Searcharxiv⌕ Search

arXiv subjects

Michael Anshelevich

Publications and source records attributed to Michael Anshelevich.

At least 19 recordsLinked to original sources
arXiv2025

Convergence of permuted products of exponentials

Let $\{A_{i,n}\}$ be a triangular array of elements in a Banach algebra, whose norms do not grow too fast, and whose row averages converge to $A$. Let $\sigma \in S(n)$ be a permutation drawn uniformly at random. If the array only contains $o(n / \log n)$ distinct elements, then almost surely, for all $0 < s < t < 1$, the permuted product of their exponentials $\prod_{i = [s n]}^{[t n]} e^{A_{\sigma(i),n}/n}$ converges in norm to $e^{(t - s) A}$. For an array of finite-dimensional matrices, convergence holds without this restriction. The proof of the latter result consists of an estimate valid in a general Banach algebra, and an application of a matrix concentration inequality.

math.FA↗
arXiv2023

Fock representation of free convolution powers

Let $B$ be a star-algebra with a state $ϕ$, and $t > 0$. Through a Fock space construction, we define two states $Φ_t$ and $Ψ_t$ on the tensor algebra $T(B, ϕ)$ such that under the natural map $(B, ϕ) \rightarrow (T(B, ϕ), Φ_t, Ψ_t)$, free independence of arguments leads to free independence, while Boolean independence of centered arguments leads to conditionally free independence. The construction gives a new operator realization of the $(1+t)$'th free convolution power of any joint (star) distribution. We also compute several von Neumann algebras which arise.

math.OA↗
arXiv2023

Higher variations for free Lévy processes

For a general free Lévy process, we prove the existence of its higher variation processes as limits in distribution, and identify the limits in terms of the Lévy-Itô representation of the original process. For a general free compound Poisson process, this convergence holds in probability. This implies joint convergence in distribution to a $k$-tuple of higher variation processes, and so the existence of $k$-fold stochastic integrals as limits in probability. If the existence of moments of all orders is assumed, the result holds for free additive (not necessarily stationary) processes and more general approximants. In the appendix we note relevant properties of symmetric polynomials in non-commuting variables.

math.OA↗
arXiv2022

Hermite trace polynomials and chaos decompositions for the Hermitian Brownian motion

For a non-zero parameter $q$, we define Hermite trace polynomials, which are multivariate polynomials indexed by permutations. We prove several combinatorial properties for them, such as expansions and product formulas. The linear functional determined by these trace polynomials is a state for $q = \frac{1}{N}$ for $N$ a non-zero integer. For such $q$, Hermite trace polynomials of different degrees are orthogonal. The product formulas extend to the closure with respect to the state. The state can be identified with the expectation induced by the $N \times N$ Hermitian Brownian motion. Hermite trace polynomials are martingales for this Brownian motion, while the elements in the closure can be interpreted as stochastic integrals with respect to it. Using the grading on the algebra, we prove several chaos decompositions for such integrals, as well as analyze corresponding creation and annihilation operators. In the univariate, pure trace polynomial case, trace Hermite polynomials can be identified with the Hermite polynomials of matrix argument.

math.PR↗
arXiv2022

Product of exponentials concentrates around the exponential of the sum

For two matrices $A$ and $B$, and large $n$, we show that most products of $n$ factors of $e^{A/n}$ and $n$ factors of $e^{B/n}$ are close to $e^{A + B}$. This extends the Lie-Trotter formula. The elementary proof is based on the relation between words and lattice paths, asymptotics of binomial coefficients, and matrix inequalities. The result holds for more than two matrices.

math.CO↗
arXiv2021

Some Fock spaces with depth two action

The subject of this paper are operators represented on a Fock space which act only on the two leading components of the tensor. We unify the constructions from arXiv:math/0702158, arXiv:0709.4334, arXiv:0812.0895, and arXiv:1003.2998, and extend a number of results from these papers to our more general setting. The results include the quadratic relation satisfied by (the kernel of) the free cumulant generating function, the resolvent form of the generating function for the Wick polynomials, and classification results for the case when the vacuum state on the operator algebra is tracial. We handle the generating functions in infinitely many variables by considering their matrix-valued versions.

math.OA↗
arXiv2018

Product formulas on posets, Wick products, and a correction for the $q$-Poisson process

We give an example showing that the product and linearization formulas for the Wick product versions of the $q$-Charlier polynomials in (Anshelevich 2004) are incorrect. Next, we observe that the relation between monomials and several families of Wick polynomials is governed by "incomplete" versions of familiar posets. We compute Möbius functions for these posets, and prove a general poset product formula. These provide new proofs and new inversion and product formulas for Wick product versions of Hermite, Chebyshev, Charlier, free Charlier, and Laguerre polynomials. By different methods, we prove inversion formulas for the Wick product versions of the free Meixner polynomials.

math.OA↗
arXiv2017

Expansion of permutations as products of transpositions

We compute the number of ways a given permutation can be written as a product of exactly $k$ transpositions. We express this number as a linear combination of explicit geometric sequences, with coefficients which can be computed in many particular cases. Along the way we prove several symmetry properties for matrices associated with bipartite graphs, as well as some general (likely known) properties of Young diagrams. The methods involve linear algebra, enumeration of border strip tableau, and a differential operator on symmetric polynomials.

math.CO↗
arXiv2016

The exponential map in non-commutative probability

The wrapping transformation $W$ is a homomorphism from the semigroup of probability measures on the real line, with the convolution operation, to the semigroup of probability measures on the circle, with the multiplicative convolution operation. We show that on a large class $\mathcal{L}$ of measures, $W$ also transforms the three non-commutative convolutions---free, Boolean, and monotone---to their multiplicative counterparts. Moreover, the restriction of $W$ to $\mathcal{L}$ preserves various qualitative properties of measures and triangular arrays. We use these facts to give short proofs of numerous known, and new, results about multiplicative convolutions.

math.PR↗
arXiv2015

Operator-valued Jacobi parameters and examples of operator-valued distributions

In the setting of distributions taking values in a $C^\ast$-algebra $\mathcal{B}$, we define generalized Jacobi parameters and study distributions they generate. These include numerous known examples and one new family, of $\mathcal{B}$-valued free binomial distributions, for which we are able to compute free convolution powers. Moreover, we develop a convenient combinatorial method for calculating the joint distributions of $\mathcal{B}$-free random variables with Jacobi parameters, utilizing two-color non-crossing partitions. This leads to several new explicit examples of free convolution computations in the operator-valued setting. Additionally, we obtain a counting algorithm for the number of two-color non-crossing pairings of relative finite depth, using only free probabilistic techniques. Finally, we show that the class of distributions with Jacobi parameters is not closed under free convolution.

math.OA↗
arXiv2015

Operator-Valued Monotone Convolution Semigroups and an Extension of the Bercovici-Pata Bijection

In a 1999 paper, Bercovici and Pata showed that a natural bijection between the classically, free and Boolean infinitely divisible measures held at the level of limit theorems of triangular arrays. This result was extended to include monotone convolution by the authors. In recent years, operator-valued versions of free, Boolean and monotone probability have also been developed. Belinschi, Popa and Vinnikov showed that the Bercovici-Pata bijection holds for the operator-valued versions of free and Boolean probability. In this article, we extend the bijection to include monotone probability theory even in the operator-valued case. To prove this result, we develop the general theory of composition semigroups of non-commutative functions and largely recapture Berkson and Porta's classical results on composition semigroups of complex functions in operator-valued setting. As a biproduct, we deduce that operator-valued monotonically infinitely divisible distributions belong to monotone convolution semigroups. Finally, in the appendix, we extend the result of the second author on the classification of Cauchy transforms for non-commutative distributions to the Cauchy transforms associated to more general completely positive maps.

math.OA↗
arXiv2014

Local limit theorems for multiplicative free convolutions

This paper describes the quality of convergence to an infinitely divisible law relative to free multiplicative convolution. We show that convergence in distribution for products of identically distributed and infinitesimal free random variables implies superconvergence of their probability densities to the density of the limit law. Superconvergence to the marginal law of free multiplicative Brownian motion at a specified time is also studied. In the unitary case, the superconvergence to free Brownian motion and that to the Haar measure are shown to be uniform over the entire unit circle, implying further a free entropic limit theorem and a universality result for unitary free Lévy processes. Finally, the method of proofs on the positive half-line gives rise to a new multiplicative Boolean to free Bercovici-Pata bijection.

math.FA↗
arXiv2014

Free evolution on algebras with two states II

Denote by $J$ the operator of coefficient stripping. We show that for any free convolution semigroup of measures $ν_t$ with finite variance, applying a single stripping produces semicircular evolution with non-zero initial condition, $J[ν_t] = ρ\boxplus σ^{\boxplus t}$, where $σ$ is the semicircular distribution with mean $β$ and variance $γ$. For more general freely infinitely divisible distributions $τ$, expressions of the form $ρ\boxplus τ^{\boxplus t}$ arise from stripping $μ_t$, where the pairs $(μ_t, ν_t)$ form a semigroup under the operation of two-state free convolution. The converse to this statement holds in the algebraic setting. Numerous examples illustrating these constructions are computed. Additional results include the formula for generators of such semigroups.

math.OA↗
arXiv2013

Limit theorems for monotonic convolution and the Chernoff product formula

Bercovici and Pata showed that the correspondence between classically, freely, and Boolean infinitely divisible distributions holds on the level of limit theorems. We extend this correspondence also to distributions infinitely divisible with respect to the additive monotone convolution. Because of non-commutativity of this convolution, we use a new technique based on the Chernoff product formula. In fact, the correspondence between the Boolean and monotone limit theorems extends from probability measures to positive measures of total weight at most one. Finally, we study this correspondence for multiplicative monotone convolution, where the Bercovici-Pata bijection no longer holds.

math.OA↗
arXiv2013

Generators of some non-commutative stochastic processes

A fundamental result of Biane (1998) states that a process with freely independent increments has the Markov property, but that there are two kinds of free Levy processes: the first kind has stationary increments, while the second kind has stationary transition operators. We show that a process of the first kind (with mean zero and finite variance) has the same transition operators as the free Brownian motion with appropriate initial conditions, while a process of the second kind has the same transition operators as a monotone Levy process. We compute an explicit formula for the generators of these families of transition operators, in terms of singular integral operators, and prove that this formula holds on a fairly large domain. We also compute the generators for the $q$-Brownian motion, and for the two-state free Brownian motions.

math.OA↗
arXiv2011

The free Meixner class for pairs of measures

We investigate in more detail the two-state free convolution semigroups of pairs of measures whose Jacobi parameters are linear in the convolution parameter $t$. These semigroups were constructed in arXiv:1001.1540, where we also showed that measures with the analogous property for the usual and free convolution are exactly the classical, resp. free Meixner classes. The class of measures in this paper has not been considered explicitly before, but we show that it also has Meixner-type properties. Specifically, it appears in limit theorems, has a Laha-Lukacs-type characterization, and is related to the $q=0$ case of quadratic harnesses.

math.OA↗
arXiv2011

Semigroups of distributions with linear Jacobi parameters

We show that a convolution semigroup of measures has Jacobi parameters polynomial in the convolution parameter $t$ if and only if the measures come from the Meixner class. Moreover, we prove the parallel result, in a more explicit way, for the free convolution and the free Meixner class. We then construct the class of measures satisfying the same property for the two-state free convolution. This class of two-state free convolution semigroups has not been considered explicitly before. We show that it also has Meixner-type properties. Specifically, it contains the analogs of the normal, Poisson, and binomial distributions, has a Laha-Lukacs-type characterization, and is related to the $q=0$ case of quadratic harnesses.

math.CO↗