Accession Number : ADA326735

Title :   Alternating Direction Multiplier Decomposition of Convex Problems,

Corporate Author : THINKING MACHINES CORP CAMBRIDGE MA

Personal Author(s) : Eckstein, Jonathan

PDF Url : ADA326735

Report Date : OCT 1991

Pagination or Media Count : 26

Abstract : This paper focuses on two kinds of problems, monotropic programs and block-seperable problems. For block-seperable problems, the algorithm obtained bears some resemblance to an earlier method due to Spingarn, but solves a larger number of simpler subproblems at each iteration. Its fundamental operation is projection onto the epigraph of a convex function. For monotropic programs, one obtains a compact method that has some interesting properties when specialized to linear programming, and, for quadratic problems, has been shown to be competitive in the massively parallel environment.

Descriptors :   *OPTIMIZATION, *LINEAR PROGRAMMING, ALGORITHMS, MATRICES(MATHEMATICS), PARALLEL PROCESSING, SYSTEMS ANALYSIS.

Subject Categories : Operations Research
      Computer Programming and Software

Distribution Statement : APPROVED FOR PUBLIC RELEASE