Random Matrix: From Wigner to Quantum Chaos

Similar documents
Universality of local spectral statistics of random matrices

Universality for random matrices and log-gases

Universality of Random Matrices and Dyson Brownian Motion

From the mesoscopic to microscopic scale in random matrix theory

Homogenization of the Dyson Brownian Motion

Universality of distribution functions in random matrix theory Arno Kuijlaars Katholieke Universiteit Leuven, Belgium

The Matrix Dyson Equation in random matrix theory

Semicircle law on short scales and delocalization for Wigner random matrices

Local semicircle law, Wegner estimate and level repulsion for Wigner random matrices

Markov operators, classical orthogonal polynomial ensembles, and random matrices

Dynamical approach to random matrix theory

Exponential tail inequalities for eigenvalues of random matrices

Universality for random matrices and log-gases Lecture Notes for Current Developments in Mathematics, 2012

Lecture Notes on the Matrix Dyson Equation and its Applications for Random Matrices

Determinantal point processes and random matrix theory in a nutshell

RANDOM MATRIX THEORY AND TOEPLITZ DETERMINANTS

1 Intro to RMT (Gene)

Spectral Universality of Random Matrices

Painlevé Representations for Distribution Functions for Next-Largest, Next-Next-Largest, etc., Eigenvalues of GOE, GUE and GSE

Duality, Statistical Mechanics and Random Matrices. Bielefeld Lectures

Comparison Method in Random Matrix Theory

Random Matrices: Invertibility, Structure, and Applications

Quantum Diffusion and Delocalization for Random Band Matrices

Fluctuations of random tilings and discrete Beta-ensembles

Update on the beta ensembles

arxiv: v3 [math-ph] 21 Jun 2012

DLR equations for the Sineβ process and applications

Eigenvalue variance bounds for Wigner and covariance random matrices

Concentration Inequalities for Random Matrices

Extreme eigenvalue fluctutations for GUE

Universality of sine-kernel for Wigner matrices with a small Gaussian perturbation

Fluctuations of random tilings and discrete Beta-ensembles

Isotropic local laws for random matrices

Bulk Universality for Random Matrices via the local relaxation flow

Chapter 29. Quantum Chaos

Lectures 6 7 : Marchenko-Pastur Law

III. Quantum ergodicity on graphs, perspectives

Stochastic Differential Equations Related to Soft-Edge Scaling Limit

Local Kesten McKay law for random regular graphs

Quantum chaos on graphs

A new type of PT-symmetric random matrix ensembles

A Note on the Central Limit Theorem for the Eigenvalue Counting Function of Wigner and Covariance Matrices

Random matrices: Distribution of the least singular value (via Property Testing)

Numerical analysis and random matrix theory. Tom Trogdon UC Irvine

Random band matrices

A Remark on Hypercontractivity and Tail Inequalities for the Largest Eigenvalues of Random Matrices

arxiv:cond-mat/ v1 [cond-mat.stat-mech] 13 Mar 2003

Random Matrices: Beyond Wigner and Marchenko-Pastur Laws

Statistical Inference and Random Matrices

Quantum Billiards. Martin Sieber (Bristol) Postgraduate Research Conference: Mathematical Billiard and their Applications

Large sample covariance matrices and the T 2 statistic

Random matrices and the Riemann zeros

Eigenvalues and Singular Values of Random Matrices: A Tutorial Introduction

Fluctuations from the Semicircle Law Lecture 4

Free Probability and Random Matrices: from isomorphisms to universality

Spectral properties of random geometric graphs

The circular law. Lewis Memorial Lecture / DIMACS minicourse March 19, Terence Tao (UCLA)

Anderson Localization Looking Forward

Recent results in quantum chaos and its applications to nuclei and particles

DEVIATION INEQUALITIES ON LARGEST EIGENVALUES

Random regular digraphs: singularity and spectrum

Invertibility of random matrices

Random Fermionic Systems

Rigidity of the 3D hierarchical Coulomb gas. Sourav Chatterjee

Random Correlation Matrices, Top Eigenvalue with Heavy Tails and Financial Applications

arxiv: v2 [math.pr] 16 Aug 2014

Beyond the Gaussian universality class

Lecture I: Asymptotics for large GUE random matrices

Applications of random matrix theory to principal component analysis(pca)

Universality for a class of random band matrices. 1 Introduction

TOP EIGENVALUE OF CAUCHY RANDOM MATRICES

NUMERICAL SOLUTION OF DYSON BROWNIAN MOTION AND A SAMPLING SCHEME FOR INVARIANT MATRIX ENSEMBLES

arxiv:math/ v2 [math.pr] 10 Aug 2005

The Tracy-Widom distribution is not infinitely divisible.

Large deviations of the top eigenvalue of random matrices and applications in statistical physics

A Generalization of Wigner s Law

Spectral Statistics of Erdős-Rényi Graphs II: Eigenvalue Spacing and the Extreme Eigenvalues

The Eigenvector Moment Flow and local Quantum Unique Ergodicity

Fredholm determinant with the confluent hypergeometric kernel

arxiv:chao-dyn/ v1 3 Jul 1995

Density of States for Random Band Matrices in d = 2

Fluctuations of the free energy of spherical spin glass

Quantum chaos in composite systems

Random Toeplitz Matrices

Random matrix pencils and level crossings

Quantum Chaos and Nonunitary Dynamics

Central Limit Theorems for linear statistics for Biorthogonal Ensembles

arxiv: v4 [math.pr] 10 Sep 2018

Airy and Pearcey Processes

Convergence of spectral measures and eigenvalue rigidity

arxiv: v2 [cond-mat.dis-nn] 9 Feb 2011

COMPLEX HERMITE POLYNOMIALS: FROM THE SEMI-CIRCULAR LAW TO THE CIRCULAR LAW

NUMERICAL CALCULATION OF RANDOM MATRIX DISTRIBUTIONS AND ORTHOGONAL POLYNOMIALS. Sheehan Olver NA Group, Oxford

The Transition to Chaos

Ihara zeta functions and quantum chaos

Introduction to Theory of Mesoscopic Systems

Random matrices and determinantal processes

Bulk scaling limits, open questions

Valerio Cappellini. References

Emergence of chaotic scattering in ultracold lanthanides.

Transcription:

Random Matrix: From Wigner to Quantum Chaos Horng-Tzer Yau Harvard University Joint work with P. Bourgade, L. Erdős, B. Schlein and J. Yin 1

Perhaps I am now too courageous when I try to guess the distribution of the distances between successive levels (of energies of heavy nuclei). Theoretically, the situation is quite simple if one attacks the problem in a simpleminded fashion.the question is simply what are the distances of the characteristic values of a symmetric matrix with random coefficients. Eugene Wigner, 1956 Nobel prize 1963 2

Classical Example of Universality -Central Limit Theorem Gaussian distribution with variance σ 2 : µ(x) = 1 [ 2πσ 2 exp x2 ] 2σ 2 Suppose X j, j = 1,..., N are independent random variables with mean zero and variance one. Then the probability distribution of X 1 +... + X N N converges to the Gaussian distribution with variance one. Independence assumption can be weaken, but is crucial. 4

Model for independent particles, Poisson statistics: Law of putting K particles independently in an interval of size N so that K/N ρ is fixed. Question: How to model systems with high correlation? Gaussian unitary ensemble: H = (h jk ) 1 j,k N hermitian with h jk = 1 N (x jk + iy jk ) and h jj = 2 N x jj and where x jk, y jk, j > k, and x jj are independent centered Gaussian random variables with variance 1/2. Classical ensembles: Gaussian unitary ensemble (GUE), Gaussian orthogonal ensemble (GOE), Gaussian symplectic ensemble(gse), sample covariance ensembles (Important for statistics applications). 5

E. Wigner (1955): The excitation spectra of heavy nuclei have the same spacing distribution as the eigenvalues of GOE. Experimental data for excitation spectra of heavy nuclei: typical Poisson statistics: Typical random matrix eigenvalues 6

Global Statistics: Density of state ρ(x) follows the Wigner semicircle law for GUE, GOE and GSE. 1 ρ (x) = 4 x 2 2π 2 2 Eigenvalues: λ 1 λ 2...... λ N, λ i+1 λ i 1/N. Moment method: for all k fixed: 1 N TrHk 1 2π 2 2 xk 4 x 2 dx 7

Wigner surmise (1956): the gap distribution πx 2 exp [ πx2 4 Idea: guess the probability law from the eigenvalues of 2 2 matrices. Correct up to a few percentage points when compared with the GOE gap distribution. ] dx 8

9

Probability density of eigenvalues (w.r.t. Lebesgue measure) p N (λ 1,..., λ N ) Correlation function for two eigenvalues: p (2) N (x 1, x 2 ) = Density of states: R N 2 p N(x 1, x 2,..., x N )dx 3...dx N ρ N (x) = p N(x, x 2,..., x N )dx 2...dx N R N 1 Dyson, Gaudin, Mehta [ 60]: local statistics of level correlation lim [ρ N(E)] 2 p (2) ( N N E + a 1 Nρ N (E), E + a ) 2, Nρ N (E) = det { S(a i a j ) } 2 i,j=1, sin πa S(a) = πa (for GUE), E < 2 Spacing distribution can be computed from the correlation functions. 10

Quantum Chaos conjecture Energy levels in quantum billiards or H = + V. Laplace equation: ψ n = λ n ψ n. Related Question: Random Schrodinger equation V is random by P. Anderson 1958 11

Berry-Tabor conjecture (1977):If the billiard trajectories are integrable, the eigenvalue spacings statistics is given by the Poisson distribution e x dx. Bohigas-Giannoni-Schmit conjecture (1984):If the billiard is chaotic, the eigenvalue spacings statistics is given by the GOE. 12

Wireless communica- Other application of random matrices: tions, biology, finance, traffic Wishart (1928), Statistics application: Sample covariance ensembles, matrix of the form A + A, A is the data matrix. Riemann ζ-function: Gap distribution of zeros of ζ function is given by GUE (Montgomery, 1973). Odlyzko data: 13

There are essentially two different behaviors in nature: A: Poisson statistics, for systems with little or no correlations. B: Random matrix statistics: for systems with high correlations. (Edge behavior is different from the bulk.) Fundamental belief of universality of random matrices: The macroscopic statistics depend on the models, but the microscopic statistics are independent of the details of the systems except the symmetries. Classical many-body systems are too hard to solve; Boltzmann s model of classical statistical physics e βh. Quantum many-body systems (and highly correlated systems) are too hard to solve; Wigner s model of random matrices. 14

Unitary ensemble: Hermitian matrices with density P(H)dH e βntr V (H) dh Invariant under H UHU 1 for any unitary U (GUE) p(λ 1,..., λ N ) i<j(λ i λ j ) β e β j NV (λ j) e NβH N H N = 1 i 2 V (λ i) 1 N i<j log λ j λ i classical ensembles β = 1, 2, 4 GOE, GUE, GSE. The distribution of the density of λ j is given by the equilibrium measure ρ V (x)dx which is the minimizer of I(ν) = V (t)dν(t) The semicircle law is the case V (x) = x 2. Suppose the support of ρ V = [A, B]. log t s dν(s)dν(t) 15

From VanderMonde determinant structure (β = 2), p (2) N (x 1, x 2 ) = C det[k N (x i, x j )] 2 i,j=1 K N (x, y) = N ψ N(x)ψ N 1 (y) ψ N (y)ψ N 1 (x) x y ψ N orthogonal polynomials w.r.t. e βv (x) dx.. large N asymptotic of orthogonal polynomials = local eigenvalue statistics indep of V. But density of e.v. depends on V. K N (x + a 1, x + a 2 ) S(a 1 a 2 ), S(a) = sin πa πa (for GUE),

Dyson (1962-76), Gaudin-Mehta (1960- ) via Hermite polynomials and general cases by Deift-Its-Zhou (1997), Bleher-Its (1999), Deift-et al (1999-2009) [Riemann-Hilbert method], Pastur- Schcherbina (1997), Lubinsky (2008)... β = 1, 4:(Widom), Deift-Gioev, Kriecherbauer-Shcherbina: Assuming V analytic with some additional assumptions (convex). From 1960 to 2008, all results depend on explicit formulas. Why need different methods for different cases? non-classical β? what about 16

Generalized Wigner Ensembles H = (h kj ) 1 k,j N, h ji = h ij independent Eh ij = 0, E h ij 2 = σ 2 ij, i σ 2 ij = 1, σ ij N 1/2. N σ2 ij C N If h ij are i.i.d. then it is called Wigner ensembles. c (A) Universality conjecture (Wigner-Dyson-Mehta conjecture): If h ij are independent, then the local eigenvalues statistics are the same as the Gaussian ensembles. No results up to 2008. More generally, eigenvalue gap distributions depends only on symmetry classes and are independent of models. It is the same for both invariant and non-invariant models (but depend on β which is a symmetry parameter). 18

Theorem [Erdős, Knowles, Schlein, Y, Yin ] The bulk universality holds for generalized Wigner ensembles satisfying (A) and E x ij 4+ε C, then for 2 < E < 2, b = N 1+ε, ε > 0 E+b lim N E b de 2b ( p (k) )( F,N p(k) µ,n E + b 1 N,..., E + b ) k N = 0 weakly F µ generalized symmetric matrices GOE generalized hermitian GUE generalized self-dual quaternion GSE real covariance real Gaussian Wishart complex covariance complex Gaussian Wishart Variances can vary in this theorem. later. Comparison with Tao-Vu Generalization to Erdős-Reyni graphs by Knowles-Erdős-Y-Yin (some example of quantum chaos with random data) 19

Theorem[Bourgade-Erdoes-Y 2010-2011] Suppose that V is real analytic and V (x) C (1) Consider the β-ensemble µ β,v with β > 0. Let E (A, B) and E ( 2, 2). We have, as N, ( 1 ϱ V (E) 2p(2) V,N x + α 1 Nϱ(E), x + α ) 2 Nϱ(E) ( 1 ϱ G (E ) 2p(2) G,N x + α 1 Nϱ G (E ), x + α ) 2 Nϱ G (E 0. ) G stands for the Gaussian with V (x) = x 2. 20

Three Steps to the Universality: Non-invariant case Step 1. A priori estimate, Local Semicircle Law. Method: System of self-consistent equations for the Green function, control the error by large deviation methods. Step 2. Universality of Gaussian divisible ensembles H = 1 th 0 + tv, t > 0, H 0 is Wigner V is GUE i.e., the matrix entries have some Gaussian components. General method based on estimating the convergence to local equilibrium of Dyson Brownian motion. Step 3. Approximation by Gaussian divisible ensembles A density argument. Resolvent perturbation expansion to remove the Gaussian part in step 2. 21

Two Steps to the Universality: Invariant case Step 1: A priori estimate (similar to the Step 1 in the noninvariant case) Theorem [Rigidity estimate] For any α > 0 and ɛ > 0, there is a constant c > 0 such that for any N 1 and k αn, (1 α)n, (Valid only in the bulk.) P µβ,v ( λk γ k > N 1+ɛ) ce cn c. Key input: Logarithmic Sobolev inequality, Loop equation. 22

Step 2. Uniqueness of log gas. We order the particles and study the statistics of : {λ j : j I}, I = I L := L + 1, L + K, Relabeling: (λ 1, λ 2,..., λ N ) := (y 1,... y L, x L+1,... x L+K, y L+K+1,... y N ) local equilibrium measure on x with boundary condition y: µ V y (dx). µ V y (dx) exp( NβH y), H y (x) = i I 1 2 V y(x i ) 1 N i,j I i<j log x j x i V y (x) = V (x) 2 N j I log x y j. 23

Goal: µ y in the bulk is independent of y as N, K for good boundary conditions, i.e., Prove the uniqueness of Gibbs measure for good boundary conditions. Tool: Study the relaxation to equilibrium for dynamics with µ G y as the invariant measure. This is a generalization of Dyson Brownian Motion. Uniqueness of Gibbs state is established via a dynamical argument, i.e., estimate of relaxation to equilibrium of a local DBM. Fundamental reason of Universality for both invariant and noninvariant: fast relaxation to local equilibrium of DBM. A priori estimate is needed to provide an estimate of local relaxation time. 24

Date Erdos, Schlein, Y, Yin Tao and Vu 07-08 Local semicircle N 1 scale Prior results scale N 1/2 delocalization of eigenvector 09May Univ. for small N 3/4 Gaussian component v.s. Johansson s O(1) Gaussian 09May 09Jun Univ. for Hermitian matrix smooth dist., first univ. Local semicircle reproved 4 moment thm for e. values Hermitian bulk univ for vanishing 3rd moment. Bernoulli dist. excluded. 09Jun Joint paper, hermitian bulk univ. Removes all extra conditions apart from subexp tail of h ij 09Jul Univ. for symm Wigner case DBM argument appeared 25

Date Erdos, Schlein, Y, Yin Tao and Vu 09Nov Bulk univ for real and complex covariance matrices 09Dec 10Jan 10Mar 10Jul 11Mar 11Apr 11Dec Bulk univ for Wigner matrices with varying variances, except Bernoulli Green fn comparison thm Real Bernoulli solved Eigenvalue rigidity proved Dyson conjecture for optimal relaxation proved univ. for sparse matrices finite 4 + ɛ moments for Wigner Bulk univ. for general β ensemble with convex V Convexity condition removed Bulk univ for complex cov. matrices reproved.

Main Open Problems Universality random Schrödinger Universality of band matrices Random band matrices: H is symmetric with independent but not identically distributed entries with mean zero and variance E W h kl 2 = e k l /W Narrow band, W N = Broad band, W N = localization, Poisson statistics delocalization, GOE statistics Even the Gaussian case is open. d-regular graphs Prove some examples of quantum chaos. 26