Infinitely divisible distributions for rectangular free convolution: classification and matricial interpretation

In a previous paper (Benaych-Georges in Related Convolution 2006), we defined the rectangular free convolution ?_λ. Here, we investigate the related notion of infinite divisibility, which happens to be closely related the classical infinite divisibility: there exists a bijection between the set of classical symmetric infinitely divisible distributions and the set of ?_λ -infinitely divisible distributions, which preserves limit theorems. We give an interpretation of this correspondence in terms of random matrices: we construct distributions on sets of complex rectangular matrices which give rise to random matrices with singular laws going from the symmetric classical infinitely divisible distributions to their ?_λ-infinitely divisible correspondents when the dimensions go from one to infinity in a ratio λ.     

Tensor Approximation Approach to Calculation of Singular Values and Vectors for SVD Problem

A new method of calculation of singular values and left and right singular vectors of arbitrary nonsquare matrices is proposed. The method allows one to avoid solutions of high rank systems of linear equations of singular value decomposition problems, which makes it not sensitive to ill-conditioness of the decomposed matrix. On the base of the Eckart–Young theorem, it was shown that each second order r-rank tensor can be represented as a sum of the first rank r-order “coordinate” tensors. A new system of equations for “coordinate” tensor generator vectors was obtained. An iterative method of solution of the system was elaborated. Results of the method were compared with classical methods of solutions of singular value decomposition problems.     

Regularization by Free Additive Convolution, Square and Rectangular Cases

The free convolution \boxplus\boxplus is the binary operation on the set of probability measures on the real line which allows to deduce, from the individual spectral distributions, the spectral distribution of a sum of independent unitarily invariant square random matrices or of a sum of free operators in a non commutative probability space. In the same way, the rectangular free convolution \boxplus_l\boxplus_{\lambda} allows to deduce, from the individual singular distributions, the singular distribution of a sum of independent unitarily invariant rectangular random matrices. In this paper, we consider the regularization properties of these free convolutions on the whole real line. More specifically, we try to find continuous semigroups (μ_t) of probability measures such that μ₀ = δ₀ and such that for all t > 0 and all probability measure n, m_t\boxplusn\nu, \mu_t\boxplus\nu (or, in the rectangular context, m_t\boxplus_ln\mu_t\boxplus_{\lambda}\nu) is absolutely continuous with respect to the Lebesgue measure, with a positive analytic density on the whole real line. In the square case, for \boxplus\boxplus, we prove that in semigroups satisfying this property, no measure can have a finite second moment, and we give a sufficient condition on semigroups to satisfy this property, with examples. In the rectangular case, we prove that in most cases, for μ in a \boxplus_l\boxplus_{\lambda}-continuous semigroup, m\boxplus_ln\mu\boxplus_{\lambda}\nu either has an atom at the origin or doesn’t put any mass in a neighborhood of the origin, and thus the expected property does not hold. However, we give sufficient conditions for analyticity of the density of m\boxplus_ln\mu\boxplus_{\lambda}\nu except on a negligible set of points, as well as existence and continuity of a density everywhere.     

Remark on the norm of random Hankel matrices

In recent years, the asymptotic properties of structured random matrices have attracted the attention of many experts involved in probability theory. In particular, R. Adamczak (J. Theor. Probab., Vol. 23, 2010) proved that, under fairly weak conditions, the squared spectral norms of large square Hankel matrices generated by independent identically distributed random variables grow with probability 1, as Nln(N), where N is the size of a matrix. On the basis of these results, by using the technique and ideas of Adamczak’s paper cited above, we prove that, under certain constraints, the squared spectral norms of large rectangular Hankel matrices generated by linear stationary sequences grow almost certainly no faster than Nln(N), where N is the number of different elements in a Hankel matrix. Nekrutkin (Stat. Interface, Vol. 3, 2010) pointed out that this result may be useful for substantiating (by using series of perturbation theory) so-called “signal subspace methods,” which are often used for processing time series. In addition to the main result, the paper contains examples and discusses the sharpness of the obtained inequality.     

Concentration of norms and eigenvalues of random matrices

We prove concentration results for ?_pⁿ operator norms of rectangular random matrices and eigenvalues of self-adjoint random matrices. The random matrices we consider have bounded entries which are independent, up to a possible self-adjointness constraint. Our results are based on an isoperimetric inequality for product spaces due to Talagrand.     

Smallest singular value of random matrices and geometry of random polytopes

We study the behaviour of the smallest singular value of a rectangular random matrix, i.e., matrix whose entries are independent random variables satisfying some additional conditions. We prove a deviation inequality and show that such a matrix is a “good” isomorphism on its image. Then, we obtain asymptotically sharp estimates for volumes and other geometric parameters of random polytopes (absolutely convex hulls of rows of random matrices). All our results hold with high probability, that is, with probability exponentially (in dimension) close to 1.     

The Limiting Spectral Measure for Ensembles of Symmetric Block Circulant Matrices

Given an ensemble of N×N random matrices, a natural question to ask is whether or not the empirical spectral measures of typical matrices converge to a limiting spectral measure as N→∞. While this has been proved for many thin patterned ensembles sitting inside all real symmetric matrices, frequently there is no nice closed form expression for the limiting measure. Further, current theorems provide few pictures of transitions between ensembles. We consider the ensemble of symmetric m-block circulant matrices with entries i.i.d.r.v. These matrices have toroidal diagonals periodic of period m. We view m as a “dial” we can “turn” from the thin ensemble of symmetric circulant matrices, whose limiting eigenvalue density is a Gaussian, to all real symmetric matrices, whose limiting eigenvalue density is a semi-circle. The limiting eigenvalue densities f _m show a visually stunning convergence to the semi-circle as m→∞, which we prove. In contrast to most studies of patterned matrix ensembles, our paper gives explicit closed form expressions for the densities. We prove that f _m is the product of a Gaussian and a certain even polynomial of degree 2m?2; the formula is the same as that for the m×m Gaussian Unitary Ensemble (GUE). The proof is by derivation of the moments from the eigenvalue trace formula. The new feature, which allows us to obtain closed form expressions, is converting the central combinatorial problem in the moment calculation into an equivalent counting problem in algebraic topology. We end with a generalization of the m-block circulant pattern, dropping the assumption that the m random variables be distinct. We prove that the limiting spectral distribution exists and is determined by the pattern of the independent elements within an m-period, depending not only on the frequency at which each element appears, but also on the way the elements are arranged.     

Asymptotics of products of nonnegative random matrices

Asymptotic properties of products of random matrices ξ _k = X _k …X ₁ as k → ∞ are analyzed. All product terms X _i are independent and identically distributed on a finite set of nonnegative matrices A = {A ₁, …, A _m }. We prove that if A is irreducible, then all nonzero entries of the matrix ξ _k almost surely have the same asymptotic growth exponent as k→∞, which is equal to the largest Lyapunov exponent λ(A). This generalizes previously known results on products of nonnegative random matrices. In particular, this removes all additional “nonsparsity” assumptions on matrices imposed in the literature.We also extend this result to reducible families. As a corollary, we prove that Cohen’s conjecture (on the asymptotics of the spectral radius of products of random matrices) is true in case of nonnegative matrices.     

von Neumann’s trace inequality for Hilbert–Schmidt operators

von Neumann’s inequality in matrix theory refers to the fact that the Frobenius scalar product of two matrices is less than or equal to the scalar product of the respective singular values. Moreover, equality can only happen if the two matrices share a joint set of singular vectors, and this latter part is hard to find in the literature. We extend these facts to the separable Hilbert space setting, and provide a self-contained proof of the “latter part”.     

Random Triangle Theory with Geometry and Applications

What is the probability that a random triangle is acute? We explore this old question from a modern viewpoint, taking into account linear algebra, shape theory, numerical analysis, random matrix theory, the Hopf fibration, and much more. One of the best distributions of random triangles takes all six vertex coordinates as independent standard Gaussians. Six can be reduced to four by translation of the center to \((0,0)\) or reformulation as a \(2\times 2\) random matrix problem. In this note, we develop shape theory in its historical context for a wide audience. We hope to encourage others to look again (and differently) at triangles. We provide a new constructive proof, using the geometry of parallelians, of a central result of shape theory: triangle shapes naturally fall on a hemisphere. We give several proofs of the key random result: that triangles are uniformly distributed when the normal distribution is transferred to the hemisphere. A new proof connects to the distribution of random condition numbers. Generalizing to higher dimensions, we obtain the “square root ellipticity statistic” of random matrix theory. Another proof connects the Hopf map to the SVD of \(2\times 2\) matrices. A new theorem describes three similar triangles hidden in the hemisphere. Many triangle properties are reformulated as matrix theorems, providing insight into both. This paper argues for a shift of viewpoint to the modern approaches of random matrix theory. As one example, we propose that the smallest singular value is an effective test for uniformity. New software is developed, and applications are proposed.     

Spectral norm of products of random and deterministic matrices

We study the spectral norm of matrices W that can be factored as W?=?BA, where A is a random matrix with independent mean zero entries and B is a fixed matrix. Under the (4?+???)th moment assumption on the entries of A, we show that the spectral norm of such an m × n matrix W is bounded by ${\sqrt{m} + \sqrt{n}}$ , which is sharp. In other words, in regard to the spectral norm, products of random and deterministic matrices behave similarly to random matrices with independent entries. This result along with the previous work of Rudelson and the author implies that the smallest singular value of a random m × n matrix with i.i.d. mean zero entries and bounded (4?+???)th moment is bounded below by ${\sqrt{m} - \sqrt{n-1}}$ with high probability.     

Record values with constraints

Record values are very popular in probability and mathematical statistics. There are many books and papers concerned with classical record values and record times, i.e., records in sequences of independent equally distributed random variables. In recent times, new types of record values (records in the F ^α-scheme, record values in sequences of unequally distributed random variables, records with confirmations, exceedance record values) have been proposed and examined. The present paper proposes another record scheme (so-called “records with constraint”). Various cases are studied in which these records may be useful. For these record values, we give their joint density functions and discover some of their properties. For particular cases of utmost importance, when the initial random variables are independent and have equal exponential distribution, we obtain fairly simple representations of records with constraints as sums of independent equally distributed random terms.     

The complete classification of fibres in pencils of curves of genus two

This article contains geometrical classification of all fibres in pencils of curves of genus two, which is essentially different from the numerical one given by Ogg ([11]) and Iitaka ([7]). Given a family π:X→D of curves of genus two which is smooth overD′=D?{0}, we define a multivalued holomorphic mapT _π fromD′ into the Siegel upper half plane of degree two, and three invariants called “monodromy”, “modulus point” and “degree”. We assert that the family π is completely determined byT _π, and its singular fibre by these three invariants. Hence all types of fibres are classified by these invariants and we list them up in a table, which is the main part of this article.     

Linear operators preserving certain functions on singular values of matrices

We study the structure of those linear operators on the rectangular complex or real matrix spaces that preserve certain functions on singular values. We first do a brief survey on the existing results in the area and then prove a theorem which covers and extends all of them. In particular. our theorem confirms two conjectures about the structure of those linear operators preserving the completely symmetric functions on powers of singular values of matrices.     

Random orthogonal matrix simulation

This paper introduces a method for simulating multivariate samples that have exact means, covariances, skewness and kurtosis. We introduce a new class of rectangular orthogonal matrix which is fundamental to the methodology and we call these matrices L matrices. They may be deterministic, parametric or data specific in nature. The target moments determine the L matrix then infinitely many random samples with the same exact moments may be generated by multiplying the L matrix by arbitrary random orthogonal matrices. This methodology is thus termed “ROM simulation”. Considering certain elementary types of random orthogonal matrices we demonstrate that they generate samples with different characteristics. ROM simulation has applications to many problems that are resolved using standard Monte Carlo methods. But no parametric assumptions are required (unless parametric L matrices are used) so there is no sampling error caused by the discrete approximation of a continuous distribution, which is a major source of error in standard Monte Carlo simulations. For illustration, we apply ROM simulation to determine the value-at-risk of a stock portfolio.     

Linear operators preserving certain functions on singular values of matrices

We study the structure of those linear operators on the rectangular complex or real matrix spaces that preserve certain functions on singular values. We first do a brief survey on the existing results in the area and then prove a theorem which covers and extends all of them. In particular. our theorem confirms two conjectures about the structure of those linear operators preserving the completely symmetric functions on powers of singular values of matrices.     

Limit distributions of random matrices

We study limit distributions of independent random matrices as well as limit joint distributions of their blocks under normalized partial traces composed with classical expectation. In particular, we are concerned with the ensemble of symmetric blocks of independent Hermitian random matrices which are asymptotically free, asymptotically free from diagonal deterministic matrices, and whose norms are uniformly bounded almost surely. This class contains symmetric blocks of unitarily invariant Hermitian random matrices whose asymptotic distributions are compactly supported probability measures on the real line. Our approach is based on the concept of matricial freeness which is a generalization of freeness in free probability. We show that the associated matricially free Gaussian operators provide a unified framework for studying the limit distributions of sums and products of independent rectangular random matrices, including non-Hermitian Gaussian matrices and matrices of Wishart type.     

On Complex Split Quaternion Matrices

In this paper, we present some important properties of complex split quaternions and their matrices. We also prove that any complex split quaternion has a 4 × 4 complex matrix representation. On the other hand, we give answers to the following two basic questions “If AB =  I, is it true that BA =  I for complex split quaternion matrices?” and “How can the inverse of a complex split quaternion matrix be found?”. Finally, we give an explicit formula for the inverse of a complex split quaternion matrix by using complex matrices.     

The Littlewood-Offord problem and invertibility of random matrices

We prove two basic conjectures on the distribution of the smallest singular value of random n×n matrices with independent entries. Under minimal moment assumptions, we show that the smallest singular value is of order n−1/2, which is optimal for Gaussian matrices. Moreover, we give a optimal estimate on the tail probability. This comes as a consequence of a new and essentially sharp estimate in the Littlewood-Offord problem: for i.i.d. random variables Xk and real numbers ak, determine the probability p that the sum k_∑akXk lies near some number v. For arbitrary coefficients ak of the same order of magnitude, we show that they essentially lie in an arithmetic progression of length 1/p.     

Estimating Averages of Order Statistics of Bivariate Functions

We prove uniform estimates for the expected value of averages of order statistics of bivariate functions in terms of their largest values by a direct analysis. As an application, uniform estimates for the expected value of averages of order statistics of sequences of independent random variables in terms of Orlicz norms are obtained. In the case where the bivariate functions are matrices, we provide a “minimal” probability space which allows us to C-embed certain Orlicz spaces \(\ell _M^n\) into \(\ell _1^{cn^3}\), with \(c,C>0\) being absolute constants.