Accession Number : AD0770399

Title :   Dual Approaches to Quadratically Constrained Quadratic Programming,

Descriptive Note : Research rept.,

Corporate Author : FLORIDA UNIV GAINESVILLE DEPT OF INDUSTRIAL AND SYSTEMS ENGINEERING

Personal Author(s) : Hearn,Donald W. ; Randolph,William D.

Report Date : SEP 1973

Pagination or Media Count : 33

Abstract : For a convex quadratically constrained quadratic program with n variables and m constraints it is shown using generalized inverses that the Wolfe dual can be reduced to a concave problem explicitly in the m dual variables. Peterson and Ecker, and Rockafellar, have developed a conjugate dual in (n+1)(m+1) variables. The authors show that these duals are derivable from each other. Computationally it appears that the explicit form of the Wolfe dual is the more tractable approach, especially when the primal objective is strictly convex and in certain location problems. Computational experience shows that problems up to 40 primal variables and 30 constraints can be solved in less than two minutes of computer time. (Author)

Descriptors :   *QUADRATIC PROGRAMMING, *CONVEX SETS, MATRICES(MATHEMATICS), COMPUTATIONS, OPTIMIZATION, THEOREMS

Subject Categories : Operations Research

Distribution Statement : APPROVED FOR PUBLIC RELEASE