Upper and Lower bounds for matrix discrepancy

Abstract: The aim of this paper is to study the matrix discrepancy problem. Assume that

x i_{1}, l d o t s, x i_{n}

are independent scalar random variables with finite support and

m a t h b f u_{1}, l d o t s, m a t h b f u_{n} i n m a t h b b C^{d}

. Let

m a t h c a l C_{0}

be the minimal constant for which the following holds: [ { m Disc}(mathbf{u}_1mathbf{u}_1^*,ldots,mathbf{u}_nmathbf{u}_n^*; xi_1,ldots,xi_n),,:=,,min_{varepsilon_1in mathcal{S}_1,ldots,varepsilon_nin mathcal{S}_n}�igg|sum_{i=1}^nmathbb{E}[xi_i]mathbf{u}_imathbf{u}_i^*-sum_{i=1}^nvarepsilon_imathbf{u}_imathbf{u}_i^*�igg|leq mathcal{C}_0cdotsigma, ] where

and

m a t h c a l S_{j}

denotes the support of

x i_{j}, j = 1, l d o t s, n

. Motivated by the technology developed by Bownik, Casazza, Marcus, and Speegle, we prove

m a t h c a l C_{0} l e q 3

. This improves Kyng, Luh and Song's method with which

m a t h c a l C_{0} l e q 4

. For the case where

{m a t h b f u_{i}}_{i = 1}^{n} s u b s e t m a t h b b C^{d}

is a unit-norm tight frame with

n l e q 2 d - 1

and

x i_{1}, l d o t s, x i_{n}

are independent Rademacher random variables, we present the exact value of

m D i s c (m a t h b f u_{1} m a t h b f u_{1}^{*}, l d o t s, m a t h b f u_{n} m a t h b f u_{n}^{*}; x i_{1}, l d o t s, x i_{n}) = s q r t f r a c n d c d o t s i g m a

, which implies

m a t h c a l C_{0} g e q s q r t 2

.

This page was built for publication: Upper and Lower bounds for matrix discrepancy

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6343426)