Random walks in Beta random environment

Transcription

1 Random walks in Beta random environment Guillaume Barraquand 22 mai 2015 (Based on joint works with Ivan Corwin)

2 Consider the simple random walk X t on Z, starting from 0. We note P ( X t+1 = X t + 1 ) = The Central Limit Theorem says that α α + β, P( X t+1 = X t 1 ) = β α + β. X t t α β α+β σ t = N (0,1). Theorem (Cramér) For α β α+β < x < 1, ) log( P(X t > xt) where I(x) is the Legendre transform of ( λ(z) := log E [ e zx ]) 1 = log t t I(x), ( αe z + βe z α + β ).

3 In random environment? Question What can we say for a random walk in random environment? In this talk, we consider simple random walks on Z in space-time i.i.d. environment: P ( X t+1 = x + 1 X t = x ) = B t,x, P ( X t+1 = x 1 X t = x ) = 1 B t,x, where (B t,x ) t,x are i.i.d. We note P,E (resp. P,E) the measure and expectation with respect to the random walk (resp. the environment) Answer All results from the previous slide still hold, even conditionally on the environment, for almost every realization of the environment.

4 Quenched large deviation principle Theorem (Rassoul-Agha, Seppäläinen and Yilmaz, 2013) Assume that log(b t,x ) have a finite third moment. Then, the limiting moment generating function exists a.s., and 1 ( λ(z) := lim t t log [ ] ) E e zx t, ) log( P(Xt > xt) a.s. t I(x). t where I(x) is the Legendre transform of λ.

5 An exactly solvable model: the Beta RWRE We assume that (B t,x ) follow the Beta(α,β) distribution. P ( B [x,x + dx] ) = x α 1 β 1 Γ(α + β) (1 x) Γ(α)Γ(β) dx. Exactly solvable means that we can exactly compute the law of P(X t > xt) (and more). The annealed law of the Beta RWRE is the simple random walk from the first slide. 0 x B x,t (x,t) 1 B x,t t X t

6 For simplicity, assume α = β = 1. (Uniform case) Theorem (B.-Corwin) The LDP rate function is I(x) = 1 1 x 2. We have the convergence in distribution as t, log( P ( Xt > xt )) + I(x)t σ(x) t 1/3 = L GUE, where L GUE is the GUE Tracy-Widom distribution, and σ(x) 3 = 2I(x)2 1 I(x), under the (technical) hypothesis that x > 4/5. The theorem should extend to the general parameter case α,β and when x covers the full range of large deviation events (i.e. x (0,1)).

7 Fredholm determinant Theorem (B.- Corwin) Let u C \ R +, and t,x with the same parity. Then for any parameters α,β > 0 one has [ ] E e up(x t>x) = det(i + K u ) L 2 (C 0 ) where C 0 is a small positively oriented circle containing 0 but not α β nor 1, and K u : L 2 (C 0 ) L 2 (C 0 ) is defined by its integral kernel K u (w,w ) = 1 1/2+i 2iπ 1/2 i π g(w) ds sin(πs) ( u)s g(w + s) s + w w where ( ) Γ(w) (t x)/2 ( ) Γ(α + β + w) (t+x)/2 g(w) = Γ(w). Γ(α + w) Γ(α + w) ( ) det(i+k u ) L 2 (C 0 ) := n [ ] n... det K u (w i,w j ) n! 2iπ dw 1...dw n. C 0 C 0 i,j=1 n=1

8 Idea of the proof Direct proof 1. Interpret the r.v. P(X t > x) as the partition function Z(t,x) of some polymer model (a particular random average process). 2. Find a recurrence relation for Z(t, x). 3. It yields an evolution equation for t E[Z(t,x 1 )...Z(t,x k )]. 4. Solve the equation using a variant of Bethe ansatz. 5. Take moment generating series. It works! 6. Write it as a Fredholm determinant using ideas from Macdonald processes. Origin Z(t,x) is a limit of observables of the q-hahn TASEP, a Bethe ansatz solvable interacting particle system introduced by Povolotsky. (like the strict-weak polymer, cf Hao Shen s talk)

9 Extreme value theory Fact The order of the maximum of N i.i.d. random variables is the quantile or order 1 1/N. Relation LDP / extreme values Second order corrections to the LDP have an interpretation in terms of second order fluctuations of the maximum of i.i.d. drawings. Corollary (B.-Corwin) Let X (1) t,...,x (N) t be random walks drawn independently in the same environment. Set N = e ct. Then, for α = β = 1, max i=1,...,e ct { X (i) t } t 1 (1 c) 2 d(c) t 1/3 = L GUE, where d(c) is an explicit function (proved under assumption c > 2/5).

10 Zero temperature limit D n,m We define the first passage-time T(n, m) from (0,0) to the half-line D n,m by T(n,m) = min t e π:(0,0) D n,m e π m (0,0) n Passage times For (ξ i,j ) i.i.d. Bernoulli and (E e ) i.i.d. Exponential, t e = { ξ i,j E e if e is horizontal, (1 ξ i,j )E e if e is vertical. Theorem (B.-Corwin) For any κ > a/b and parameters a,b > 0, there exist constants ρ(κ) and τ(κ), s.t. T(n,κn) τ(κ)n ρ(κ)n 1/3 = L GUE.

11 Dynamical construction Alternative description At time 0, only one random walk trajectory (in black). One adds to the percolation cluster portions of branching-coalescing random walks at exponential rate, at each branching point.

12 Outlook We have seen A first exactly solvable model of space-time RWRE. Second order corrections to the LDP converge to L GUE. Limit theorem for the max of N = e ct trajectories. Results propagate to the zero temperature model. Questions KPZ universality for RWRE and random average process, to which extent? Integrability : determinantal structure? Analogue of Schur/Macdonald processes? Link with a random matrix model? Tracy-Widom distribution and extreme value theory...