eScholarship Repository eScholarship Repository California Digital Library
eScholarship > UCLASTAT > PAPERS > Paper 2006041602

Statistics Papers

Statistics Website

Policies

Search Statistics

Submit a Paper

Notify me of new papers
institute_logo

Department of Statistics, UCLA
University of California, Los Angeles
Statistics Papers  •  Statistics Website  •  Policies  •  Search Statistics  •  Submit a Paper

A Decomposition Method for Weighted Least Squares Low-rank Approximation of Symmetric Matrices
Jan de Leeuw, Department of Statistics, UCLA

Download the Paper (169 K, PDF file) - April 16, 2006 Tell a colleague about it.
Printing Tips: Select 'print as image' in the Acrobat print dialog if you have trouble printing.

ABSTRACT:

We discuss an alternating least squares algorithm that uses both decomposition and block relaxation to find the optimal positive semidefinite approxation of given rank p to a known symmetric matrix of order n. Each iteration of the algorithm involves minimizing n quartics and solving n secular equations of order p.

SUGGESTED CITATION:
Jan de Leeuw, "A Decomposition Method for Weighted Least Squares Low-rank Approximation of Symmetric Matrices" (April 16, 2006). Department of Statistics, UCLA. Department of Statistics Papers. Paper 2006041602.
http://repositories.cdlib.org/uclastat/papers/2006041602

 
bar
Open Archives Initiative eScholarship is a service of the California Digital Library bepress