Using SeDuMi 1.02, a MATLAB toolbox for optimization over symmetric cones-英文文献
Using SeDuMi a MATLAB to olb ox for optimization over
symmetric cones
Jos F Sturm
Communications Research Lab oratory McMaster University Hamilton Canada
Supp orted by Netherlands Organization for Scientic Research NWO
Email sturmcauchycrlmcmasterca
August
Abstract
SeDuMi is an addon for MATLAB that lets you solve optimization problems with linear
quadratic and semideniteness constraints It is p ossible to have complex valued data and
variables in SeDuMi Moreover large scale optimization problems are solved eciently by
exploiting sparsity This pap er describ es how to work with this to olb ox
KEY W ORDS Symmetric cone semidenite programming second order cone program
ming selfduality MATLAB SeDuMi
SeDuMi stands for SelfDualMinimization it implements the selfdual emb edding technique
for optimization over selfdual homogeneous cones The selfdual emb edding technique as pro
p osed Ye To dd and Mizuno essentially makes it p ossible to solve certain optimization
problems in a single phase leading either to an optimal solution or a certicate of infeasibility
Optimization over selfdual homogeneous cones or more concisely optimization over symmetric
cones was rst studied by Nesterov and To dd and is currently an active area of research
Semidenite programming is a sp ecial case of optimization over symmetric cones The p op
ular package SP by Vandenb erghe and Boyd is one of the rst software to ols that was de
velop ed for semidenite programming Some control theorists use SP indirectly via LMITOOL
by El Ghaoui Nikoukhah and Deleb ecque or MRCT by Dus