![Arxiv:2104.11201V1 [Math-Ph] 22 Apr 2021](http://data.docslib.org/img/a4f294e85c62e26dfe695414f67d76c5-1.webp)
arXiv:2104.11201v1 [math-ph] 22 Apr 2021 asdniy hnfo ( from then density; mass fw nrdc h ubeparameter Hubble the introduce we If Herein iem,Fnt iebo-pv lbleitne ubepa Hubble existence, global vs blow-up time Finite dilemma, (3) tensor stress energy-momentum the and eemnsi ucino ie o h aeo completenes a of be sake to the considered For [ is in time. studied parameter of Hubble genera function the stochastic deterministic is, a of that means equation; by approached theoretically is (4) R (1) equations field scretya netit niseprmna quantificat experimental its in uncertainty an currently importance its is Despite cosmology. in quantity fundamental (2) Friedmann-Lemaˆıtre-Robertson-Walker metric the with equation k 00 steGusa uvtr,and curvature, Gaussian the is 2010 phrases. and words Key Date h ubeprmtri esr ftert twihteuniv the which at rate the of measure a is parameter Hubble The = uut2,2021. 26, August : − S:6H5 70,6H0 83F05 60H10, 37H05, 60H05, MSC: G fw utmdf h os nepeaint hto Itˆo, th of foun is that behavior o to asymptotic interpretation is, expected noise that the the solution, Moreover, surely. the modify just of that we blow-up prove if time we finite Herein of sense. a probability Hubbl Stratonovich as the the literature of in quantification cosmology interpreted experimental the the in in proposed found tainty been has model This Abstract. 3 C eoe h etngaiainlconstant, gravitational Newton the denotes C ′′ 10 n rm( from and , ;t hsedw olw[ follow we end this to ]; nti okw td tcatcvrino h remn ac Friedmann the of version stochastic a study we work this In NASOHSI OMLGCLMODEL COSMOLOGICAL STOCHASTIC A IN ds 2 3 tcatcdffrnileutos admdnmclsyste dynamical Random equations, differential Stochastic 1 ), = efind we ) T − IT ALSECDR,CRO MANADA CARLOS ESCUDERO, CARLOS dt = 2 ESSSTRATONOVICH VERSUS O ˆ T + p λ λ R h rsue If pressure. the C H = T µν ˙ µν 2 . − H T ( = 3 8 = t ,[ ], 1. λν 3 ) ( = − := C g C 9 πG Introduction H νλ ′′ 1 p ,[ ], C C 2 − 4 = + dr ′ 3 = − 10 hseuto eoe h remn acceleration Friedmann the becomes equation this , kr g 2 T 1 ) ,ad[ and ], πG 4 µν p δ 2 πG 3 µ 0 − aee,Cosmology. rameter, + R δ − ( ρ ν 0 µν ρ ( r g ρ 1 2 + and , 3 + 2 µν g T dθ 3 + 12 steRcitno o erc( metric for tensor Ricci the is pg ne hsitrrtto too. interpretation this under d p h erctensor, metric the 2 aaee.Isniehsbe tacitly been has noise Its parameter. e .Ltu tr osdrn h Einstein the considering start us Let ]. rprasa osqec fi,there it, of consequence a as perhaps or , nteslto lblyeit almost exists globally solution the en µν h ubeprmtr ncontrast, In parameter. Hubble the f o.I eeec [ reference In ion. hsitrrtto ed oapositive a to leads interpretation this µν ) p + iaino h remn acceleration Friedmann the of lization . T ) eoe eiaino h model the of derivation a offer we s . . 00 osbeepaaino h uncer- the of explanation possible r , 2 = sin tcatcpoesrte hna than rather process stochastic ρ 2 + reepns ssc,i sa is it such, As expands. erse θdϕ p + 2 eeainequation. celeration pg , 00 s tov Stratonovich vs Itˆo ms, C 10 ( = t ,ti uncertainty this ], h cl factor, scale the ) ρ where , 2 ,w have we ), ρ sthe is 2 C. ESCUDERO, C. MANADA
CC′′+2(C′)2+2k C2p On the other hand for µ = ν = 1, we have R11 = 1−kr2 and T11 = pg11 = 1−kr2 , and from (1) we find (5) CC′′ + 2(C′)2 + 2k = 4πG(ρ p)C2. − Now multiply (4) by C2/3 and sum it to (5) to obtain the Friedmann equation 8πG (6) (C′)2 + k = ρC2. 3 From (5) and (6) we deduce that p and ρ only depend on t. If we moreover restrict ourselves to flat models (i.e. k = 0), and assume the proportionality p = ξρ, with ξ a parameter yet to be specified, we get 3 H˙ = H2 (1 + ξ) . −2 If we were to consider a deterministic model, and furthermore, following [10], a model in which 1 ′′ the condition 3p + ρ = 0 is fulfilled, then we had to conclude ξ = 3 . Then C = 0 by (4), and therefore, if C(0) > 0 and C′(0) > 0, then H(0) > 0 and − H H(t)= 0 , 1+ H0t where H0 := H(0) and we have H(t)t 1 when t . → →∞ Some of the properties of this model, such as uniqueness and global existence of the solution, along with its asymptotic behavior, could be expected to hold in a stochastic counterpart of it. It is one of the goals of this work to specify some conditions that guarantee so. Now, again following [10], if we assume that the condition 3p + ρ = 0 has only to be fulfilled ′ 1 ′ on average, we could write ξ = ξ + 3 , where ξ could be any zero mean stochastic process. For simplicity, and as in [10], consider it is a standard Gaussian white noise to arrive at the model 3 2 (7) H˙ = H2 + ξ′ , −2 3 which is a stochastic version of the Friedmann acceleration equation. Of course, this model is not precise, and it could only be made so by means of the introduction of a notion of stochastic integration. This type of matter has been studied in much detail in physics [11]. In [10], the authors tacitly (not explicitly) assume the Stratonovich interpretation of noise, known to be the only one able to preserve the classical chain rule from ordinary differential calculus. However, as we will show herein, this choice is problematic. Precisely, it opens the possibility of finite time blow-ups for the Hubble parameter, phenomena of difficult physical interpretation, while this problematic disappears when the Itˆointerpretation of noise is chosen. The remainder of this article is as follows. In section 2 we give a concise summary on the theory of stochastic dynamical systems that we will subsequently employ. In section 3 we compare the consequences of making precise (7) via either the Stratonovich or the Itˆoprescription; in particular, the Stratonovich interpretation of noise (the one tacitly assumed in [10]) shows finite time blow-ups of the Hubble parameter with positive probability. In contrast, the Itˆoprescription shows global existence almost surely and the expected asymptotic behavior. Finally, in section 4, we draw our main conclusions.
2. Stochastic dynamical systems In this section we provide a summary on the mathematical framework we will use, that of random and stochastic dynamical systems. To this end we follow [1], [2], and [5]. We start introducing the definition of random dynamical system. From now on we denote R := [0, ). + ∞ Definition 2.1. Let (Ω, , P) be a probability space and θ = θt t∈R a map θ : R Ω Ω, satisfying F { } × 7→ ITOˆ VS STRATONOVICH IN COSMOLOGY 3
(1) θ (ω)= ω, ω Ω. 0 ∀ ∈ (2) θt τ (ω)= θt θτ (ω), t,τ R . + ◦ ∀ ∈ + (3) (t,ω) θt(ω) is measurable in ( (R) , ), and θtP = P for all t R (i.e. P(θt(ω)) = P(ω)).7→ B × F F ∈ Then (Ω, , P; θ) is called a driving dynamical system and θ is called a driver. F Definition 2.2. Let (Ω, , P) be a probability space. A random dynamical system (RDS) over Rd F d d (d = 1, 2, 3, ) consists on a pair (θ, ϕ) with θ a driver and a cocycle mapping ϕ : R+ Ω R R satisfying: · · · × × 7→ (1) ϕ(0,ω,x)= x, ω Ω,x Rd. ∀ ∈ ∈ d (2) ϕ(t + s,ω,x)= ϕ(t,θs(ω), ϕ(s,ω,x)), t,s R+,ω Ω,x R . (3) (t,ω,x) ϕ(t,ω,x) is measurable. ∀ ∈ ∈ ∈ (4) x ϕ(t,ω,x7→ ) is continuous t R a.s. 7→ ∀ ∈ + Henceforth we fix d = 1 and take the sample space Ω = C0(R, R), the space of continuous functions that depend on a real variable, take values on the reals, and vanish at the origin; let be its Borel sigma-field. Moreover we take P to be the Wiener measure, θ the Wiener shift F
θtω( )= ω(t + ) ω(t), · · − and ω(t) two-sided Brownian motion, with the usual definition (two independent Wiener processes, respectively with positive and negative time, glued at the origin). In this setting consider the one dimensional linear stochastic differential equation (SDE)
(8) dz(t)= λz(t)dt + dWt, − for some λ> 0 and Wt(ω)= ω(t). Note that this is an Orstein-Uhlenbeck SDE. Then, the following result holds [2].
Proposition 2.3. Let λ> 0. Then there exists a θt t R-invariant (in the sense of (3) in Defini- { } ∈ tion 2.1) subset Ω of Ω= C0(R, R) with unit measure such that for all ω Ω: ω(∈t) F ∈ (1) limt→±∞ t = 0. (2) The random variable given by z∗(ω) := λ 0 eλτ ω(τ)dτ is well defined, and the map − −∞ 0 0 ∗ λτ R λτ (t,ω) z (θtω)= λ e θt(ω(τ))dτ = λ e ω(t + τ)dτ + ω(t), 7→ − − Z−∞ Z−∞ is a stationary solution of (8), with continuous trajectories satisfying: ∗ |z (θtω)| (a) limt→±∞ |t| = 0, 1 t ∗ (b) limt→±∞ t 0 z (θτ ω)dτ = 0, 1 t ∗ (c) limt z (θτ ω) dτ < . →±∞ t R0 | | ∞ Proof. We just give a sketchR of the proof and refer the reader to the references for more details. (1) follows from the law of iterated logarithms (see for instance [8]): ω(t) ω(t) P lim sup = 1 = P lim sup = 1 = 1. t √2t log log t t →∞ ( →∞ 2t log log 1/t ) t −λ(t−τ) (2) is a consequence of the fact that the process (ω,tp) e dWτ (ω) is a solution 7→ −∞ to (8) with z(0) = 0 and of integration by parts. The stationarity of this solution follows R from the invariance of the Wiener measure dWt with respect to the flow θt t. The other statements follow from the ergodic theorem and the Burkholder inequality{ (see} for instance [8]). At this point, we will employ the theory of conjugated RDSs to SDEs with additive noise. To that end we will need the following result [2]. 4 C. ESCUDERO, C. MANADA
Proposition 2.4. Let (θ, ϕ) be a RDS on R. Suppose that the mapping T :Ω R R possesses the following properties: × 7→ (1) The mapping T (ω, ) is a homeomorphism over R for every ω Ω. (2) The mappings T ( ,x· ) and T −1( ,x) are measurable for every x∈ R. · · −1 ∈ Then, the mapping (t,ω,x) φ(t,ω,x) := T (θt(ω), ϕ(t,ω,T (ω,x))) defines a conjugated random dynamical system. 7→ Proof. First, note that −1 −1 T (θ0(ω), ϕ(0,ω,T (ω,x))) = T (ω, T (ω,x)) = x. Then, let us see that the property of cocycles is fulfilled: −1 φ(t + τ,ω,x)= T (θt+τ (ω), ϕ(t + τ,ω,T (ω,x))) −1 = T (θt(θτ (ω)), ϕ(t,θτ (ω), ϕ(τ,ω,T (ω,x)))
= φ(t,θτ (ω), φ(τ,ω,x)).
Now, following [2], we apply this theory to a SDE with additive noise of the form
(9) dx(t,ω)= f(x)dt + dWt, x(0,ω)= x R, 0 ∈ where we suppose that f is continuously differentiable with bounded derivative, so that a unique strong solution to (9) exists almost surely (see [13] for milder conditions for existence and unique- ness). Subsequently perform the change of variables ∗ (10) y(t,ω)= x(t,ω) z (θt(ω)), − where z∗ is defined as in Proposition 2.3. Then it holds that y(0,ω)= x(0,ω) z∗(ω)= x z∗(ω). − 0 − So we obtain dy(t,ω)= dx(t,ω) dz∗ − ∗ = f(x(t,ω))dt + dWt(ω) ( λz (θt(ω))dt + dWt(ω)) ∗ − − = (f(x(t,ω)) + λz (θt(ω)))dt ∗ ∗ = (f(y(t,ω)+ z (θt(ω))) + λz (θt(ω)))dt, and thus dy (11) (t,ω)= f(y(t,ω)+ z∗(θ (ω))) + λz∗(θ (ω)). dt t t Equation (11) is a random differential equation, and f is regular enough so that it generates a RDS ϕ [1]; then from Proposition 2.4 we obtain a conjugated RDS for equation (9). Indeed, in this case, the map T :Ω R R is given by × 7→ T (ω,x) := x z∗(ω), − which satisfies the hypotheses of Proposition 2.4. Therefore, −1 φ(t,ω,x0)= T (θt(ω), ϕ(t,ω,T (ω,x0))) −1 ∗ = T (θt(ω), ϕ(t,ω,x z (ω))) − = y(t,ω,x z∗(ω)) + z∗(ω) 0 − = x(t,ω,x0), is our conjugated RDS for the SDE (9). ITOˆ VS STRATONOVICH IN COSMOLOGY 5
3. Blow-up versus global existence 3.1. Stratonovich equation. Equation (7) can be written in the precise manner:
2 3 2 (12) dH = H dt H dWt, − − 2 ◦ if we assume the stochastic integral of Stratonovich. This is tacitly assumed by the authors in [10], who employ the usual Leibnitz-Newton calculus rules with this equation, known to be valid only in the case of Stratonovich. Due to the regularity of its terms, this equation will possess a unique strong solution as long as it remains bounded [13]. If we further follow reference [10] by formally 1 performing the change of variables x := H , then (12) transforms into 3 dx = dt + dWt 2 ◦ 3 (13) = dt + dW , 2 t where we have just employed the Stratonovich (or Leibnitz-Newton) calculus rules on the first line and the fact that an additive noise equation has not to be interpreted (at least, in other sense rather than the one provided by the Wiener integral). For this stochastic differential equation we have the following result:
Lemma 3.1. The SDE (13) subject to an initial condition x(0) = 1/H0 > 0 possesses a unique solution which is both strong and global and moreover fulfils x(t) lim = 1 a.s. t→∞ t Proof. Existence and uniqueness of a strong and global solution is direct from the standard theory [13]. To find the asymptotic behavior multiply the SDE by 2/3 to get 2x 2 d = dt + dW , 3 3 t which is of the form (9) with f = 2/3. Now we will apply to this equation the developments in the last part of the previous section; first, doing the change of variables (10), which in the present case reads 2 y = x z∗, 3 − leads to (11), which now takes the form 2 t y(t)= t + λ z∗dτ. 3 Z0 It defines a random dynamical system [1] and, by Proposition 2.4, it gives rise to a conjugated random dynamical system for equation (13). Subsequently applying Proposition 2.3 yields y(t) 2 lim = a.s. t→∞ t 3 The statement follows from undoing the change of variables to recover x and applying Proposition 2.3 once more. From this result one would be tempted to conclude Ht 1 a.s. when t , → →∞ which is in fact the behavior found for the deterministic model in the Introduction. However, this conclusion would go through undoing the change of variables H = 1/x, which is possible only if x(t) does not cross (or even touch) zero. Unfortunately, this is not the case as we show in the next result. 6 C. ESCUDERO, C. MANADA
Lemma 3.2. The stochastic process x(t) that solves SDE (13) subject to x(0) = 1/H0 > 0 crosses the origin in finite time with a probability 8 exp . −9H 0 3 1 Proof. Note that we have the explicit formula x(t) = t + 2 Wt + H0 . Therefore the statement can be seen as a particular case of the first passage time problem for Brownian motion with drift solved in [8] by means of the Girsanov theorem. Consider the stopping time 2 Ta := inf t 0 : t + Wt = a , ≥ 3 for any a = 0. It possesses the density [8]: 6 2 2 a (a 3 t) P(Ta dt)= | | exp − dt, t> 0, ∈ √2πt3 − 2t ! and therefore 2 2 P(Ta < ) = exp a a . ∞ 3 − 3 2 As in our case a = 3H0 we conclude − 8 P 2 T− < = exp . 3H0 ∞ −9H 0
Since exp 8 (0, 1) for H (0, ), there is a positive probability that H(t) blows up in − 9H0 ∈ 0 ∈ ∞ finite time. So, in summary, we obtain the following dual behavior for the solution to SDE (12): 8 H(t)t 1 as t with probability 1 exp , → →∞ − −9H 0 8 2 H(t) as t T− with probability exp ; → ∞ → 3H0 −9H 0 i.e., the solution blows up in finite time with a positive probability exp 8 , so the solution − 9H0 ceases to exist in finite time and therefore the asymptotic behavior is never achieved for these samples. Consequently, we may conclude that this model presents difficulties in its application to cosmology. 3.2. Itˆoequation. Equation (7) can also be written in the different precise manner:
2 3 2 (14) dH = H dt H dWt, − − 2 if we assume the stochastic integral of Itˆo. As for the Stratonovich model, the terms are so regular that this equation will possess a unique strong solution as long as it remains bounded [13]. If in this case we change variables again x = 1/H then, by the Itˆochain rule, we find 9 3 (15) dx = 1+ dt + dW . 4x 2 t Note that this equation, again due to the regularity of its terms, will possess a unique strong solution as long as it remains bounded and bounded away from zero [13]. We will show that it is well behaved in the sense we are looking for. In order to analyze it we need to introduce the following family of stochastic processes. 1 n Definition 3.3. For any integer n 2, let W˜ t = (W , ,W ),t 0, be a n-dimensional Wiener ≥ t · · · t ≥ process. Then the process Rt = Wt ,t 0, is called a Bessel process of dimension n. | | ≥ ITOˆ VS STRATONOVICH IN COSMOLOGY 7
Remark 3.4. A Bessel process of dimension n satisfies the SDE n 1 dRt = − dt + dWt,R0 = 0, 2Rt where Wt is the standard one-dimensional Wiener process, see [8].
Theorem 3.5. The SDE (15) subject to an initial condition x(0) = 1/H0 > 0 possesses a unique solution which is both strong and global and moreover fulfils x(t) lim = 1 a.s. t→∞ t Furthermore, x(t) > 0 for all t 0 a.s. ≥ 2 Proof. First of all change variables z = 3 x to obtain 2 1 dz = dt + dt + dW . 3 z t Note that the SDE 1 (16) dy = dt + dW , y t subject to y(0) = 0 possesses as unique solution the Bessel process y(t) = Rt with dimension n = 3, see Definition 3.3 and Remark 3.4. Then y(t) > 0 for all t > 0 almost surely, see [8], and consequently z(t) > 0 for all t > 0 almost surely provided that, for the same realization of Wt, z(t) y(t) for all t> 0 almost surely. Note that this should be true at least in some interval [0, t], for some≥ positive stopping time t = t(ω), given the continuity of the trajectories of both z(t) and 2 y(t), and the fact that z(0) = 3H0 > 0 = y(0). Suppose that, on the contrary, there exists a set Ω˜ Ω of positive measure such that for all ω Ω˜ there is a random time tω > t > 0 such that ⊂ ∈ z(tω,ω)
y(t) W 1 2 W 2 2 W 3 2 = t + t + t 0 a.s. t s t t t → as t by Definition 3.3. Finally, from (17) we conclude →∞ z(t) 2 lim = a.s., t→∞ t 3 3 and the statement follows from undoing the change of variables x = 2 z. From Theorem 3.5 we conclude that the change of variables x(t) = 1/H(t) is well defined for all t 0 almost surely, so we can invert it to conclude ≥ H(t)t 1 as t with probability 1. → →∞ 4. Conclusions In this work we have studied a stochastic version of the Friedmann acceleration equation. This SDE was introduced in [10] as a possible theoretical explanation of the uncertainty observed in the experimental quantification of the Hubble parameter. In the original model, the Stratonovich interpretation of noise was tacitly chosen, as the authors analyzed this equation using the standard chain rule from ordinary calculus. Herein we have shown that this Stratonovich equation presents finite time blow-ups with a positive probability; in particular this probability is 8 exp . −9H 0 On the other hand, we have shown that if the interpretation is changed to that of Itˆo, then the solution exists globally in time almost surely and moreover we have the asymptotic behavior H(t)t 1 as t with probability 1, → →∞ or, in other words, we have for almost every sample the universal decay rate H(t) 1/t, with universal amplitude, for long times. This coincides with both the deterministic behavior≈ (which of course takes place surely) and the long time dynamics of the Stratonovich SDE in those cases in which the solution does not blow up in finite time. Yet another advantage of the Itˆointerpretation is that it admits the explicit bounds 1 1 3 1 H(t) 3 1 3 y(t)+ t + ≤ ≤ max Wt + t + , y(t) 2 H0 { 2 H0 2 } for all times t 0 almost surely, where y(t) is the Bessel process with dimension n = 3 that solves equation (16);≥ this is a direct consequence of the proof of Theorem 3.5. Although it seems that the Stratonovich interpretation is more accepted for applications in physics, we have herein highlighted an example in the field of cosmology in which the Itˆointerpre- tation looks more advantageous. This is not the only case, as other problems in statistical mechanics are somehow similar in this respect [4, 6, 7]. In these previous cases, however, the uniqueness of solution was lost in the Stratonovich equation, while preserved in the Itˆoone. In the present case, it is the (global in time) existence of solution what is lost for Stratonovich and preserved for Itˆo. Another difference between these previous statistical mechanical examples and the current cosmo- logical one is that the deficiency of the Stratonovich equation was of a more fundamental character there, as there is a Stratonovich equation equivalent to equation (14) due to the smoothness of its diffusion; this did not happen in the other cases because the equations under study there did not ITOˆ VS STRATONOVICH IN COSMOLOGY 9 present regular enough diffusion terms. Anyway, in the problem we have analyzed herein, it seems much simpler to employ the Itˆointerpretation rather than try to circumvent it via an equivalent Stratonovich formulation. In summary, we have illustrated one example in cosmology for which Itˆo versus Stratonovich means global existence versus finite time blow-up. Acknowledgments
This work has been partially supported by the Government of Spain (Ministerio de Ciencia, Innovaci´on y Universidades) through Project PGC2018-097704-B-I00. References
[1] L. Arnold, Random Dynamical Systems, Springer, Berlin, 1998. [2] T. Caraballo and X. Han, Applied Nonautonomous and Random Dynamical Systems, Springer Nature, Cham, 2016. [3] F. Chamizo, Seminario 2001: una odisea en el espacio-tiempo, available at: http://matematicas.uam.es/∼fernando.chamizo/libreria/fich/APseminario02.pdf. [4] A.´ Correales and C. Escudero, Itˆovs Stratonovich in the presence of absorbing states, J. Math. Phys. 60, 123301 (2019). [5] H. Crauel and F. Flandoli, Attractors for random dynamical systems, Probab. Theory Relat. Fields 100, 365–393 (1994). [6] C. Escudero, Kinetic energy of the Langevin particle, Stud. Appl. Math. 145, 719–738 (2020). [7] C. Escudero, Fluctuation-dissipation relation, Maxwell-Boltzmann statistics, equipartition theorem, and stochas- tic calculus, arXiv:2101.02543. [8] I. Karatzas and S. E. Shreve, Brownian Motion and Stochastic Calculus, Springer, New York, 1998. [9] J. Foster and J. D. Nightingale, A Short Course in General Relativity, Springer, New York, 2006. [10] M. John, C. Sivakumar, and B. Joseph, Classical stochastic approach to cosmology revisited, Pramana - Journal of Physics 60, 1–10 (2003). [11] W. Horsthemke and R. Lefever, Noise Induced Transitions, Springer, New York, 2006. [12] C. W. Misner, K. S. Thorne, and J. A. Wheeler, Gravitation, Princeton University Press, Princeton, 2017. [13] B. Øksendal, Stochastic Differential Equations: An Introduction with Applications, Springer, Berlin, 2003.
Carlos Escudero Departamento de Matem´aticas Fundamentales Universidad Nacional de Educaci´on a Distancia [email protected]
Carlos Manada Facultad de Ciencias Universidad Nacional de Educaci´on a Distancia [email protected]