SDPA-C (SemiDefinite Programming Algorithm - Completion method) User's Manual — Version 6.2.0

Classification: Mathematical ProgrammingAugust 2004Abstract. The SDPA-C (SemiDefinite Programming Algorithm – Completion method) is a soft-ware package designed for solving large scale sparse SDPs (semidefinite programming problems).In particular, the SDPA-C solves an SDP quite efficiently in computational time and memory ifthe aggregated sparsity pattern of the data matrices induces a sparse chordal extension of theaggregate sparsity pattern matrix; if not, the standard version of the SemiDefinite ProgrammingAlgorithm, SDPA solves the SDP faster. Using the positive definite matrix completion theory,the SDPA-C stores only sparse matrices and perform matrix computations that take advantagesof their sparsity. For theoretical and technical details of the SDPA-C, see the papers Fukuda-Kojima-Murota-Nakata (2000) and Nakata-Fujisawa-Fukuda-Kojima-Murota (2003) listed in thereferences of this manual.This manual, the SDPA-C, and the instructions for its installation can be found in the WWWsitehttp://www.is.titech.ac.jp/˜kojima/sdpa/index.html.Key words Semidefinite Programming, Interior-Point Method, Matrix Completion,Computer Software