Zero-One Rounding of Singular Vectors

نویسندگان

  • Amit Deshpande
  • Ravi Kannan
  • Nikhil Srivastava
چکیده

Given a matrix A, it can be shown that there is a vector z ∈ 0, 1 for which |Az|/|Z| ≥ |A|2/C log(n) (a0/1 sum of columns of A which witnesses its large spectral norm) for instance by discretizing the top singular vector of A and taking a dyadic expansion. We give a simple geometric proof of this fact by relating it to the Euclidean diameter of a polytope, and obtain a sharp bound of |Az| ≥ 2|A|2/ √ ln(n). We then show that it is possible to extract k orthogonal 0, 1 vectors z1 . . . zk such that |Azi|/|zi| ≥ σk(A)/ √ k lnn, which is tight up to a √ log k factor, and mention applications to approximating matrices by sums of rectangles. Joint work with Amit Deshpande and Ravi Kannan.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

On Graded Qr Decompositions of Products of Matrices

This paper is concerned with the singular values and vectors of a product Mm = A1A2 · · ·Am of matrices of order n. The chief difficulty with computing them directly from Mm is that with increasing m the ratio of the small to the large singular values of Mm may fall below the rounding unit, so that the former are computed inaccurately. The solution proposed here is to compute recursively the fa...

متن کامل

Construction Formulae for Singular Vectors of the Topological N = 2 Superconformal Algebra

The Topological N=2 Superconformal algebra has 29 different types of singular vectors (in complete Verma modules) distinguished by the relative U(1) charge and the BRST-invariance properties of the vector and of the primary on which it is built. Whereas one of these types only exists at level zero, the remaining 28 types exist for general levels and can be constructed already at level 1. In thi...

متن کامل

Approximating the Distributions of Singular Quadratic Expressions and their Ratios

Noncentral indefinite quadratic expressions in possibly non- singular normal vectors are represented in terms of the difference of two positive definite quadratic forms and an independently distributed linear combination of standard normal random variables. This result also ap- plies to quadratic forms in singular normal vectors for which no general representation is currently available. The ...

متن کامل

Blur identification using averaged spectra of degraded image singular vectors

In this paper we propose new blur identi cation algorithm based on singular value decomposition (SVD) of degraded image. An unknown space-invariant point-spread function (PSF) is also decomposed using SVD. Magnitude functions of PSF singular vectors (left and right) are identi ed using averaged spectra of corresponding singular vectors of degraded image. Phase functions of PSF singular vectors ...

متن کامل

A Novel Noise Reduction Method Based on Subspace Division

This article presents a new subspace-based technique for reducing the noise of signals in time-series. In the proposed approach, the signal is initially represented as a data matrix. Then using Singular Value Decomposition (SVD), noisy data matrix is divided into signal subspace and noise subspace. In this subspace division, each derivative of the singular values with respect to rank order is u...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2012