On asymptotics of large Haar distributed unitary matrices

Let U _n be an n × n Haar unitary matrix. In this paper, the asymptotic normality and independence of Tr U _n , Tr U _n ² ,..., Tr U _n ^k are shown by using elementary methods. More generally, it is shown that the renormalized truncated Haar unitaries converge to a Gaussian random matrix in distribution. This revised version was published online in August 2006 with corrections to the Cover Date.     

Maxima of entries of Haar distributed matrices

Let _n=(_ij) be an n×n random matrix such that its distribution is the normalized Haar measure on the orthogonal group O(n). Let also W_n:=max_1i,jn|_ij|. We obtain the limiting distribution and a strong limit theorem on W_n. A tool has been developed to prove these results. It says that up to n/( log n)² columns of _n can be approximated simultaneously by those of some Y_n=(y_ij) in which y_ij are independent standard normals. Similar results are derived also for the unitary group U(n), the special orthogonal group SO(n), and the special unitary group SU(n).Mathematics Subject Classification (2000):15A52, 60B10, 60B15, 60F10     

Eigenvalue distributions of random unitary matrices

 Let U be an n × n random matrix chosen from Haar measure on the unitary group. For a fixed arc of the unit circle, let X be the number of eigenvalues of M which lie in the specified arc. We study this random variable as the dimension n grows, using the connection between Toeplitz matrices and random unitary matrices, and show that (X -E [X])/(\Var (X))^1/2 is asymptotically normally distributed. In addition, we show that for several fixed arcs I ₁ , ..., I _m , the corresponding random variables are jointly normal in the large n limit. Received: 15 November 2000 / Revised version: 27 September 2001 / Published online: 17 May 2002     

Codimension-one minimal projections onto Haar subspaces

Let H_n be an n-dimensional Haar subspace of and let H_n−1 be a Haar subspace of H_n of dimension n−1. In this note we show (Theorem 6) that if the norm of a minimal projection from H_n onto H_n−1 is greater than 1, then this projection is an interpolating projection. This is a surprising result in comparison with Cheney and Morris (J. Reine Angew. Math. 270 (1974) 61 (see also (Lecture Notes in Mathematics, Vol. 1449, Springer, Berlin, Heilderberg, New York, 1990, Corollary III.2.12, p. 104) which shows that there is no interpolating minimal projection from C[a,b] onto the space of polynomials of degree n, (n2). Moreover, this minimal projection is unique (Theorem 9). In particular, Theorem 6 holds for polynomial spaces, generalizing a result of Prophet [(J. Approx. Theory 85 (1996) 27), Theorem 2.1].     

The Entries of Haar-Invariant Matrices from the Classical Compact Groups

Let Γ _n =(γ _ij ) _n×n be a random matrix with the Haar probability measure on the orthogonal group O(n), the unitary group U(n), or the symplectic group Sp(n). Given 1≤m<n, a probability inequality for a distance between (γ _ij ) _n×m and some mn independent F-valued normal random variables is obtained, where F=ℝ, ℂ, or ℍ (the set of real quaternions). The result is universal for the three cases. In particular, the inequality for Sp(n) is new.     

Convergence in Distribution of Products of I.I.D. Nonnegative Matrices

Let (X _i) be a sequence of m × m i.i.d. stochastic matrices with distribution . Then ⁿ is the distribution of X _n X _n–1 ...X ₁. Simple sufficient conditions for the weak convergence of ( ⁿ ) are presented here. An extremely simple (and verifiable) necessary and sufficient condition is provided for m= 3. The method for m= 3 works for m> 3 even though calculations are more involved for higher values of m. We also discuss the purity of the limit distribution for m2.     

Strong limit theorems for quasi-orthogonal random fields

This is a systematic and unified treatment of a variety of seemingly different strong limit problems. The main emphasis is laid on the study of the a.s. behavior of the rectangular means ζ_mn = 1/(λ₁(m) λ₂(n)) Σ_i=1^m Σ_k=1ⁿ X_ik as either max{m, n} → ∞ or min{m, n} → ∞. Here {X_ik: i, k ≥ 1} is an orthogonal or merely quasi-orthogonal random field, whereas {λ₁(m): m ≥ 1} and {λ₂(n): n ≥ 1} are nondecreasing sequences of positive numbers subject to certain growth conditions. The method applied provides the rate of convergence, as well. The sufficient conditions obtained are shown to be the best possible in general. Results on double subsequences and 1-parameter limit theorems are also included.     

A constructive approach to minimal projections in Banach spaces

Let X be a Banach space and Y a finite-dimensional subspace of X. Let P be a minimal projection of X onto Y. It is shown (Theorem 1.1) that under certain conditions there exist sequences of finite-dimensional “approximating subspaces” X_m and Y_m of X with corresponding minimal projections P_m: X_m → Y_m, such that lim_m→∞ P_m = P. Moreover, a certain related sequence of projections i_m○P_m○π_m: X → Y has cluster points in the strong operator topology, each of which is a minimal projection of X onto Y. When X = C[a, b] the result reduces to a theorem of [7.]. It is shown (Corollary 1.11) that the hypothesis of Theorem 1.1 holds in many important Banach spaces, including C[a, b], L^P[a, b] and l^P for 1 p < ∞, and c₀, the space of sequences converging to zero in the sup norm.     

Geometric interpretations of two branching theorems of D.E. Littlewood

D.E. Littlewood proved two branching theorems for decomposing the restriction of an irreducible finite-dimensional representation of a unitary group to a symmetric subgroup. One is for restriction of a representation of U(n) to the rotation group SO(n) when the given representation τ_λ of U(n) has nonnegative highest weight λ of depth n/2. It says that the multiplicity in τ_λ|_SO(n) of an irreducible representation of SO(n) of highest weight ν is the sum over μ of the multiplicities of τ_λ in the U(n) tensor product τ_μτ_ν, the allowable μ's being all even nonnegative highest weights for U(n). Littlewood's proof is character-theoretic. The present paper gives a geometric interpretation of this theorem involving the tensor products τ_μτ_ν explicitly. The geometric interpretation has an application to the construction of small infinite-dimensional unitary representations of indefinite orthogonal groups and, for each of these representations, to the determination of its restriction to a maximal compact subgroup. The other Littlewood branching theorem is for restriction from U(2r) to the rank-r quaternion unitary group Sp(r). It concerns nonnegative highest weights for U(2r) of depth r, and its statement is of the same general kind. The present paper finds an analogous geometric interpretation for this theorem also.     

The central limit theorem for weighted empirical processes indexed by sets

Sufficient conditions are found for the weak convergence of a weighted empirical process {(ν_n(C)/q(P(C))) 1 _{[P(C) λn]}: C }, indexed by a class of sets and weighted by a function q of the size of each set. We find those functions q which allow weak convergence to a sample-continuous Gaussian process, and, given q, determine the fastest rate at which one may allow λ_n → 0.     

Convergence of greedy approximation for the trigonometric system

Summary We study the following nonlinear method of approximation by trigonometric polynomials in this paper. For a periodic function f we take as an approximant a trigonometric polynomial of the form G_m(f ) := ∑_kЄΛ f^(k) e ^{(i k,x)}, where ΛˆZ^d is a set of cardinality m containing the indices of the m biggest (in absolute value) Fourier coefficients f^ (k) of function f . Note that G_m(f ) gives the best m-term approximant in the L₂-norm and, therefore, for each f ЄL₂, ║f-G_m(f )║₂→0 as m →∞. It is known from previous results that in the case of p ≠2 the condition f ЄL_p does not guarantee the convergence ║f-G_m(f )║_p→0 as m →∞.. We study the following question. What conditions (in addition to f ЄL_p) provide the convergence ║f-G_m(f )║_p→0 as m →∞? In our previous paper [10] in the case 2< p ≤∞ we have found necessary and sufficient conditions on a decreasing sequence {A_n}_n=1^∞ to guarantee the L_p-convergence of {G_m(f )} for all f ЄL_p , satisfying a_n (f ) ≤A_n , where {a_n (f )} is a decreasing rearrangement of absolute values of the Fourier coefficients of f. In this paper we are looking for necessary and sufficient conditions on a sequence {M (m)} such that the conditions f ЄL_p and ║G_M(m)(f ) - G_m(f )║_p →0 as m →∞ imply ║f - G_m(f )║_p →0 as m →∞. We have found these conditions in the case when p is an even number or p = ∞.     

Lambda-determinants and domino-tilings

Consider the 2n-by-2n matrix with m_i,j=1 for i,j satisfying |2i−2n−1|+|2j−2n−1|2n and m_i,j=0 for all other i,j, consisting of a central diamond of 1's surrounded by 0's. When n4, the λ-determinant of the matrix M (as introduced by Robbins and Rumsey [Adv. Math. 62 (1986) 169–184]) is not well defined. However, if we replace the 0's by t's, we get a matrix whose λ-determinant is well defined and is a polynomial in λ and t. The limit of this polynomial as t→0 is a polynomial in λ whose value at λ=1 is the number of domino-tilings of a 2n-by-2n square.     

Second order freeness and fluctuations of random matrices: II. Unitary random matrices

We extend the relation between random matrices and free probability theory from the level of expectations to the level of fluctuations. We show how the concept of “second order freeness”, which was introduced in Part I, allows one to understand global fluctuations of Haar distributed unitary random matrices. In particular, independence between the unitary ensemble and another ensemble goes in the large N limit over into asymptotic second order freeness. Two important consequences of our general theory are: (i) we obtain a natural generalization of a theorem of Diaconis and Shahshahani to the case of several independent unitary matrices; (ii) we can show that global fluctuations in unitarily invariant multi-matrix models are not universal.     

Comments on a result of Yin, Bai, and Krishnaiah for large dimensional multivariate F matrices

Let {v_ij} i,j = 1, 2,…, be i.i.d. standardized random variables. For each n, let V_n = (v_ij) I = 1, 2,…, n; J = 1, 2,…, S = s(n), where (n/s) → y > 0 as n → ∞, and let M_n = (1/s)V_nV_n^T. Previous results [7, 8] have shown the eigenvectors of M_n to display behavior, for n large, similar to those of the corresponding Wishart matrix. A certain stochastic process X_n on [0, 1], constructed from the eigenvectors of M_n, is known to converge weakly, as n → ∞, on D[0, 1] to Brownian bridge when v₁₁ is N(0, 1), but it is not known whether this property holds for any other distribution. The present paper provides evidence that this property may hold in the non-Wishart case in the form of limit theorems on the convergence in distribution of random variables constructed from integrating analytic function w.r.t. X_n(F_n(x)), where F_n is the empirical distribution function of the eigenvalues of M_n. The theorems assume certain conditions on the moments of v₁₁ including E(v₁₁⁴) = 3, the latter being necessary for the theorems to hold.     

Ramsey functions involving with large

For fixed integers m,k2, it is shown that the k-color Ramsey number r_k(K_m,n) and the bipartite Ramsey number b_k(m,n) are both asymptotically equal to k^mn as n→∞, and that for any graph H on m vertices, the two-color Ramsey number is at most (1+o(1))n^m+1/(logn)^m-1. Moreover, the order of magnitude of is proved to be n^m+1/(logn)^m if H≠K_m as n→∞.     

The Asymptotic Distribution of Singular-Values with Applications to Canonical Correlations and Correspondence Analysis

Let X_n, n = 1, 2, ... be a sequence of p × q random matrices, p ≥ q. Assume that for a fixed p × q matrix B and a sequence of constants b_n → ∞, the random matrix b_n(X_n − B) converges in distribution to Z. Let ψ(X_n) denote the q-vector of singular values of X_n. Under these assumptions, the limiting distribution of b_n (ψ(X_n) − ψ(B)) is characterized as a function of B and of the limit matrix Z. Applications to canonical correlations and to correspondence analysis are given.     

A limit in probability in a W-algebra is unique

Let be the space of self-adjoint Segal measurable operators affiliated to a W^*-algebra (for x=∫λe_x(dλ), x e_x((−∞, −n)(n, ∞)) is a finite projection in for n large enough). The limit in probability is unique in . x_n → x in probability e_nProj ;e_n→1 strongly and (x_n − x) e_n → 0. The following proposition proves to be important in the investigation of . If 1 − ƒ is a finite projection in and 1 − e_n → 0 strongly, e_nProj , then strongly.     

On simultaneous reduction of families of matrices to triangular or diagonal form by unitary congruences

Let {A_i }and {B_i } be two given families of n-by-n matrices. We give conditions under which there is a unitary U such that every matrix UA_iU ¹ is upper triangular. We give conditions, weaker than the classical conditions of commutativity of the whole family, under which there is a unitary U such that every matrix UA_jU ^? is upper triangular. We also give conditions under which there is one single unitary U such that every UA_iU ¹ and every UB_jU ^? is upper triangular. We give necessary and sufficient conditions for simultaneous unitary reduction to diagonal form in this way when all the A_j's are complex symmetric and all theB_j 's are Hermitian.     

Amarts: A class of asymptotic martingales B. Continuous parameter

A continuous-parameter ascending amart is a stochastic process (X_t)_{t +} such that E[X_τn] converges for every ascending sequence (τ_n) of optional times taking finitely many values. A descending amart is a process (X_t)_{t +} such that E[X_τn] converges for every descending sequence (τ_n), and an amart is a process which is both an ascending amart and a descending amart. Amarts include martingales and quasimartingales. The theory of continuous-parameter amarts parallels the theory of continuous-parameter martingales. For example, an amart has a modification every trajectory of which has right and left limits (in the ascending case, if it satisfies a mild boundedness condition). If an amart is right continuous in probability, then it has a modification every trajectory of which is right continuous. The Riesz and Doob-Meyer decomposition theorems are proved by applying the corresponding discrete-parameter decompositions. The Doob-Meyer decomposition theorem applies to general processes and generalizes the known Doob decompositions for continuous-parameter quasimartingales, submartingales, and supermartingales. A hyperamart is a process (X_t) such that E[X_τn] converges for any monotone sequence (τ_n) of bounded optional times, possibly not having finitely many values. Stronger limit theorems are available for hyperamarts. For example: A hyperamart (which satisfies mild regularity and boundedness conditions) is indistinguishable from a process all of whose trajectories have right and left limits.