Int. J. Anal. Appl. (2022), 20:42 Some New Types of Convergence Definitions for Random Variable Sequences Saadettin Aydın∗ Gülhane Medical Faculty, University of Health Sciences, Ankara, Turkey ∗Corresponding author: saadettin.aydin@sbu.edu.tr Abstract. In this paper, we introduce the concepts of invariant convergence in probability, statisti- cally invariant convergence in probability, invariant convergence almost surely, invariant convergence in distribution and invariant convergence in Lp- norm for sequences of random variables. Also, we investigate some inclusion relations. 1. Introduction In probability theory, we know several different notions of convergence of random variables. The convergence of sequences of random variables to some limit random variable is an essential concept in probability theory and its applications to statistics. The natural density of a set K of positive integers is defined by δ(K) := lim n→∞ 1 n |{k ≤ n : k ∈ K}|, where |{k ≤ n : k ∈ K}| denotes the number of elements of K not exceeding n. The concept of statistical convergence of number sequences was introduced by Fast [5]. In [25] Schoenberg established some basic features of statistical convergence and studied the notion as a summability method. A continuous linear functional ϕ on `∞, the space of real bounded sequences, is said to be a invariant limit if (a) ϕ(x)≥ 0 when the sequence x =(xn) has xn ≥ 0 for all n, Received: Jul. 12, 2022. 2010 Mathematics Subject Classification. 40A05, 40D25, 60B10. Key words and phrases. random variable; statistical convergence; invariant convergence; strongly invariant conver- gence; convergence in probability; almost surely convergence. https://doi.org/10.28924/2291-8639-20-2022-42 ISSN: 2291-8639 © 2022 the author(s). https://doi.org/10.28924/2291-8639-20-2022-42 2 Int. J. Anal. Appl. (2022), 20:42 (b) ϕ(e)=1 where e =(1,1,1, ...) and (c) ϕ(xσ(m))= ϕ(x) for all x ∈ l∞, where σ : N+ → N+. A sequence x ∈ l∞ is said to be invariant convergent to ` if all of its invariant limits are equal to `. The mappings σ are one-to-one and such that σk(m) 6= m for all k,m ∈ N+ and σk(m) = σ(σk−1(m). Thus ϕ extends the limit functional on the space of convergent sequences, in the sense that ϕ(x) = limx for all convergent sequences x. In the case σ is the translation mapping σ(m) = m+1, the σ-mean is often called a Banach limit and Vσ, the set of bounded sequences all of whose invariant means are equal, is the set of almost convergent sequences. It can be shown that Vσ = {(xk)∈ `∞ : lim n→∞ 1 n+1 n∑ k=0 xσk(m) = ` uniformly in m}. In this case, we write σ− limxn = `. The set of all strongly invariant convergent sequences is denoted by [Vσ] = {(xk)∈ `∞ : lim n→∞ 1 n+1 n∑ k=0 |xσk(m) − `|=0 uniformly in m}. In [18], the concept of σ-statistically convergent sequence was introduced as follows: A sequence x =(xk) is said to be σ-statistically convergent to ` if for every� > 0, lim n→∞ 1 n |{k ≤ n : |xσk(m) − `| ≥ �}|=0 uniformly in m. Throughout the paper, {Xn} will denote a sequence of random variables where each Xn is defined on the same sample space W with respect to a given class of events S and a given probability function P :S →R. The concept of statistical convergence in probability of a sequence of random variables {Xn} was introduced in [8] as follows: A sequence of random variables {Xn} is said to be statistically convergent in probability to a random variable X if for any ε,δ > 0 lim n→∞ 1 n |k ≤ n : P(|Xk −X| ≥ ε)≥ δ|=0. The reader can also see [9–11] for related works. A sequence of random variables {Xn} is said to be strongly Cesàro summable in probability to a random variable X if for each ε > 0, lim n→∞ 1 n n∑ k=1 P(|Xk −X| ≥ ε)=0. Let {Xn} be a sequence of random variables. We say {Xn} is bounded in probability, if for every ε > 0, there exists Mε > 0 such that P(|Xn| > ε) < Mε for all n. Int. J. Anal. Appl. (2022), 20:42 3 For more information on the concepts covered in this article, see articles [1,2,8–12,27] on random variables, articles [3–7], [13]- [26] on invariant convergence and statistical convergence. 2. Some new types of invariant convergence Let denote the set of all σ mapping by M. Then⋂ {Vσ : σ ∈ M} is equal to the set of all convergent sequences (see [21]). Therefore, the definitions and theorems we will give in this section are meaningful. The definitions we will give are more general, and the kinds of convergence previously known for random variable sequences can be derived from the definitions we give. 2.1. Invariant convergence in probability. Definition 2.1. The sequence {Xn} is said to be invariant convergent in probability to a random variable X if for any ε > 0 lim n→∞ 1 n n∑ k=1 P(|Xσk(m) −X| ≥ ε)=0 uniformly in m or lim n→∞ 1 n n∑ k=1 P(|Xσk(m) −X| < ε)=1 uniformly in m. In this case, we write σ − limP(|Xn −X| ≥ ε)= 0 or σ − limP(|Xn −X| < ε)= 1. The class of all sequences of random variables which are invariant convergent in probability will be denoted by V Pσ . ⋂{ V Pσ : σ ∈ M } is equal to the set of all random variable sequences which are convergent in probability. 2.2. Statistically invariant convergence in probability. Definition 2.2. The sequence {Xn} is said to be statistically invariant convergent in probability to a random variable X if for any ε,δ > 0 lim n→∞ 1 n |k ≤ n : P(|Xσk(m) −X| ≥ ε)≥ δ|=0 uniformly in m. In this case, we write Xn → X(SPσ ). The class of all sequences of random variables which are statistically invariant convergent in probability will be denoted by SPσ . ⋂ {SPσ : σ ∈ M} is equal to the set of all random variable sequences which are statistically convergent in probability. 4 Int. J. Anal. Appl. (2022), 20:42 Theorem 2.1. If a sequence of random variables {Xn} is strongly invariant convergent in probability to X, then {Xn} is statistically invariant convergent in probability to X. Proof. Suppose that{Xn}is strongly invariant convergent in probability to X . For an arbitrary ε > 0,δ, we get 1 n n∑ k=1 P(|Xσk(m) −X| ≥ ε) = 1 n   n∑ k=1 P(|X σk(m) −X|≥ε)≥δ P(|Xσk(m) −X| ≥ ε)+ n∑ k=1 P(|X σk(m) −X|≥ε)<δ P(|Xσk(m) −X| ≥ ε)   ≥ 1 n n∑ k=1 P(|X σk(m) −X|≥ε)≥δ P(|Xσk(m) −X| ≥ ε) ≥ δ n |{k ≤ n : P(|Xσk(m) −X| ≥ ε)≥ δ}| for each m. Hence, we have lim n→∞ 1 n |{k ≤ n : P(|Xσk(m) −X| ≥ ε)≥ δ}|=0 uniformly in m, that is, {Xn} is statistically invariant convergent in probability to X. � Theorem 2.2. If a sequence of random variables {Xn} is statistically invariant convergent in probability to X and bounded in probability, then {Xn} is strongly invariant convergent in probability to X. Proof. Suppose that {Xn} is statistically invariant convergent in probability to X and {Xn} is bounded, say P(|Xσk(m) −X| > ε) < Mε for all m and k. Given ε,δ > 0, we get 1 n n∑ k=1 P(|Xσk(m) −X| ≥ ε) = 1 n n∑ k=1 P(|X σk(m) −X|≥ε)≥δ P(|Xσk(m) −X| ≥ ε)+ 1 n n∑ k=1 P(|X σk(m) −X|≥ε)<δ P(|Xσk(m) −X| ≥ ε) ≤ ε n n∑ k=1 P(|X σk(m) −X|≥ε)≥δ 1+ δ n n∑ k=1 P(|X σk(m) −X|≥ε)<δ 1 ≤ Mε n |{k ≤ n : P(|Xσk(m) −X| ≥ ε)}|+ δ n |{k ≤ n : P(|Xσk(m) −X| ≥ ε)}| for each m, hence we have lim n→∞ 1 n n∑ k=1 P(|Xσk(m) −X| ≥ ε)=0 uniformly in m. � Int. J. Anal. Appl. (2022), 20:42 5 Theorem 2.3. Suppose that Xn is invariant convergent in probability to a random variable X and that f is a continuous function. Then {f (Xn)} is invariant convergent in probability to f (X). Proof. If f is continuous, given ε > 0 there exists δ > 0 such that |f (Xn)−f (X)| < ε for |Xn−X| < δ. Since {Xn} is invariant convergent in probability to the random variable X, then lim n→∞ 1 n n∑ k=1 P(|Xσk(m) −X| < δ)=1 uniformly in m. Thus, lim n→∞ 1 n n∑ k=1 P(|f (Xσk(m))− f (X)| < ε)=1 uniformly in m. � 2.3. Invariant convergence almost surely. We say an event happens “surely” if it always happens. We say an event happens “almost surely” if it happens with probability 1. Almost sure convergence is one of the main modes of stochastic convergence. It may be viewed as a notion of convergence for random variables that’s analogous to, but not the same as, the notion of pointwise convergence for real functions. Now we will introduce invariant convergent almost surely random variable sequence. Definition 2.3. We say that a sequence of random variables {Xn} is invariant convergent almost surely to X if P( lim n→∞ 1 n n∑ k=1 Xσk(m) = X uniformly in m)=1 that is, P(σ − limXn = X)=1. We will denote the set of all sequences of random variables {Xn} which are invariant convergent almost surely by ASσ. ⋂ {ASσ : σ ∈ M} is equal to the set of all random variable sequences which are convergent almost surely. Theorem 2.4. If a sequence of random variables {Xn} is invariant convergent almost surely to a random variable X, then {Xn} is invariant convergent in probability to X. Proof. Let {Xn} be invariant convergent almost surely to a random variable X and ε > 0. Then for every s ≥ n, we can write P(|Xn −X| < ε)≥ P(|Xs −X| < ε). (2.1) By considering the original sample space, we can write {t : |Xs(t)−X(t)| > ε, ∀s ≥ n}⊂{t : |Xn(t)−X(t)| > ε}. 6 Int. J. Anal. Appl. (2022), 20:42 Since {Xn} is invariant convergent almost surely to a random variable X, for every s ≥ n, we have σ − limP(|Xs −X| < ε)=1. If we take σ − limits of in (2.1), ∀s ≥ n, we get σ − limP(|Xn −X| < ε)≥ σ − limP(|Xs −X| < ε)=1. Thus, σ − limP(|Xn −X| < ε)=1. � 2.4. Invariant convergence in distribution. Definition 2.4. We say that a sequence of random variables {Xn} is invariant convergent in distribution to a random variable X if lim n→∞ 1 n n∑ k=1 FX σk(m) = FX(x) uniformly in m, that is P − limFXn(x)= FX(x), for all x at which FX(x) is continuous. The set of all sequences of random variables {Xn} which are invariant convergent in distribution will be denoted IDσ. ⋂ {IDσ : σ ∈ M} is equal to the set of all random variable sequences which are convergent in distribution. Theorem 2.5. If a sequence of random variables {Xn} is invariant convergent in distribution to a number a, then {Xn} is invariant convergent in probability to a. Proof. Since {Xn} is invariant convergent in distribution to the number a, for any ε > 0, we have σ − limFXn(a−ε)=0, σ− limFXn(a+ ε 2 )=1. We can write for any ε > 0, σ − limP(|Xn −a| ≥ ε) = σ − lim[P(Xn ≤ a−ε)+P(Xn ≥ a+ε)] = σ − limP(Xn ≤ a−ε)+σ − limP(Xn ≥ a+ε) = σ − limFXn(a−ε)+σ − limP(Xn ≥ a+ε) = 0+σ − limP(Xn ≥ a+ε) (since σ − limFXn(a−ε)=0) ≤ σ − limP(Xn > a+ ε 2 ) = 1−σ − limFXn(a+ ε 2 ) (since σ − limFXn(a+ ε 2 )=1) = 0. Int. J. Anal. Appl. (2022), 20:42 7 Since σ− limP(|Xn −a| ≥ ε)≥ 0, we conclude that σ− limP(|Xn −a| ≥ ε)=0, for all ε > 0, which means {Xn} is invariant convergent in probability to a. � 2.5. Invariant convergence in pth mean. Definition 2.5. Let p ≥ 1 be a fixed number. A sequence of random variables {Xn} is invariant convergent in the pth mean or in the Lp-norm to a random variable X if lim n→∞ 1 n n∑ k=1 E(|Xσk(m) −X| p)=0 uniformly in m, that is, σ − limE(|Xn −X|p)=0. If p =2, it is called the mean-square invariant convergence. The set of all invariant convergent in the pth mean or in the Lp-norm will be denoted by Epσ. ⋂ {Epσ : σ ∈ M} is equal to the set of all random variable sequences that are convergent in the pth mean or in the Lp-norm. Theorem 2.6. If a sequence of random variables {Xn} is invariant convergent in the Lp,(p ≥ 1)- norm to X, then {Xn} is invariant convergent in probability to X. Proof. For any ε > 0 and for each m, we can write P(|Xn −X| ≥ ε) = P(|Xn −X|p ≥ εp)( since p ≥ 1) ≤ E(|Xn −X|p) εp by Markov’s inequality and 1 n n∑ k=1 P(|Xσk(m) −X| ≥ ε)≤ 1 n n∑ k=1 E(|Xσk(m) −X|p) εp Since by assumption limn→∞ 1 n ∑n k=1E(|Xσk(m) −X| p) uniformly in m, we get σ − limP(|Xn −X| ≥ ε)=0, for all ε > 0. � In case σ(m) = m + 1, we have the concepts of almost convergence in probability, statisti- cally almost convergence in probability, almost convergence almost surely, almost convergence in distribution and almost convergence in Lp-norm from the Definitions 2.1, 2.2, 2.3, 2.4 and 2.5, respectively. In addition, the inclusion relations in the Theorems 2.1, 2.2, 2.3, 2.4, 2.5 and 2.6 are also valid between these concepts. These definitions and theorems have not been seen elsewhere until now. 8 Int. J. Anal. Appl. (2022), 20:42 Conflicts of Interest: The author declares that there are no conflicts of interest regarding the publi- cation of this paper. References [1] L. Breiman, Probability, SIAM, 1992. [2] P. Brockwell, R. Davis, Time Series: Theory and Methods, Springer, 1991. [3] J.S. Connor, The Statistical and Strong p-Cesàro of Sequences, Analysis, 8 (1988), 47–63. https://doi.org/10. 1524/anly.1988.8.12.47. [4] J.S. Connor, On Strong Matrix Summability with Respect to a Modulus and Statistical Convergence, Can. Math. Bull. 32 (1989), 194–198. https://doi.org/10.4153/cmb-1989-029-3. [5] H. Fast, Sur la Convergence Statistique, Colloq. Math. 2 (1951), 241–244. http://eudml.org/doc/209960. [6] J.A. Fridy, On Statistical Convergence, Analysis, 5 (1985), 301–313. https://doi.org/10.1524/anly.1985.5. 4.301. [7] J.A. Fridy, C. Orhan, Lacunary Statistical Convergence, Pac. J. Math. 160 (1993), 43–51. [8] S. Ghosal, Statistical Convergence of a Sequence of Random Variables and Limit Theorems, Appl. Math. 58 (2013), 423–437. https://doi.org/10.1007/s10492-013-0021-7. [9] S. Ghosal, I-statistical convergence of a sequence of random variables in probability, Afr. Mat. 25 (2013), 681–692. https://doi.org/10.1007/s13370-013-0142-x. [10] S. Ghosal, Sλ-Convergence of a Sequence of Random Variables, J. Egypt. Math. Soc. 23 (2015), 85–89. https: //doi.org/10.1016/j.joems.2014.03.007. [11] S. Ghosal, Weighted Statistical Convergence of Order α and Its Applications, J. Egypt. Math. Soc. 24 (2016), 60–67. https://doi.org/10.1016/j.joems.2014.08.006. [12] P.G. Hoel, S.C. Port, C.J. Stone, Introduction to Probability Theory, Houghton Mifflin Company, 1971. [13] G.G. Lorentz, A Contribution to the Theory of Divergent Sequences, Acta Math. 80 (1948), 167–190. https: //doi.org/10.1007/bf02393648. [14] I.J. Maddox, A New Type of Convergence, Math. Proc. Camb. Phil. Soc. 83 (1978), 61–64. https://doi.org/ 10.1017/s0305004100054281. [15] M. Mursaleen, Matrix Transformation Between Some New Sequence Spaces, Houston J. Math. 9 (1983), 505–509. [16] M. Mursaleen, On Infinite Matrices and Invariant Means, Indian J. Pure Appl. Math. 10 (1979), 457–460. [17] M. Mursaleen, O.H.H. Edely, On the Invariant Mean and Statistical Convergence, Appl. Math. Lett. 22 (2009), 1700–1704. https://doi.org/10.1016/j.aml.2009.06.005. [18] F. Nuray, E. Savaş, Invariant Statistical Convergence and A-Invariant Statistical Convergence, Indian J. Pure Appl. Math. 25 (1994), 267–274. [19] E. Savaş, Some Sequence Spaces Involving Invariant Means, Indian J. Math. 31 (1989), 1–8. [20] E. Savaş, Strongly σ-Convergent Sequences, Bull. Calcutta Math. 81 (1989), 295–300. [21] R.A. Raimi, Invariant Means and Invariant Matrix Methods of Summability, Duke Math. J. 30 (1963), 81–94. https://doi.org/10.1215/s0012-7094-63-03009-6. [22] S. Ross, A First Course in Probability, Macmillan Publishing Company, 2nd ed., 1984. [23] P. Schaefer, Infinite Matrices and Invariant Means, Proc. Amer. Math. Soc. 36 (1972), 104–110. https://doi. org/10.1090/s0002-9939-1972-0306763-0. [24] T. Šalát, On Statistically Convergent Sequences of Real Numbers, Math. Slovaca, 30 (1980), 139–150. http: //dml.cz/dmlcz/136236. [25] I.J. Schoenberg, The Integrability of Certain Functions and Related Summability Methods, Amer. Math. Mon. 66 (1959), 361–775. https://doi.org/10.1080/00029890.1959.11989303. https://doi.org/10.1524/anly.1988.8.12.47 https://doi.org/10.1524/anly.1988.8.12.47 https://doi.org/10.4153/cmb-1989-029-3 http://eudml.org/doc/209960 https://doi.org/10.1524/anly.1985.5.4.301 https://doi.org/10.1524/anly.1985.5.4.301 https://doi.org/10.1007/s10492-013-0021-7 https://doi.org/10.1007/s13370-013-0142-x https://doi.org/10.1016/j.joems.2014.03.007 https://doi.org/10.1016/j.joems.2014.03.007 https://doi.org/10.1016/j.joems.2014.08.006 https://doi.org/10.1007/bf02393648 https://doi.org/10.1007/bf02393648 https://doi.org/10.1017/s0305004100054281 https://doi.org/10.1017/s0305004100054281 https://doi.org/10.1016/j.aml.2009.06.005 https://doi.org/10.1215/s0012-7094-63-03009-6 https://doi.org/10.1090/s0002-9939-1972-0306763-0 https://doi.org/10.1090/s0002-9939-1972-0306763-0 http://dml.cz/dmlcz/136236 http://dml.cz/dmlcz/136236 https://doi.org/10.1080/00029890.1959.11989303 Int. J. Anal. Appl. (2022), 20:42 9 [26] H. Steinhaus, Sur la Convergence Ordinaire et la Convergence Asymptotique, Colloq. Math. 2 (1951), 73–74. [27] D. Williams, Probability with Martingales, Cambridge University Press, 1991. 1. Introduction 2. Some new types of invariant convergence 2.1. Invariant convergence in probability 2.2. Statistically invariant convergence in probability 2.3. Invariant convergence almost surely 2.4. Invariant convergence in distribution 2.5. Invariant convergence in pth mean References