() CUBO A Mathematical Journal Vol.17, No¯ 02, (97–122). June 2015 Spacetime singularity, singular bounds and compactness for solutions of the Poisson’s equation Carlos Cesar Aranda Blue Angel Navire research laboratory, Rue Eddy 113 Gatineau, QC, Canada carloscesar.aranda@gmail.com ABSTRACT A black hole is a spacetime region in whose interior lies a structure known as a space- time singularity whose scientific description is profoundly elusive, and which depends upon the still missing theory of quantum gravity. Using the classical weak comparison principle we are able to obtain new bounds, compactness results and concentration phe- nomena in the theory of Newtonian potentials of distributions with compact support which gives a suitable mathematical theory of spacetime singularity. We derive a rig- orous renormalization of the Newtonian gravity law using nonlinear functional analysis and we have a solid set of astronomical observations supporting our new equation. This general setting introduces a new kind of ill posed problem with a very simple physical interpretation. RESUMEN Un hoyo negro es una región espacio-temporal en cuyo interior hay una estructura llamada singularidad espacio-temporal cuya descripción cient́ıfica es dif́ıcil de encontrar, y que depende de la aún inexistente teoŕıa de la gravedad cuántica. Usando el clásico principio de comparación débil, aqúı probamos nuevas cotas, resultados de compacidad y fenómenos de concentración en la teoŕıa de potenciales Newtonianos de distribuciones de soporte compacto, que dan una teoŕıa matemática adecuada de la singularidad espacio-temporal. Derivamos una rigurosa renormalización de la ley de gravitación Newtoniana usando análisis funcional no lineal y tenemos un contundente conjunto de datos de observaciones astronómicas que apoyan nuestra nueva ecuación. Este marco general introduce una nueva forma de problema mal-puesto con una interpretación f́ısica muy simple. Keywords and Phrases: Black hole, spacetime singularity, quantum field theory, Newtonian potentials, elliptic equations, compact imbedding, Sobolev’s spaces. 2010 AMS Mathematics Subject Classification: 35J25, 35J60, 35J75. 98 Carlos Cesar Aranda CUBO 17, 2 (2015) 1 Introduction. In [2, 3], the authors introduced a new concentration phenomena for the Poisson’s equation using techniques from nonlinear functional analysis. In this article we are concerned with several sim- ple consequences of this new concentration of compactness results. Using the classical theory of Newtonian potentials of distributions with compact support we are able to derive concentration of compactness for Newtonian potentials with singular behaviour. For a review of this topic see [10]. Newtonian potentials are useful in the description of gravity fields of celestial bodies [5, 20, 21]. Today black holes in gravity theory and astronomy plays a central role [8, 12, 15, 27, 30, 31, 33, 40, 46]. The interior of a black hole is usually called spacetime singularity [7, 31]. In [3] we obtain the existence of a sequence {Pj} ∞ j=1 ∈ C 2(Ω) for any Ω bounded domain in RN such that − limj→∞ ∆Pj = ∞ uniformly on Ω and 0 < Pj(x) ≤ Pj+1(x) ≤ Cte for all x ∈ Ω. This sequence proof that it is possible to do rigorous treatment of divergence to infinite in the frame of Newtonian potentials extending rigorous quantum field theories on large scale [37, 38]. Black holes are complicated real objects, and our Newtonian equation is a mathematical object but we have astronomical observations of supermassive black holes given a concrete physical significance to this new theoretical frame [27]. This is a remarkable fact in gravity theory [13, 41]. Lemma 1.1 (Lemma 1 page 277 [11]). Let Ω be an open set of RN, f ∈ D′(Ω) and u a solution (in the sense of distribution) of Poisson’s equation ∆u = f on Ω. Then for every bounded open set Ω1 with Ω1 ⊂ Ω there exists f1 ∈ E ′ the space of distributions on RN with compact support, such that f1 = f on Ω, u = the Newtonian potential of f1 on Ω1. (1) Therefore for Ω1 ⊂ Ω and our sequence {Pj} ∞ j=1 there exists a sequence {f1,j} ∞ j=1 ∈ E ′ the space of distributions on RN with compact support, such that f1,j = ∆Pj on Ω, Pj = the Newtonian potential of f1,j on Ω1. (2) We have associated to each pair {Pj, Ω1} a gravitational Newtonian potential defined on all R N. Using this lemma we have a very simple Newtonian interpretation of the interior of a black hole. We have a set of solid astronomic observations supporting our new equation. CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 99 Black hole name Solar mass (sun=1) References Holmberg 15A 170.000.000.000 [23] S5 0014+813 40.000.000.000 [16] SDSS J085543.40-001517.7 25.000.000.000 [47] APM 08279+5255 23.000.000.000 [35] NGC 4889 21.000.000.000 [25] Central black hole of Phoenix cluster 20.000.000.000 [26] SDSS J07451.78+734336.1 19.500.000.000 [47] OJ 285 primary 18.000.000.000 [39] SDSS J08019.69+373047.3 15.140.000.000 [47] SDSS J115954.33+20192.1 14.120.000.000 [47] SDSS J075303.34+423130.8 13.500.000.000 [47] SDSS J081855.77+095848.0 12.000.000.000 [47] SDSS J0100+2802 12.000.000.000 [44] SDSS J082535.19+512706.3 11.220.000.000 [47] SDSS J013127.34-0321000.1 11.000.000.000 ♦ Central black hole of RX J1532.9+3021 10.000.000.000 ♦ QSO B2126-158 10.000.000.000 [16] SDSS J015741.57-010629.6 9.800.000.00 [47] NGC 3842 9.700.000.000 [25] SDSS J2330301.45-093930.7 9.120.000.000 [47] SDSS J075819.70+202300.9 7.800.000.000 [47] SDSS J080956.02+50200.9 6.450.000.000 [47] SDSS J0142114.75+0023224.2 6.310.000.000 [47] Messier 87 6.300.000.000 ♦ QSO B0746+254 5.000.000.000 [16] QSO B2149-306 5.000.000.000 [16] NGC 1277 5.000.000.000 ♦ SDSS J090033.50+421547.0 4.700.000.000 [47] Messier 60 4.500.000.000 ♦ SDSS J011521.20+152453.3 4.100.000.000 [47] QSO B0222+185 4.000.000.000 [16] Hercules A (3C 348) 4.000.000.000 ♦ SDSS J213023.61+122252.0 3.500.000.000 [47] 100 Carlos Cesar Aranda CUBO 17, 2 (2015) J173352.23+540030.4 3.400.000.000 [47] SDSS J025021.76-075749.9 3.100.000.000 [47] SDSS J030341.04-002321.9 3.000.000.000 [47] QSO B0836+710 3.000.000.000 [16] SDSS J224956.08+000218.0 2.630.000.000 [47] SDSS J030449.85-000813.4 2.400.000.000 [47] SDSS J234625.66-001600.4 2.240.000.000 [47] ULAS J1120+0641 2.000.000.000 ♦ QSO 0537-286 2.000.000.000 [16] NGC 3115 2.000.000.000 ♦ Q0906+6930 2.000.000.000 ♦ QSO B0805+614 1.500.000.000 [16] Messier 84 1.500.000.000 ♦ QSO B225155+2217 1.000.000.000 [16] QSO B1210+330 1.000.000.000 [16] NGC 6166 1.000.000.000 ♦ Cygnus A 1.000.000.000 ♦ Sombrero Galaxy 1.000.000.000 ♦ Markarian 501 900.000.000-3.400.000.000 ♦ PG 1426+015 467.740.000 [28] 3C 273 550.000.000 [28] Messier 49 560.000.000 ♦ PG 0804+761 190.550.000 [28] PG 1617+175 275.420.000 [28] PG 1700 + 518 60.260.000 [28] NGC 4261 400.000.000 ♦ NGC 1275 340.000.000 ♦ 3C 390.3 338.840.000 [28] II Zwicky 136 144.540.000 [28] PG 0052+251 218.780.000 [28] Messier 59 270.000.000 ♦ PG 1411+442 79.430.000 [28] Markarian 876 240.000.000 [28] Andromeda Galaxy 230.000.000 ♦ PG 0953+414 182.000.000 [28] PG 0026+129 53.700.000 [28] Fairall 9 79.430.000 [28] CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 101 Markarian 1095 182.000.000 [28] Messier 105 200.000.000 ♦ Markarian 509 57.550.000 [28] OJ 287 secondary 100.000.000 [39] RX J124236.9-1111935 100.000.000 ♦ Messier 85 100.000.000 ♦ NGC 5548 123.000.000 ♦ PG 1221+143 40.740.000 [28] Messier 88 80.000.000 ♦ Messier 81 70.000.000 ♦ Markarian 771 75.860.000 [28] Messier 58 70.000.000 ♦ PG 0844+349 21.380.000 [28] Centaurus A 55.000.000 ♦ Markarian 79 52.500.000 [28] Messier 96 48.000.000 ♦ Markarian 817 43.650.000 [28] NGC 3227 38.900.000 [28] NGC 4151 primary 40.000.000 ♦ 3C 120 22.900.000 ♦ Markarian 279 41.700.000 ♦ NGC 3516 23.000.000 ♦ NGC 863 17.700.00 ♦ Messier 82 30.000.000 ♦ Messier 108 24.000.000 ♦ NGC 3783 9.300.000 ♦ Markarian 110 5.620.000 ♦ Markarian 335 6.310.000 ♦ NGC 4151 secondary 10.000.000 ♦ NGC 7469 6.460.000 ♦ IC 4329 A 5.010.000 ♦ NGC 4593 8.130.000 ♦ Messier 61 5.000.000 ♦ Messier 32 1.500.000-5.000.000 ♦ Sagittarius A* 4.100.000 ♦ NGC 4051 1.300.000 ♦ ♦ Only main 102 Carlos Cesar Aranda CUBO 17, 2 (2015) references are provided. Our function P = limj→∞ Pj is very ’rough’ from the point of view of regularity on Sobolev’s spaces in opposition of the Poisson’s equation ∆u = 0 on Ω. Theorem 1.2 (Weyl page 118[10]). Let Ω ⊂ RN be open and u ∈ L1loc(Ω) satisfy ∫ Ω u∆v = 0 for all v ∈ C∞0 (Ω). Then u ∈ C ∞(Ω) and ∆u = 0. A simpler use of Green’s identity allow us to imply the discontinuity at infinitum of the functional G : C2(Ω) × C2(Ω) → R, G(u, v) = ∫ Ω (v∆u − u∆v) dx − ∫ ∂Ω ( v ∂u ∂n − u ∂v ∂n ) dγ, (3) where Ω is a bounded domain with C1 boundary ∂Ω. The functional G for a fixed pair (u, v) ∈ C2(Ω) × C2(Ω) in a smooth bounded domain Ω ⊂ RN satisfies the Green’s identity G(u, v) = 0 (for a proof of Green’s formula see page 20 [29]). This equality is a consequence of the divergence Theorem, (for a proof of the divergence Theorem stated by E. Heinz see page 46 [36], also page 570 [11]). Let us to remember several classical results: Theorem 1.3. [Theorem 6.6 [18]] Let Ω be a C2,α domain in RN and let u ∈ C2,α(Ω) be a solution of the equation Lu = N∑ i,j=1 aijuxi,xj + N∑ i=1 biuxi + cu = f, (4) where f ∈ Cα(Ω) and the coefficients of L satisfy, for positive constants λ, Λ N∑ i,j=1 aijξiξj ≥ λ | ξ | 2, for all x ∈ Ω, ξ ∈ RN, (5) | ai,j |0,α;Ω, | bi |0,α;Ω, | ci |0,α;Ω≤ Λ. (6) Let ϕ ∈ C2,α(Ω) and suppose u = ϕ on ∂Ω. Then | u |2,α;Ω≤ C{| u |0,Ω + | ϕ |2,α;Ω + | f |0,α;Ω}, (7) where C = C(N, α, λ, Λ, Ω). Lu = Di ( aij(x)Dju + b i(x)u ) + ci(x)Diu + d(x)u, (8) L∗u = Di ( aijDju − c iu ) − biDiu + du, (9) aij(x)ξiξj ≥ λ | ξ | 2, for all x ∈ Ω, ξ ∈ RN, (10) ∑ | aij(x) |2≤ Λ2, (11) λ−2 ∑ ( | bi(x) |2 + | ci(x) |2 ) + λ−1 | d(x) |≤ v2, (12) CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 103 Theorem 1.4. [Theorem 9.11 [18]] Let Ω be an open set in RN and W2,p(Ω)∩Lp(Ω), 1 < p < ∞, a strong solution of the equation Lu = f in Ω where the coefficients of L satisfy for positive constants λ, Λ aij ∈ C0(Ω), bi, c ∈ L∞(Ω), f ∈ Lp(Ω); (13) aijξiξj ≥ λ | ξ | 2 for all ξ ∈ RN, (14) | ai,j |, | bi |, | c |≤ Λ, (15) where i, j = 1, . . . , N. Then for any domain Ω′ ⊂⊂ Ω, ‖ u ‖2,p;Ω′≤ C (‖ u ‖p;Ω + ‖ f ‖p;Ω) , (16) where C depends on N, p, λ, Λ, Ω′, Ω and the moduli of continuity of the coefficients aij on Ω′. Now the existence of our concentrating sequence of non negative, bounded functions {Pj} ∞ j=1 is a compactness property not established and noncontradictory with the statements in Theorems 1.3 and 1.4. This is a singular property that using very elementary techniques, allow us to obtain new bounds and several interesting results related to the Newtonian potential in the theory of distributions [11]. Moreover we built a sequence Fj : [a, b] → [0,∞) satisfying limj→∞ F′′j (x) = −∞ in measure, Cte ≥ Fj+1 ≥ Fj ≥ 0 and each Fj non increasing on [a, b]. By considering Fj(x1, x2, . . . , xN) = Fj(x1) we obtain a sequence for bounded smooth domains in R N, N ≥ 2 with similar properties. 2 Preliminaries. The results of this section are contained in [3]. For the sake of the readability, we stated and prove it here again. Lemma 2.1. Let B(0, R) be a ball of radius R > 0 in RN, N > 2. Consider the singular nonlinear elliptic equation −∆Pj = Hj(Pj) in B(0, R) Pj = 0 on ∂B(0, R). (17) where Hj : (0,∞) → (0,∞) is locally Hölder continuous function Hj(s) = { s−j if 0 < s < 1, s−1 if s ≥ 1. Then the next properties holds: (i) The sequence {Pj} ∞ j=1 ∈ C 2(B(0, R)) ∩ C(B(0, R)) are radial functions with ∂P ∂r < 0. (ii) The sequence {Pj} ∞ j=1 satisfies Pj ≤ Pj+1. (iii) The sequence {Hj(Pj)} ∞ j=1 satisfies Hj(Pj) ≤ Hj+1(Pj+1). (iv) The sequence {Pj} ∞ j=1 satisfies w ≤ Pj ≤ e, where −∆v = v −1 in B(0, R), v = 0 on ∂B(0, R), 104 Carlos Cesar Aranda CUBO 17, 2 (2015) −∆e = e−1 in B(0, R), e = 1 on ∂B(0, R) and −∆w = e−1 in B(0, R), w = 0 on ∂B(0, R). Proof. The enunciate (i) is a consequence of classical results on radial symmetry. The points (ii) and (iii) have been stated at [2, 3] and the point (iv) is proved in [1]. Theorem 2.2 ([3]). Let B(0, R) ⊂ RN, a ball of radius R, with N ≥ 3. Then there exists a sequence of radial, nonnegative and bounded functions {Pj} ∞ j=1 and 0 ≤ R0 < R such that − lim j→∞ ∆Pj = ∞ on A(R0, R), (18) where A(R0, R) is the annulus of external radius R and internal radius R0. Moreover Pj ∈ C∞(A(R0, R)) and Pj ≤ Pj+1. Proof. Our proof is a reductio ad absurdum procedure. Let us to consider P = lim j→∞ Pj. (19) Now if there exists no sequence such that it is stated in our Theorem 2.2, then lim rրR P(r) ≥ 1, (20) because if limrրR P(r) < 1 for a all nonnegative and small enough ǫ, there exist δ > 0 such that P(r) ≤ 1 − ǫ for all r ∈ (R − δ, R). Therefore Pj(r) ≤ P(r) ≤ 1 − ǫ for all r ∈ (R − δ, R) and − limj→∞ ∆Pj = limj→∞ Hj(Pj) = limj→(Pj) −j ≥ limj→∞(1 − ǫ) −j = ∞ on A(R − δ, R). (21) Similarly if there exists no sequence satisfying our Theorem 2.2 then lim j→∞ Hj(Pj(r)) < ∞ for all r ∈ [0, R), (22) because if lim j→∞ Hj(Pj(r)) = ∞ for r0 ∈ [0, R), (23) from Hj(Pj(r)) ≥ Hj(Pj(r0)) for all | x |= r ∈ [r0, R), (24) we deduce − limj→∞ ∆Pj = limj→∞ Hj(Pj) ≥ limj→∞ Hj(Pj(r0)) = ∞ on A(r0, R). (25) CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 105 Contradiction so 22 holds and we obtain that P ∈ C1(B(0, R)), because using Theorem 9.11 page 235 in [18] we derive ‖ Pj ‖H2,p(Ω′′) ≤ C(N, p, Ω ′, Ω′′) { ‖ Pj ‖Lp(Ω′) + ‖ Hj(Pj) ‖Lp(Ω′) } ≤ C(N, p, Ω′, Ω′′) { ‖ e ‖Lp(Ω′) + ‖ limj→∞ Hj(Pj(r3)) ‖Lp(Ω′) } (26) where Ω′ ⊂ Ω′′, Ω′′ ⊂ B(0, R), p > N and r3 = supx∈Ω′ | x |. According to Theorem 7.26 page 171 [18], we have the bounds ‖ Pj ‖ C 1,1− N p (Ω′′) ≤‖ Pj ‖H2,p(Ω′′) . (27) Therefore P ∈ C1(B(0, R)) and Pj → P in C1,αloc(B(0, R) for 0 < α < 1 − N p . One more time if there exist no sequence satisfying the conclusions of Theorem 2.2 we imply P : [0, R) → [0,∞) is a strictly nonincreasing radial function because if there exist 0 ≤ r1 < r2 < R with P(r1) = P(r2) and the fact of being P nonincreasing implies −∆P = 0 on the annulus A(r1, r2). Using ‖ Pj ‖C1,α(A(r1,r2)≤ C and a nonnegative test function ϕ with support contained in A(r1, r2): 0 = ∫ A(r1,r2) ∇P · ∇ϕdx = limj→∞ ∫ A(r1,r2) ∇Pj · ∇ϕdx = ∫ A(r1,r2) Hj(P1)ϕdx ≥ ∫ A(r1,r2) H1(P1)ϕdx > 0. (28) Contradiction. So from the negation of the conclusion of Theorem 2.2 we derive that the function P : [0, R) → R satisfies P ∈ C1(0, R), P is strictly nonincreasing in (0, R) and limrրR P ≥ 1 and therefore P(r) > 1 for all r ∈ [0, R). (29) Finally we are ready to finish the proof of Theorem 2.2. Independently of the hypothesis in the reductio ad absurdum, there exists 0 < r0 < R such that H1(P1(r0)) > H1(1) = 1 and therefore using (iii) of Lemma 2.1 1 < H1(P1(r0)) ≤ Hj(Pj(r0)) for all j ≥ 1. (30) But limj→∞ Pj(r0) = P(r0) > 1 and therefore for j big enough Pj(r0) > 1 and Hj(Pj(r0)) = H1(Pj(r0)) < H1(1) = 1. (31) Therefore 1 < H1(P1(r0)) ≤ Hj(Pj(r0)) < 1, (32) for j big enough. In page 291 [11] it is stated that for every function f ∈ Cm,α(Ω), 0 < α < 1, m ≥ 1 the solutions of Poisson’s equation ∆u = f on Ω are of class Cm+2,α on Ω, and so Pj are of class C∞(A(r, R)). This end the proof. 106 Carlos Cesar Aranda CUBO 17, 2 (2015) The next lemma is new Lemma 2.3. The sequence {Hj(Pj)} ∞ j=1 is unbounded in C α loc(A(r, R)). Proof. In Theorem 4.6 page 60 [18] it is stated that: let Ω be a domain in RN and let u ∈ C2(Ω), f ∈ Cα(Ω) satisfy Poisson’s equation ∆u = f. Then u ∈ C2,α(Ω) and for any two concentric balls B1 = B(x0, R), B2 = B(x0, 2R) ⊂⊂ Ω we have | u |′2,α;B1≤ C(| u |0;B2 +R 2 | f |′0,α;B2). (33) Therefore limj→∞ | Hj(Pj) | ′ 0,α;B2 = ∞. 3 The construction of the sequence of functions Fj. In [19] it is stated that [page XVIII, [19]] When we refer to a set F as a fractal, therefore, we will typically have the following in mind. (i) F has a fine structure, i. e. detail on arbitray small scales. (ii) F is too irregular to be described in traditional geometrical language, both locally and globally. (iii) Often F has some form of self-similarity, perhaps approximate or statistical. (iv) Usually, the ‘fractal dimension’ of F (defined in some way) is greater than its topological dimension. (v) In most cases of interest F is defined in a very simple way, perhaps recursively. Moreover [Page XXII, [19]] The highly intricate structure of the Julia set illustrated in figure 0.6 stems from the single quadratic function f(z) = z2 + c for a suitable constant c. Although the set is not strictly self-similar in the sense that the Cantor’s set and von Koch curve are, it is ‘quasi-self-similar’ in that arbitrarily small portions of the set can be magnified an distorted smoothly to coincide with a large part of the set. CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 107 Let [0, a] be a bounded interval in R. Let us to consider, the infinite sequence of linear functions h(x) = a − x, h0(x) = a + 1 − x, h1(x) = a + 1 + 1 2 − x, · · · hj(x) = a + j∑ n=0 1 2n − x. We introduce a infinite sequence of functions defined on [0, a], p(x) = p(x; s; k) = −t(x − s k a)2 + cs for all x ∈ [a s k , a s + 1 k ], s = 0, 1, 2, . . . k − 1, p0(x) = p0(x; s; k0) = −t0(x − s k0 a)2 + c0,s for all x ∈ [a s k0 , a s + 1 k0 ], s = 0, 1, 2, . . . k0 − 1, p1(x) = p1(x; s; k1) = −t1(x − s k1 a)2 + c1,s for all x ∈ [a s k1 , a s + 1 k1 ], s = 0, 1, 2, . . . k1 − 1, . . . , pj(x) = pj(x; s; kj) = −tj(x − s kj a)2 + cj,s for all x ∈ [a s kj , a s + 1 kj ], s = 0, 1, 2, . . . kj − 1. Where the sequence c0 = 1, c0,0 = 1 + 1, c1,0 = 1 + 1 + 1 2 , . . . , cj,0 = 1 + j∑ s=0 1 2j . satisfies cs = p ( s k ; s; k ) for all s = 1, . . . , k − 1, (34) c0,s = p0 ( s k ; s; k0 ) for all s = 1, . . . , k0 − 1, (35) c1,s = p1 ( s k ; s; k1 ) for all s = 1, . . . , k1 − 1, (36) · · · , (37) cj,s = pj ( s kj ; s; kj ) for all s = 1, . . . , kj − 1. (38) 108 Carlos Cesar Aranda CUBO 17, 2 (2015) We have the association: {cs} ∞ s=0 ←→ h ←→ p, {c0,s} ∞ s=0 ←→ h0 ←→ p0, {c1,s} ∞ s=0 ←→ h1 ←→ p1, . . . , {cj,s} ∞ s=0 ←→ hj ←→ pj. The choice of the sequence of non negative numbers k ∪ {kj} ∞ j=1 determines t ∪ {tj} ∞ j=1. We keep k such that the equation p(x; 0, k) = h0(x), (39) has no solutions. Similarly we divide this first interval [0, a k ] in k0 intervals and setting k0 = kk0 such that the equation p0(x; 0, k) = h1(x), (40) has no solution. Now we complete the procedure by induction. Therefore lim j→∞ tj = ∞. (41) It is follow that the non decreasing, bounded, sequence of functions {p}∞j=0 defined on [0, a] has second derivative defined almost everywhere and p′′j (x) = −2tj. Using the Rolle’s Theorem for the functions pj(·; s; kj), hj(·) in the interval [ sa kj , (s+1)a kj ], we deduce the existence of xj,s ∈ ( sa kj , (s+1)a kj ) such that p′j(xj,s; s; kj) = h ′ j(xj,s) = −1, (42) where xj,s = 1 2tj + sa kj . (43) Similarly for x̃j,s = 1 20tj + sa kj , (44) we have p′j(x̃j,s; s; kj) = − 1 10 . (45) For j big enough, let us to consider the sequence of intervals (sa kj − δj, sa kj + δj) ⊂ [0, a] for s = 1, 2, . . . , kj − 1 where δj = min{ 1 200tj , a kjj 2 , a kj − 1 2tj 10 }. (46) CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 109 By smoothing each pj on ( sa kj − δj, sa kj + δj), we obtain the desired sequence {Fj} ∞ j=1 ∈ C ∞[a, b]. We point that C∞ extension is a nontrivial task, see for example page 136 [18] or the Nikolskii’s extension method page 69 [29] and the Calderon’s extension method page 72 [29]. Therefore we give a full description of our procedure. For the smoothing method we use the functions S(x; a; b) = x − ∫x 0 g(t)dt, (47) where g(x) = 0 for x < a − δ, g(x) = 1 for a < x < b, g(x) = 0 for x > b + δ, 0 < δ ∈ R, and g ∈ C∞(R) (see the C∞ Urysohn Lemma page 245, [14]). It it follows that S(x; a; b) = x for x < a − δ, S(x; a; b) = x − c for x > b + δ and S′(x; a; b) ≥ 0. Taking a suitable composition with functions S(·; a; b) we accomplish with the smoothing procedure and moreover F′j(x) ≤ 0. The sequence {∆Fj} ∞ j=1 converges in measure to −∞. Note that the set of numbers {xj,s, x̃j,s} are numerable and dense in [0, a]. Therefore by 42 and 45 we imply if Fj(x1, x2, . . . , xN) = Fj(x1) then lim j→∞ ‖ Fj ‖C1,α(B(x,r))= ∞, (48) for any ball B(x, r) with center at x with radius r and B(x, r) ⊂ Ω. It follows that lim j→∞ ‖ Fj ‖C2,α(B(x,r))= ∞. (49) Let us to consider the function IAj(x) = { 1 if x ∈ Aj, 0 if x /∈ Aj, (50) where Aj = {x ∈ ( sa kj + δj, (s+1)a kj − δj) | s = 0, 1, 2, . . . kj − 1}. Therefore ∫a 0 F′′j IAjdx = tj kj−1∑ s=0 ( a kj − 2δj ) = tja − 2 kj−1∑ s=0 δj ≥ tja − 2 a j2 . So limj→∞ ∫a 0 F′′j IAjdx = ∞ and lim j→∞ ‖ F′′j ‖Lp[0,a]= ∞ for all 1 ≤ p ≤ ∞. (51) 4 A primer analysis. 4.1 A new kind of ill-posed problem. The concept of a well-posed problem of mathematical physics was introduced by J. Hadamard. The solution of any quantitative problem usually ends in a equation z = R(u) where u is the 110 Carlos Cesar Aranda CUBO 17, 2 (2015) initial data and z is the solution, R : U → Z, U and Z are metric spaces with distances ρU and ρZ respectively. The problem of determining the solution z in the space Z from the initial data u in the space U is said to be well-posed on the pair of metric space (Z, U) if the following three conditions are satisfied: (i) For every element u ∈ U there exists a solution z in the space Z. (ii) The solution is unique. (iii) For every positive number ǫ > 0 there exists a positive number δ such that ρU(u1, u2) ≤ δ implies ρZ(z1, z2) ≤ ǫ, where z1 = S(u1), z2 = S(u2). Problems that do not satisfy them are said ill-posed. The sequence {Pj} ∞ j=1 is a new kind of ill- posed problem related to Sobolev’s spaces or even for the Laplacian operator in the the context of distributions. Given Ω a bounded domain in RN and u ∈ W1,1(Ω) whose laplacian is a bounded measure µ on Ω, we call normal derivative in the sense of distributions of u on ∂Ω the distribution v1 defined on RN by 〈v1, ϕ〉 = ∫ Ω ϕdµ + ∫ Ω ∇ϕ · ∇udx, ϕ ∈ D(RN) (52) The distribution v1 defined by 52 is of compact support in ∂Ω, if suppϕ ∩ ∂Ω = ∅ then ϕIΩ and ∇(ϕIΩ) = (∇ϕ)IΩ. Then by the definition of Laplacian in the sense of distributions ∫ Ω ϕdµ = 〈∆u, ϕIΩ〉 = − ∫ ∇(ϕIΩ) · ∇udx = − ∫ Ω ∇ϕ · ∇udx. (53) Proposition 4.1 (page 500 [11]). Let Ω be a regular bounded open set with boundary of class W2,∞, µ a bounded Radon measure and v1 a Radon measure on ∂Ω. Let us to consider the Neumann problem u ∈ W1,1(Ω), ∆u = µ in D ′(Ω), (54) v1 is the normal derivative on ∂Ω of u. (55) (i) There exists a solution of 54,55 if and only if ∫ ∂Ω dv1 = ∫ Ω dµ. (ii) If ∫ ∂Ω dv1 = ∫ Ω dµ then the solution of 54, 55 is defined to whitin an additive constant and u ∈ W1,p(Ω) for all 1 ≤ p < N N−1 . Our sequence {∆Pj} ∞ j=1 is non bounded in L ∞(Ω) therefore the limit P is outside of the scope of application of Proposition 4.1. CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 111 4.1.1 Green’s identities. We show the nature of the discontinuity of the the sequence {Pj} ∞ j=1 in Sobolev’s spaces and in the context of distribution theory. From (page 17 [18] or [6]) ∫ Ω (∆u) vdx = ∫ ∂Ω ∂u ∂n vdγ − ∫ Ω ∇u · ∇vdx for all u ∈ C2(Ω), for all v ∈ C1(Ω), (56) we calculate lim j→∞ ∫ Ω ∇Pj · ∇vdx = ∞ for all non negative v ∈ C 1(Ω) with v = 0 on ∂Ω. (57) Taking v ≡ 1 in 56, we derive lim j→∞ ∫ ∂Ω ∂Pj ∂n dγ = −∞. (58) Now we use the second Green’s identity ∫ Ω (v∆u − u∆v) dx = ∫ ∂Ω ( v ∂u ∂n − u ∂v ∂n ) dγ for all u, v ∈ C2(Ω), (59) it follows lim j→∞ ∫ ∂Ω ( v ∂Pj ∂n − Pj ∂v ∂n ) dγ = −∞ for all nonnegative v ∈ C2(Ω). (60) Therefore for non negative v ∈ C2(Ω) with v = 0 on ∂Ω, we deduce ∫ Ω ((v + ǫ)∆Pj − Pj∆v) dx = ǫ ∫ ∂Ω ∂Pj ∂n dγ − ∫ ∂Ω Pj ∂v ∂n dγ, (61) ∫ Ω (v∆Pj − Pj∆v) dx = − ∫ ∂Ω Pj ∂v ∂n dγ, (62) and letting j →∞ we demonstrate the discontinuity at infinitum of the functional 3. Theorem 4.2 (Theorem 4.11 page 85 [29]). Let Ω ∈ RN a bounded domain with smooth boundary; if 1 ≤ p < N put 1 q = 1 p − 1 N−1 p−1 p ; if p = N, put q ≥ 1. There exists a unique mapping Z ∈ [W2,p(Ω) → W1,q(∂Ω)] such that u ∈ C∞(Ω) =⇒ Zu = u. The second Green’s identity 59 is valid on W2,2(Ω), it follows from 58 and Theorem 4.2 that our sequence {Pj} ∞ j=1 is unbounded in W 1,2 N−1 N−2 (∂Ω). From 57 we imply lim j→∞ ‖ Pj ‖W1,p(Ω)= ∞ for all 1 ≤ p ≤ ∞, (63) and lim j→∞ ‖ Pj ‖W2,p(Ω)= ∞ for all 1 ≤ p ≤ ∞. (64) Therefore the sequence is unbounded in the domain of definition of the trace operator stated in Theorem 4.2. Similar considerations are implied easily from 112 Carlos Cesar Aranda CUBO 17, 2 (2015) Theorem 4.3 (page 5 [29]). Let Ω be a bounded domain with lipschitzian boundary. Then there exists a uniquely defined, linear an continuous mapping T : Wk,2(Ω) → L2(∂Ω) such that for x ∈ ∂Ω and v ∈ C∞(Ω), it is defined by T(v)(x) = v(x) Theorem 4.4 (page 135 [36]). We consider a solution u = u(x) ∈ C2(Ω) of Poisson’s differential equation ∆u(x) = f(x), x ∈ Ω in the domain Ω ⊂ RN, N ≥ 3. For each ball BR(a) ⊂⊂ Ω then we have the identity u(a) = 1 RN−1ωN ∫ |x−a|=R u(x)dσ − 1 (N − 2)ωN ∫ |x−a|≤R ( | x − a |2−N −R2−N ) f(x)dx (65) The same discontinuity at infinitum appears in the context of singular phenomena in nonlinear elliptic problems [1, 2, 9, 17, 34]. ∫ Ω ((v + ǫ)∆Uj − Uj∆v) dx = ǫ ∫ ∂Ω ∂Uj ∂n dγ − ∫ ∂Ω Uj ∂v ∂n dγ, (66) ∫ Ω (v∆Uj − Uj∆v) dx = − ∫ ∂Ω Uj ∂v ∂n dγ, (67) where Uj solves the problem −∆Uj = U −γ j in Ω, (68) Uj = 1 j on ∂Ω, (69) with γ > 1. Moreover limj→∞ Uj = U ∈ C 2(Ω) ∩ C0(Ω) and ∫ Ω U−γdx = ∞, showing the same kind of discontinuity at infinitum. This discontinuity property is an interesting example in the Friedrichs method of extension of semibounded operators to self-adjoint operators (page 228 [4], see also page 205 [24]). 4.1.2 Integration by parts. In the one dimension if u, v ∈ W1,p(I) with 1 ≤ p ≤ ∞, then ∫y x u′v = u(x)v(x)−u(y)v(x)− ∫x y uv′ for all x, y ∈ I but even if a distribution has second distributional derivative the integration by parts is not true. For example for fh = ∫x 0 1 h I[−h,0](t)dt, we have: ∫y x [f′h]ϕ ′ = [f′h(y)]ϕ(y) − [f ′ h(x)]ϕ(x) − ∫y x [f′′h]ϕ = 0 for all ϕ ∈ C ∞ 0 (x, y)∫y x [f′h]ϕ ′ = [fh(y)]ϕ ′(y) − [fh(x)]ϕ ′(x) − ∫y x [fh]ϕ ′′ = ϕ ′(−h)−ϕ′(0) −h for all ϕ ∈ C∞0 (x, y). (70) Taking a radial ϕ ∈ D(A(r, R)) we get ∫ A(r,R) ∇Pj · ∇ϕdx = ∫ A(r,R) ∂Pj ∂r xi r ϕ ∂r xi r dx = ∫ A(r,R) ∂Pj ∂r ∂ϕ ∂r rdx = ∫R r rN−1dr ∫ SN−1 ∂Pj ∂r ∂ϕ ∂r rdω = ∫R r ∂Pj ∂r ∂ϕ ∂r rNdr ∫ SN−1 dω. (71) CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 113 So lim j→∞ ∫R r ∂Pj ∂r ∂ϕ ∂r dr = ∞ (72) and the integration by parts rule not hold in the limit because ∫R r ∂Pj ∂r ∂ϕ ∂r dr = Pj(R) ∂ϕ ∂r (R) − Pj(r) ∂ϕ ∂r (r) − ∫R r Pj ∂2ϕ ∂r2 dr. (73) Also the Cantor’s function not satisfies the integration by part rule (this function is monotone and it has zero derivative almost everywhere). 4.1.3 A detour with monotone functions. Let Ω be a domain in RN, P a line verifying P ∩ Ω is a nonempty set. A function defined almost everywhere in Ω is said absolutely continuous on the line P if it is continuous on each closed interval of P ∩ Ω. Theorem 4.5 (page 55 [29]). Suppose u ∈ L1loc(Ω) and ∂u ∂xi ∈ Lp(Ω), p ≥ 1. This function changed on a set of measure zero is absolutely continuous on almost all lines parallel to axis xi. Let us denote by [ ∂u ∂xi ] the usual derivative and by ∂u ∂xi the distribution derivative. Then we have almost everywhere [ ∂u ∂xi ] = ∂u ∂xi . Conversely, if u ∈ L1loc(Ω) is absolutely continuous on almost all lines parallel to the axis xi with [ ∂u ∂xi ] ∈ Lp(Ω), then we have ∂u ∂xi = [ ∂u ∂xi ] . By the Lebesgue’s Differentiation Theorem the function P has derivative almost everywhere with respect the radius. VarPj = supr≤r0 0 and limh→0+ ‖ [ dfh dt ] ‖Lp[−1,1]= ∞ for all 1 < p ≤ ∞. Moreover for all nonnegative test function ϕ ∈ C∞0 (−1, 1), we have lim h→0+ ∫1 −1 [f′h(x)] ϕ ′(x)dx = lim h→0+ ∫0 −h ϕ′(x) h dx = lim h→0+ ϕ(−h) − ϕ(0) −h = (−1)(Dδ0)ϕ. (81) Therefore if we define the distribution Λh(ϕ) = ∫1 −1 [f′h(x)] ϕ ′(x)dx then limh→0+ Λh = −(Dδ0) where Dδ0 is the distributional derivative of Dirac’s δ distribution and it is well known that distribution has not weak derivative. The space of functions of pointwise bounded variation admits discontinuous functions and therefore both topologies on the same set C∞(r, R) produce completely different objects in metrics and associated functionals. Our functions The sequence {Pj} ∞ j=1 is bounded in W1,1(A(r, R)) but is it not ensured the strong or weak convergence. The function P has derivative [ ∂P ∂r ] almost everywhere on (r, R) because is a monotone function and moreover we have P = PAC + PC + Pj, where PAC is an absolutely continuous function, Pj is continuous and singular, and PJ is the jump function of P. Theorem 4.6 (page 3 [22]). Let I ⊂ R be an interval and let u : I → R be a monotone function. Then u has as most countable many discontinuity points. Conversely, given a countable set E ⊂ R, there exists a monotone function u : R → R whose set of discontinuity points is exactly E. So by Theorem 4.5 the function u has derivative [u′] but no weak derivative if for example E is dense on I. 4.2 The solid mean value. Despite the difficulty posed by the discontinuity of Green’s identities on the sequence {Pj} ∞ j=1 we can obtain several properties. If a function u is absolutely continuous on the interval (a, b) page 225 [22], then u(x) − 1 b − a ∫b a u(t)dt = 1 b − a [∫x a (t − a)u′(t)dt − ∫b x (b − t)u′(t)dt ] . (82) Using the Lebesgue’s Dominated Convergence Theorem, we obtain a one dimensional solid mean average identity F(x) − 1 b − a ∫b a F(t)dt = lim j→∞ { 1 b − a [∫x a (t − a)F′j(t)dt − ∫b x (b − t)F′j(t)dt ]} . (83) CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 115 Now we recall the extension to Poisson’s equation of the solid mean value for Laplace’s equation in [18] Chapter 4: Let v ∈ C2(Ω) ∩ C0(Ω) satisfy −∆v = f then for any ball B = BR(y), we have v(y) = 1 | B | ∫ B vdx + 1 NωN ∫ B f(x)Θ(r, R)dx, r =| x − y |, (84) where Θ(r, R) = 1 N − 2 ( r2−N − R2−N ) − 1 2RN ( R2 − r2 ) , (85) for N > 2 and Θ(r, R) = log ( R r ) − 1 2 ( 1 − r2 R2 ) , (86) for N = 2, where ωN is the volume of the unit ball in R N. We deduce that if P(y) = limj→∞Pj(y), using the Lebesgue’s Dominated Convergence Theorem then P(y) − 1 | B | ∫ B Pdx = lim j→∞ 1 NωN ∫ B Hj(Pj(x))Θ(r, R)dx, (87) r =| x − y | . (88) This elementary result involves several strong indeterminations. Lemma 4.7 (Lemma 3.1.1 page 113 [45]). Let u ∈ W1,p[B(x0, r)], p ≥ 1, where x0 ∈ R N and r > 1. Let 0 < δ < r. Then ∫ B(x0,r) u(y)dy rN − ∫ B(x0,δ) u(y)dy δN = ∫ B(x0,r) [∇u(y)·(y−x0)]dy NrN − ∫ B(x0,δ) [∇u(y)·(y−x0)]dy NδN . (89) The Lebesgue’s Dominated Convergence Theorem implies ∫ B(x0,r) P(y)dy rN − ∫ B(x0,δ) P(y)dy δN = limj→∞ {∫ B(x0,r) [∇Pj(y)·(y−x0)]dy NrN − ∫ B(x0,δ) [∇Pj(y)·(y−x0)]dy NδN } . (90) 4.3 Newtonian potentials. The theory of Newtonian potentials for distributions with compact support are defined on Ω ⊂ R N, N ≥ 1. We shall make use of the following results: Proposition 4.8 (Proposition 5 page 281 [11]). Let Ω be a regular bounded open set and let u ∈ C2(Ω) ∩ C1n(Ω) with ∆u ∈ L 1(Ω). Then u = u0 + u1 + u2 on Ω, (91) 116 Carlos Cesar Aranda CUBO 17, 2 (2015) where u0, u1, u2 are the Newtonian potentials of the distributions f0, f1, f2 on R N defined by 〈f0, ζ〉 = ∫ Ω ζ∆udx, (92) 〈f1, ζ〉 = ∫ ∂Ω ζ ( − ∂u ∂n ) dγ, (93) 〈f2, ζ〉 = ∫ ∂Ω ∂ζ ∂n udγ. (94) We note that f0, f1, f2 ∈ E ′: f0 is an integrable function on R N with support contained in Ω, f1 is a measure on R N with support contained in ∂Ω, f2 is a distribution of order 1 on R N with support contained in ∂Ω. We say that u1 is the simple layer (respectively double layer) potential defined by the function − ∂u ∂n (respectively u) continuous on ∂Ω. We apply this results to our sequence {Pj} ∞ j=1. Using Proposition 4.8, we have Pj = P0,j + P1,j + P2,j on Ω, (95) where P0,j, P1,j, P2,j are the Newtonian potentials of the distributions f0,j, f1,j, f2,j. Therefore ∆P0,j = f0,j on R N, (96) ∆P1,j = f1,j on R N, (97) ∆P2,j = f2,j on R N. (98) We obtain lim j→∞ 〈f0,j, ζ〉 = −∞, (99) lim j→∞ 〈f1,j, 1〉 = ∫ ∂Ω 1 ( − ∂Pj ∂n ) dγ = ∞, simple layer potentials, (100) lim j→∞ 〈f2,j, ζ〉 = ∫ ∂Ω ∂ζ ∂n ( lim j→∞ Pj ) dγ, double layer potentials. (101) Proposition 4.9 (Proposition 2 page 278 [11]). Let f ∈ E ′ and let u be the Newtonian potential of f. Then for every multi-index α ∈ N ∂αu ∂xα (x) = 〈f, 1〉 + O ( 1 | x |n+|α|−1 ) when | x |→∞. (102) In particular if N ≥ 3, lim |x|→∞ u(x) = 0, (103) if N ≥ 2, lim |x|→∞ ∇u(x) = 0. (104) The last proposition is useful in the description of the sequence of Newtonian potentials {f1,j} ∞ j=1, {f0,j} ∞ j=1, {f1,j} ∞ j=1 and {f2,j} ∞ j=1. CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 117 4.4 The spherical average. For u in 56 with {0} ∈ Ω ⊂ RN, N ≥ 2, let u be the spherical average of u, i.e., u(r) = 1 ωNr N−1 ∫ |x|=r u(x)dγx. (105) With the change of variable x → y, we have u(r) = 1 ωN ∫ |y|=1 u(ry)dγy, (106) and du dr = 1 ωN ∫ |y|=1 ∇u(ry) · ydγy. (107) Hence du dr = 1 ωN ∫ |y|=1 ∂u ∂r (ry)dγy = 1 ωNr N−1 ∫ |x|=r ∂u ∂r (x)dγx, (108) that is du dr = 1 ωNr N−1 ∫ B(0,r) ∆u(x)dx. (109) Therefore from Theorem 2.2 it follows that lim j→∞ dPj dr = −∞. (110) 5 Statement and proof of the main results. Our main Theorem states the concentration of compactness. It can be regarded as a classical counterpart of Helly’s Selection Theorem in the space of functions of bounded point variations (Theorem 2.35 page 59 [22]). Theorem 5.1. Let Ω be a bounded domain in RN, N ≥ 3 and any sequence of functions {uj} ∞ j=1 in C2(Ω) ∩ C0(Ω) satisfying −∆Pj ≥ −∆uj in Ω, (111) Pj ≥ uj on ∂Ω. (112) Then there exist a constant C depending only on the sequence {P}∞j=1, such that uj ≤ C for all j = 1, . . . ,∞. Proof. It is a simple consequence of Theorem 3.3 page 33 in [18]. 118 Carlos Cesar Aranda CUBO 17, 2 (2015) Theorem 5.2 (Strong concentration of compactness for Newtonian potentials). Let Ω be a bounded smooth domain in RN, N > 2. Then there exist a sequence {f1,j} ∞ j=1 ∈ E ′ the space of distributions on RN with compact support such that: (i) f1,j = ∆Pj on Ω in the sense of the distributions. (ii) The sequence of functions {Pj} ∞ j=1 ∈ C ∞(Ω) is non decreasing and bounded. (iii) limj→∞ ∆Pj = −∞ uniformly on Ω. (iv) Pj is the Newtonian potential of f1,j on Ω. (v) The simple layer potential a f1,j ∈ E ′ associated to Pj satisfies limj→∞〈f1,j, 1〉 = ∫ ∂Ω 1 ( − ∂Pj ∂n ) dγ = ∞. (vi) limj→∞ ‖ ∆Pj ‖ ′ Cα(B(x0,R)) = ∞ for all B(x0, R) ⊂⊂ Ω. (vii) Solid mean value property. P(y) − 1 | B | ∫ B Pdx = lim j→∞ (−1) NωN ∫ B ∆Pj(x)Θ(r, R)dx, (113) r =| x − y | . (114) where Θ(r, R) = 1 N − 2 ( r2−N − R2−N ) − 1 2RN ( R2 − r2 ) , (115) for N > 2, where ωN is the volume of the unit ball in R N. (viii) For N > 2 we have ∫ B(x0,r) P(y)dy rN − ∫ B(x0,δ) P(y)dy δN = limj→∞ {∫ B(x0,r) [∇Pj(y)·(y−x0)]dy NrN − ∫ B(x0,δ) [∇Pj(y)·(y−x0)]dy NδN } . (116) (ix) The spherical average Pj satisfy lim j→∞ dPj dr = −∞. (117) Proof. This theorem is a collection of results stated in the A primer analysis section. Theorem 5.3 (Weak concentration of compactness for Newtonian potentials). Let Ω be a bounded smooth domain in RN, N ≥ 1. Then there exist a sequence {f1,j} ∞ j=1 ∈ E ′ the space of distributions on RN with compact support such that: (i) f1,j = ∆Fj on Ω in the sense of the distributions. (ii) The sequence of functions {Fj} ∞ j=1 ∈ C ∞(Ω) is non decreasing and bounded. (iii) limj→∞ ∆Fj = −∞ in measure on Ω. (iv) Fj is the Newtonian potential of f1,j on Ω. (v) limj→∞ ‖ F ‖C1,α(B(x0,R))= ∞ for all B(x0, R) ⊂⊂ Ω. CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 119 (vi) One dimensional mean value property. F(x) − 1 b − a ∫b a F(t)dt = lim j→∞ { 1 b − a [∫x a (t − a)F′j(t)dt − ∫b x (b − t)F′j(t)dt ]} . (118) (vii) Solid mean value property. F(y) − 1 | B | ∫ B Fdx = lim j→∞ (−1) NωN ∫ B ∆Fj(x)Θ(r, R)dx, (119) r =| x − y | . (120) where Θ(r, R) = 1 N − 2 ( r2−N − R2−N ) − 1 2RN ( R2 − r2 ) , (121) for N > 2 and Θ(r, R) = log ( R r ) − 1 2 ( 1 − r2 R2 ) , (122) for N = 2, where ωN is the volume of the unit ball in R N. (viii) For N > 2 ∫ B(x0,r) F(y)dy rN − ∫ B(x0,δ) F(y)dy δN = limj→∞ {∫ B(x0,r) [∇Fj(y)·(y−x0)]dy NrN − ∫ B(x0,δ) [∇Fj(y)·(y−x0)]dy NδN } . (123) Proof. This theorem is a collection of results stated in the A primer analysis section. Received: July 2014. Accepted: April 2015. References [1] C. C. Aranda and T. Godoy. Existence and multiplicity of positive solutions for a singular problem associated to the p-laplacian operator. Electronic Journal of Differential Equations, Vol. 2004(2004), No. 132, pp. 1-15. [2] C. C. Aranda. Bounds and compactness for solutions of second-order elliptic equations. Elec- tronic Journal of Differential Equations, Vol. 2012 (2012), No. 134, pp. 1-5. [3] C. C. Aranda. On the Poisson’s equation −∆u = ∞. CUBO A Mathematical Journal, Vol. 15 N. 01, (151-158), March 2013. [4] M. S. Birman and M. Z. Solomjak. Spectral theory of self-adjoint operators in Hilbert space. D. Reidel Publishing Company, Dordrecht, Holland 1987. 120 Carlos Cesar Aranda CUBO 17, 2 (2015) [5] R. J. Blakely. Potential theory in gravity and magnetic applications. Cambridge University Press 1996. [6] H. Brezis. Analyse fonctionnelle. Théorie et applications. Dunod, Paris 1999. [7] L. M. Burko and A. Ori. Internal Structure of Black Holes and Space Time Singularities: Annals of the Israel Physical Society Vol. 13. Israel Physical Society 1997. [8] S. Chandrasekhar. The mathematical theory of black holes. Oxford University Press 1983. [9] M. G. Crandall, P. H. Rabinowitz and L. Tartar. On a Dirichlet problem with singular non- linearity. Comm. Partial Differential Differential Equations 2 (2) 1977, 193-222. [10] B. Dacorogna. Introduction to the calculus of varitions. Imperial College Press 2004. [11] R. Dautray and J. L. Lions. Mathematical analysis and numerical methods for science and technology. Vol. 1 Physical Origins and classical methods. Springer Verlag 1990. [12] N. Denegaar, J. M. Miller, J. Kennea, N. Gehrels, M. T. Reynolds, and R. Wijnands. The X-ray flaring properties of Srg A* during six years of monitoring with Swift. To appear in Astrophysical Journal. [13] B. Fauser, J. Tolksdorf and E. Zeidler editors. Quantum Gravity Mathematical Models and Experimental Bounds. Birkhauser Verlag 2006. [14] G. B. Folland. Real analysis. John Wiley & Sons Inc. 1999. [15] V. P. Frolov and A. Zelnikov. Introduction to Black Hole Physics. Oxford University Press 2011. [16] G, Ghisellini, R. D. Ceca, M Volonteri, G. Ghirlanda, F. Tavecchio, L. Foschini, G. Tagliaferri, F. Haardt, G. Pareschi and J. Grindlay, . Chasing the heaviest black holes in active galactic nuclei. Monthly Notices of the Royal Astronomical Society 405: 387, (2010). [17] J. Giacomoni and K. Saoudi. Multiplicity of positive solutions for a singular and critical problem. Nonlinear Analysis 71 (2009) 4060-4077. [18] D. Gilbarg and N. S. Trudinger. Elliptic Partial Differential Equations of Second Order. Clas- sics in mathematics reprint of 1998 edition Springer. [19] K. Falconer. Fractal Geometry Mathematical Foundations and Applications. Second Edition, Wiley 2003. [20] P. M. Fitzpatrick. Principles of Celestial Equations. Academic Press 1970. [21] N. S. Koshlyakov, M. M. Smirnov and E. B. Gliner. Differential Equations of Mathematical Physics. North-Holland Publishing Company 1964. CUBO 17, 2 (2015) Spacetime singularity, singular bounds and compactness . . . 121 [22] G. Leoni. A first course in Sobolev spaces. AMS 2009. [23] O. Lpez-Cruz, C. Aorve, M. Birkinshaw, D.M. Worrall, H.J. Ibarra-Medel, W.A. Barkhouse, J.P. Torres-Papaqui, V. Motta (2014). The Central Galaxy of Abell 85-The Largest Galaxy Core Ever Observed. Preprint. [24] V. Maz’ya. Sobolev Spaces With Applications to Elliptic Partial Differential Equations. 2nd. Edition Springer 2011. [25] N. J. McConnell et al. Two ten billion solar mass black holes at the centres of giant elliptical galaxies Nature 480 (7376): 2158, 2011. [26] M. McDonald et al. A Massive, Cooling Flow Induced Starburst in the Core of a Highly Luminous Galaxy Cluster, Nature, Aug 2012. [27] F. Melia. The Galactic Supermassive Black Hole. Princeton University Press 2007. [28] C. H. Nelson. Black Hole Mass, Velocity Dispersion, and the Radio Source in Active Galactic Nuclei. The Astrophysical Journal 544 (2): L91, 2000. [29] J. Nečas. Direct Methods in The Theory of Elliptic Equations. Springer-Verlag 2012. [30] B. O’Neill. The Geometry of Kerr Black Holes. A. K. Peters Massachusetts 1995. [31] R. Penrose. The Road to Reality. Jonathan Cape London 2004. [32] M. H. P. M van Putten. Gravitational Radiation Luminus Black Holes and Gamma-Ray Burst Supernovae. Cambridge University Press 2005. [33] D. Raine and E. Thomas. Black Holes. Imperial College Press, second edition 2009. [34] V. D. Radulescu. Singular Phenomena in Nonlinear Elliptic problems. In Handbook of dif- ferential equations, stationary partial differential equations. Volume 4 edited by M. Chipot Elsevier 2007. [35] D. A. Riechers, F. Walter, C. L. Carilli, and G. F. Lewis. Imaging The Molecular Gas in a z = 3.9 Quasar Host Galaxy at 0.”3 Resolution: A Central, SubKiloparsec Scale Star Formation Reservoir in APM 08279+5255. The Astrophysical Journal 690 (1): 463485, (2009). [36] F. Sauvigny. Partial differential equations 1. Springer 2006. [37] G. Scharf. Finite Quantum Electrodynamics. Springer Verlag 1988. [38] G. Scharf. Quantum Gauge Theories A True Ghost Story. John Wiley & Sons, Inc 2001. [39] D. Shiga. Biggest black hole in the cosmos discovered. NewScientist.com news service (10 January 2008). 122 Carlos Cesar Aranda CUBO 17, 2 (2015) [40] I. S. Sokolnikoff. Tensor Analysis Theory and Applications. New York John Wiley & Sons 1951. [41] H. Stephani, D. Kramer, M. Maccallum, C. Hoenselaers and E. Herlt. Exact Solutions of Einstein’s. Second Edition, Cambridge University Press 2003. [42] M. Struwe. Variational Methods: Applications to Nonlinear Partial Differential Equations and Hamiltonian systems. Springer 2000. [43] J. L. Walsh, A. J. Barth, L. C. Ho and M. Sarzi. The M87 Black Hole Mass from Gas dynamical Models of Space Telescope Imaging Spectrograph Observations. The Astrophysical Journal 770 (2): 86, (June 2013). [44] X. Wu, F. Wang, X. Fan, W. Yi, W. Zuo, F Bian, L. Jiang, I. D. McGreer, R. Wang, J. Yang, Q. Yang, D. Thompson and Y. Beletsky (25 February 2015). An ultraluminous quasar with a twelve-billion-solar-mass black hole at redshift 6.30. Nature 518 [45] W. P. Ziemer. Weakly Differentiable Functions. Springer-Verlag 1989. [46] K. Zubovas, S. Nayakshin and S. Markoff. Sgr A* flares: tidal disruption of asteroids and planets? to appear in Mon. Not. R. Astron. Soc. [47] W. Zuo, X. B. Wu, X. Fan, R. Green, R. Wang and F. Bian. Black Hole Mass Estimates and Rapid Growth of Supermassive Black Holes in Luminous z ∼ 3.5 Quasars. Preprint (2014). Introduction. Preliminaries. The construction of the sequence of functions Fj. A primer analysis. A new kind of ill-posed problem. Green's identities. Integration by parts. A detour with monotone functions. The solid mean value. Newtonian potentials. The spherical average. Statement and proof of the main results.