Searcharxiv⌕ Search

arXiv subjects

V. V. Peller

Publications and source records attributed to V. V. Peller.

27 records · Page 2Linked to original sources

Multiple operator integrals and higher operator derivatives

In this paper we consider the problem of the existence of higher derivatives of the function $t\mapsto\f(A+tK)$, where $\f$ is a function on the real line, $A$ is a self-adjoint operator, and $K$ is a bounded self-adjoint operator. We improve earlier results by Sten'kin. In order to do this, we give a new approach to multiple operator integrals. This approach improves the earlier approach given by Sten'kin. We also consider a similar problem for unitary operators.

math.SP↗

Approximation by analytic operator functions. Factorizations and very badly approximable functions

This is a continuation of our earlier paper \cite{PT3}. We consider here operator-valued functions (or infinite matrix functions) on the unit circle $\T$ and study the problem of approximation by bounded analytic operator functions. We discuss thematic and canonical factorizations of operator functions and study badly approximable and very badly approximable operator functions. We obtain algebraic and geometric characterizations of badly approximable and very badly approximable operator functions. Note that there is an important difference between the case of finite matrix functions and the case of operator functions. Our criteria for a function to be very badly approximable in the case of finite matrix functions also guarantee that the zero function is the only superoptimal approximant. However in the case of operator functions this is not true.

math.FA↗

Very badly approximable matrix functions

We study in this paper very badly approximable matrix functions on the unit circle $\T$, i.e., matrix functions $Φ$ such that the zero function is a superoptimal approximation of $Φ$. The purpose of this paper is to obtain a characterization of the continuous very badly approximable functions. Our characterization is more geometric than algebraic characterizations earlier obtained in \cite{PY} and \cite{AP}. It involves analyticity of certain families of subspaces defined in terms of Schmidt vectors of the matrices $Φ(\z)$, $\z\in\T$. This characterization can be extended to the wider class of {\em admissible} functions, i.e., the class of matrix functions $Φ$ such that the essential norm $\|H_Φ\|_{\rm e}$ of the Hankel operator $H_Φ$ is less than the smallest nonzero superoptimal singular value of $Φ$. In the final section we obtain a similar characterization of badly approximable matrix functions.

math.FA↗

Distorted Hankel integral operators

For $\a,\b>0$ and for a locally integrable function (or, more generally, a distribution) $\f$ on $(0,\be)$, we study integral ooperators ${\frak G}^{\a,\b}_\f$ on $L^2(\R_+)$ defined by $\big({\frak G}^{\a,\b}_\f f\big)(x)=\int_{\R_+}\f\big(x^\a+y^\b\big)f(y)dy$. We describe the bounded and compact operators ${\frak G}^{\a,\b}_\f$ and operators ${\frak G}^{\a,\b}_\f$ of Schatten--von Neumann class $\bS_p$. We also study continuity properties of the averaging projection $\Q_{\a,\b}$ onto the operators of the form ${\frak G}^{\a,\b}_\f$. In particular, we show that if $\a\le\b$ and $\b>1$, then ${\frak G}^{\a,\b}_\f$ is bounded on $\bS_p$ if and only if $2\b(\b+1)^{-1}<p<2\b(\b-1)^{-1}$.

math.FA↗

An Interesting Class of Operators with unusual Schatten-von Neumann behavior

We consider the class of integral operators $Q_\f$ on $L^2(\R_+)$ of the form $(Q_\f f)(x)=\int_0^\be\f (\max\{x,y\})f(y)dy$. We discuss necessary and sufficient conditions on $ϕ$ to insure that $Q_ϕ$ is bounded, compact, or in the Schatten-von Neumann class $\bS_p$, $1<p<\infty$. We also give necessary and sufficient conditions for $Q_ϕ$ to be a finite rank operator. However, there is a kind of cut-off at $p=1$, and for membership in $\bS_{p}$, $0<p\leq1$, the situation is more complicated. Although we give various necessary conditions and sufficient conditions relating to $Q_ϕ\in\bS_{p}$ in that range, we do not have necessary and sufficient conditions. In the most important case $p=1$, we have a necessary condition and a sufficient condition, using $L^1$ and $L^2$ modulus of continuity, respectively, with a rather small gap in between. A second cut-off occurs at $p=1/2$: if $\f$ is sufficiently smooth and decays reasonably fast, then $\qf$ belongs to the weak Schatten-von Neumann class $\wS{1/2}$, but never to $\bS_{1/2}$ unless $\f=0$. We also obtain results for related families of operators acting on $L^2(\R)$ and $\ell^2(\Z)$. We further study operations acting on bounded linear operators on $L^{2}(\R^{+})$ related to the class of operators $Q_\f$. In particular we study Schur multipliers given by functions of the form $ϕ(\max\{x,y\}) $ and we study properties of the averaging projection (Hilbert-Schmidt projection) onto the operators of the form $Q_\f$.

math.FA↗

Hankel and Toeplitz-Schur Multipliers

We study the problem of characterizing Hankel-Schur multipliers and Toeplitz-Schur multipliers of Schatten-von Neumann class $\bS_p$ for $0<p<1$. We obtain various sharp necessary conditions and sufficient conditions for a Hankel matrix to be a Schur multiplier of $\bS_p$. We also give a characterization of the Hankel-Schur multipliers of $\bS_p$ whose symbols have lacunary power series. Then the results on Hankel-Schur multipliers are used to obtain a characterization of the Toeplitz-Schur multipliers of $\bS_p$. Finally, we return to Hankel-Schur multipliers and obtain new results in the case when the symbol of the Hankel matrix is a complex measure on the unit circle.

math.FA↗

Invariance properties of thematic factorizations of matrix functions

We study the problem of invariance of indices of thematic factorizations. Such factorizations were introduced in [PY1] for studying superoptimal approximation by bounded analytic matrix functions. As shown in [PY1], the indices may depend on the choice of a thematic factorization. We introduce the notion of a monotone thematic factorization. The main result shows that under natural assumptions a matrix function that admits a thematic factorization also admits a monotone thematic factorization and the indices of a monotone thematic factorization are uniquely determined by the matrix function itself. We obtain similar results for so-called partial thematic factorizations.

math.FA↗

Unitary interpolants and factorization indices of matrix functions

For an $n\times n$ bounded matrix function $Φ$ we study unitary interpolants $U$, i.e., unitary-valued functions $U$ such that $\hat U(j)=\hatΦ(j)$, $j<0$. We are looking for unitary interpolants $U$ for which the Toeplitz operator $T_U$ is Fredholm. We give a new approach based on superoptimal singular values and thematic factorizations. We describe Wiener--Hopf factorization indices of $U$ in terms of superoptimal singular values of $Φ$ and thematic indices of $Φ-F$, where $F$ is a superoptimal approximation of $Φ$ by bounded analytic matrix functions. The approach essentially relies on the notion of a monotone thematic factorization introduced in [AP]. In the last section we discuss hereditary properties of unitary interpolants. In particular, for matrix functions $Φ$ of class $H^\be+C$ we study unitary interpolants $U$ of class $QC$.

math.FA↗

Badly approximable matrix functions and canonical factorizations

We continue studying the problem of analytic approximation of matrix functions. We introduce the notion of a partial canonical factorization of a badly approximable matrix function $Φ$ and the notion of a canonical factorization of a very badly approximable matrix function $Φ$. Such factorizations are defined in terms of so-called balanced unitary-valued functions which have many remarkable properties. Unlike the case of thematic factorizations studied earlier in [PY1], [PY2], [PT], [AP1], the factors in canonical factorizations (as well as partial canonical factorizations) are uniquely determined by the matrix function $Φ$ up to constant unitary factors. We study many properties of canonical factorizations. In particular we show that under certain natural assumptions on a function space $X$ the condition $Φ\in X$ implies that all factors in a canonical factorization of $Φ$ belong to the same space $X$. In the last section we characterize the very badly approximable unitary-valued functions $U$ that satisfy the condition $\|H_U\|_{\text e}<1$.

math.FA↗