Progress in the method of Ghosts and Shadows for Beta Ensembles Alan Edelman (MIT) Alex Dubbs (MIT) and Plamen Koev (SJS) Oct 8, 2012 1/47
Wishart Matrices (arbitrary covariance) G=mxn matrix of Gaussians Σ=mxn semidefinite matrix G G Σ is similar to A=Σ ½ G GΣ ½ For β=1,2,4, the joint eigenvalue density of A has a formula: Known for β=2 in some circles as Harish Chandra Itzykson Zuber 2/47
Main Purpose of this talk Eigenvalue density of G G Σ ( similar to A=Σ ½ G GΣ ½ ) Present an algorithm for sampling from this density Show how the method of Ghosts and Shadows can be used to derive this algorithm Further evidence that β=1,2,4 need not be special 3/47
Eigenvalues of GOE (β=1) Naïve Way: MATLAB : A=randn(n); S=(A+A )/sqrt(2*n);eig(s) R: A=matrix(rnorm(n*n),ncol=n);S=(a+t(a))/sqrt(2*n);eigen(S,symmetric=T,only.values=T)$values; Mathematica: A=RandomArray[NormalDistribution[],{n,n}];S=(A+Transpose[A])/Sqrt[n];Eigenvalues[s] 4/47
Eigenvalues of GOE/GUE/GSE and beyond Real Sym. Tridiagonal: β=1:real, β=2:complex, β=4:quaternion Diagonals N(0,2), Off diagonals chi random variables General β>0 Computations significantly faster 5/47
Introduction to Ghosts G 1 is a standard normal N(0,1) G 2 is a complex normal (G 1 +ig 1 ) G 4 is a quaternion normal (G 1 +ig 1 +jg 1 +kg 1 ) G β (β>0) seems to often work just fine Ghost Gaussian 6/47
Chi squared 2 Defn: χ β is the sum of β iid squares of standard normals if β=1,2, Generalizes for non integer β as the gamma function interpolates factorial χ β is the sqrt of the sum of squares (which generalizes) (wikipedia chi distriubtion) G 1 is χ 1, G 2 is χ 2, G 4 is χ 4 So why not G β is χ β? I call χ β the shadow of G β 7/47
Linear Algebra Example Given a ghost Gaussian, G β, G β is a real chibeta variable χ β variable. Gram Schmidt: G β G β G β χ 3β G β G β χ 3β G β G β χ 3β G β G β G β G β G β G β G β χ 2β G β H 3 H 2 H 1 χ 2β G β G β G β G β G β G β G β Χ β or G β G β G β G β G β G β G β G β G β = [Q] * χ 3β G β G β χ 2β G β Χ β 8/47
Linear Algebra Example Given a ghost Gaussian, G β, G β is a real chibeta variable χ β variable. Gram Schmidt: G β G β G β χ 3β G β G β χ 3β G β G β χ 3β G β G β G β G β G β G β G β χ 2β G β H 3 H 2 H 1 χ 2β G β G β G β G β G β G β G β Χ β or G β G β G β G β G β G β G β G β G β = [Q] * χ 3β G β G β χ 2β G β Χ β Q has β Haar Measure! We have computed moments! 9/47
Tridiagonalizing Example G β G β G β G β Symmetric Part of G β G β G β G β G β G β G β G β G β G β G β G β G β (Silverstein, Trotter, etc) 10/47
Histogram without Histogramming: Sturm Sequences Count #eigs < 0.5: Count sign changes in Det( (A 0.5*I)[1:k,1:k] ) Count #eigs in [x,x+h] Take difference in number of sign changes at x+h and x Mentioned in Dumitriu and E 2006, Used theoretically in Albrecht, Chan, and E 2008 11/47
A good computational trick is a good theoretical trick! 12/47
Stochastic Differential Eigen Equations Tridiagonal Models Suggest SDE s with Brownian motion as infinite limit: E: 2002 E and Sutton 2005, 2007 Brian Rider and (Ramirez, Cambronero, Virag, etc.) (Lots of beautiful results!) (Not today s talk) 13/47
Tracy Widom Computations Eigenvalues without the whole matrix! Never construct the entire tridiagonal matrix! Just say the upper 10n 1/3 by 10n 1/3 Compute largest eigenvalue of that, perhaps using Lanczos with shift and invert strategy! Can compute for n amazingly large!!!!! And any beta. E: 2003 Persson and E: 2005 14/47
Other Computational Results Infinite Random Matrix Theory The Free Probability Calculator (Raj) Finite Random Matrix Theory MOPS (Ioana Dumitriu) (β: orthogonal polynomials) Hypergeometrics of Matrix Argument (Plamen Koev) (β: distribution functions for finite stats such as the finite Tracy Widom laws for Laguerre) Good stuff, but not today. 15/47
Scary Ideas in Mathematics Zero Negative Radical Irrational Imaginary Ghosts: Something like a commutative algebra of random variables that generalizes random Reals, Complexes, and Quaternions and inspires theoretical results and numerical computation 16/47
Did you say commutative?? Quaternions don t commute. Yes but random quaternions do! If x and y are G 4 then x*y and y*x are identically distributed! 17/47
Hermite: RMT Densities c λ i λ j β e λ i 2 /2 (Gaussian Ensemble) Laguerre: Jacobi: c λ i λ j β λ m i e λ i (Wishart Matrices) c λ i λ j β λ m 1 i (1 λi ) m 2 (Manova Matrices) Fourier: c λ i λ j β (on the complex unit circle) Jack Polynomials Traditional Story: Count the real parameters β=1,2,4 for real, complex, quaternion Applications: β=1: All of Multivariate Statistics β=2:tons, β=4: Almost nobody cares Dyson 1962 Threefold Way Three Associative Division Rings 18/47
β Ghosts β=1 has one real Gaussian (G) β=2 has two real Gaussians (G+iG) β=4 has four real Gaussians (G+iG+jG+kG) β 1 has one real part and (β 1) Ghost parts 19/47
Introductory Theory There is an advanced theory emerging A β ghost is a spherically symmetric random variable defined on R β A shadow is a derived real or complex quantity 20/47
Wishart Matrices (arbitrary covariance) G=mxn matrix of Gaussians Σ=mxn semidefinite matrix G G Σ is similar to A=Σ ½ G GΣ ½ For β=1,2,4, the joint eigenvalue density of A has a formula: 21/47
Joint Eigenvalue density of G G Σ The 0 F 0 function is a hypergeometric function of two matrix arguments that depends only on the eigenvalues of the matrices. Formulas and software exist. 22/47
Laguerre: Generalization of Laguerre Versus Wishart: 23/47
General β? The joint density: is a probability density for all β>0. Goals: Algorithm for sampling from this density Get a feel for the density s ghost meaning 24/47
Main Result An algorithm derived from ghosts that samples eigenvalues A MATLAB implementation that is consistent with other beta ized formulas Largest Eigenvalue Smallest Eigenvalue 25/47
Real quantity Working with Ghosts 26/47
More practice with Ghosts 27/47
Bidiagonalizing Σ=I Z Z has the Σ=I density giving a special case of 28/47
The Algorithm for Z=GΣ ½ 29/47
The Algorithm for Z=GΣ ½ 30/47
Removing U and V 31/47
Algorithm cont. 32/47
Completion of Recursion 33/47
Numerical Experiments Largest Eigenvalue Analytic Formula for largest eigenvalue dist E and Koev: software to compute 34/47
1 m3n3beta5.000m150.stag.a.fig 0.9 0.8 0.7 0.6 F(x) 0.5 0.4 0.3 0.2 0.1 0 0 20 40 60 80 100 120 x 35 35/47
1 m4n4beta2.500m130.stag.a.fig 0.9 0.8 0.7 0.6 F(x) 0.5 0.4 0.3 0.2 0.1 0 0 10 20 30 40 50 60 70 80 90 100 x 36/47
1 m5n4beta0.750m120.1234.a.fig 0.9 0.8 0.7 0.6 F(x) 0.5 0.4 0.3 0.2 0.1 0 0 20 40 60 80 100 120 140 x 37 37/47
Smallest Eigenvalue as Well The cdf of the smallest eigenvalue, 38/47
Cdf s of smallest m5n4beta3.000.stag.a.least.fig eigenvalue 1 0.9 0.8 0.7 0.6 F(x) 0.5 0.4 0.3 0.2 0.1 0 0 2 4 6 8 10 12 14 16 x 39 39/47
Goals Continuum of Haar Measures generalizing orthogonal, unitary, symplectic New Definition of Jack Polynomials generalizing the zonals Computations! E.g. Moments of the Haar Measures Place finite random matrix theory β into same framework as infinite random matrix theory: specifically β as a knob to turn down the randomness, e.g. Airy Kernel d 2 /dx 2 +x+(2/β ½ )dw White Noise 40/47
Formally Let S n =2π/Γ(n/2)= surface area of sphere Defined at any n= β>0. A β ghost x is formally defined by a function f x (r) such that f x (r) r β 1 S β 1 dr=1. r=0 Note: For β integer, the x can be realized as a random spherically symmetric variable in β dimensions Example: A β normal ghost is defined by f(r)=(2π) β/2 e r2 /2 Example: Zero is defined with constant*δ(r). Can we do algebra? Can we do linear algebra? Can we add? Can we multiply? 41/47
A few more operations ΙΙxΙΙ is a real random variable whose density is given by f x (r) (x+x )/2 is real random variable given by multiplying ΙΙxΙΙ by a beta distributed random variable representing a coordinate on the sphere 42/47
Addition of Independent Ghosts: Addition returns a spherically symmetric object Have an integral formula Prefer: Add the real part, imaginary part completed to keep spherical symmetry 43/47
Multiplication of Independent Ghosts Just multiply ΙΙzΙΙ s and plug in spherical symmetry Multiplication is commutative (Important Example: Quaternions don t commute, but spherically symmetric random variables do!) 44/47
Understanding λ i λ j β Define volume element (dx)^ by (r dx)^=r β (dx)^ (β dim volume, like fractals, but don t really see any fractal theory here) Jacobians: A=QΛQ (Sym Eigendecomposition) Q daq=dλ+(q dq)λ Λ(Q dq) (da)^=(q daq)^= diagonal ^ strictly upper diagonal = dλ i =(dλ)^ off diag = ((Q dq) ij (λ i λ j ))^=(Q dq)^ λ i λ j β 45/47
Haar Measure β=1: E Q (trace(aqbq ) k )= C κ (A)C κ (B)/C κ (I) Forward Method: Suppose you know C κ s a priori. (Jack Polynomials!) Let A and B be diagonal indeterminants (Think Generating Functions) Then can formally obtain moments of Q: Example: E( q 11 2 q 22 2 ) = (n+α 1)/(n(n 1)(n+α)) α:=2/ β Can Gram Schmidt the ghosts. Same answers coming up! 46/47
Further Uses of Ghosts Multivariate Orthogonal Polynomials Tracy Widom Laws Largest Eigenvalues/Smallest Eigenvalues Expect lots of uses to be discovered 47/47