95	)	)	PUNCT
ejpam-515	406	1	=	=	SYM
ejpam-515	406	2	et	et	PROPN
ejpam-515	406	3	−	−	PROPN
ejpam-515	406	4	es−2∆	es−2∆	PROPN
ejpam-515	406	5	>	>	X
ejpam-515	406	6	0	0	PUNCT
ejpam-515	407	1	therefore	therefore	ADV
ejpam-515	407	2	,	,	PUNCT
ejpam-515	407	3	f(t	f(t	PROPN
ejpam-515	407	4	,	,	PUNCT
ejpam-515	407	5	s	s	X
ejpam-515	407	6	)	)	PUNCT
ejpam-515	407	7	≥	≥	NOUN
ejpam-515	407	8	0	0	NUM
ejpam-515	407	9	with	with	ADP
ejpam-515	407	10	f(t	f(t	PROPN
ejpam-515	407	11	,	,	PUNCT
ejpam-515	407	12	s	s	NOUN
ejpam-515	407	13	)	)	PUNCT
ejpam-515	407	14	=	=	SYM
ejpam-515	407	15	0	0	PUNCT
ejpam-515	408	1	only	only	ADV
ejpam-515	408	2	if	if	SCONJ
ejpam-515	408	3	t	t	PROPN
ejpam-515	408	4	=	=	SYM
ejpam-515	408	5	s	s	PROPN
ejpam-515	408	6	,	,	PUNCT
ejpam-515	408	7	namely	namely	ADV
ejpam-515	408	8	k(t	k(t	PROPN
ejpam-515	408	9	,	,	PUNCT
ejpam-515	408	10	s	s	X
ejpam-515	408	11	)	)	PUNCT
ejpam-515	408	12	≥	≥	NOUN
ejpam-515	408	13	1	1	NUM
ejpam-515	408	14	2	2	NUM
ejpam-515	408	15	es−2∆(t	es−2∆(t	NOUN
ejpam-515	408	16	−	−	NOUN
ejpam-515	408	17	s)2	s)2	NOUN
ejpam-515	408	18	.	.	PUNCT
ejpam-515	409	1	similarly	similarly	ADV
ejpam-515	409	2	,	,	PUNCT
ejpam-515	409	3	we	we	PRON
ejpam-515	409	4	can	can	AUX
ejpam-515	409	5	show	show	VERB
ejpam-515	409	6	that	that	SCONJ
ejpam-515	409	7	g(t	g(t	PROPN
ejpam-515	409	8	,	,	PUNCT
ejpam-515	409	9	s	s	NOUN
ejpam-515	409	10	)	)	PUNCT
ejpam-515	409	11	≤	≤	NOUN
ejpam-515	409	12	0	0	NUM
ejpam-515	409	13	with	with	ADP
ejpam-515	409	14	g(t	g(t	PROPN
ejpam-515	409	15	,	,	PUNCT
ejpam-515	409	16	s	s	PART
ejpam-515	409	17	)	)	PUNCT
ejpam-515	409	18	=	=	SYM
ejpam-515	409	19	0	0	PUNCT
ejpam-515	410	1	only	only	ADV
ejpam-515	410	2	if	if	SCONJ
ejpam-515	410	3	t	t	PROPN
ejpam-515	410	4	=	=	SYM
ejpam-515	410	5	s	s	PROPN
ejpam-515	410	6	,	,	PUNCT
ejpam-515	410	7	namely	namely	ADV
ejpam-515	410	8	k(t	k(t	PROPN
ejpam-515	410	9	,	,	PUNCT
ejpam-515	410	10	s	s	X
ejpam-515	410	11	)	)	PUNCT
ejpam-515	410	12	≤	≤	NUM
ejpam-515	410	13	1	1	NUM
ejpam-515	410	14	2	2	NUM
ejpam-515	410	15	es+2∆(t	es+2∆(t	NOUN
ejpam-515	410	16	−	−	NOUN
ejpam-515	410	17	s)2	s)2	NOUN
ejpam-515	410	18	.	.	PUNCT
ejpam-515	411	1	this	this	PRON
ejpam-515	411	2	concludes	conclude	VERB
ejpam-515	411	3	the	the	DET
ejpam-515	411	4	proof	proof	NOUN
ejpam-515	411	5	.	.	PUNCT