SDPA (Semidefinite Programming Algorithm)
Katsuki Fujisawa and Masakazu Kojima
The SDPA (SemiDefinite Programming Algorithm) is an advanced
version of the PINPAL (Primitive INterior Point ALgorithm) which is
based on the interior-point algorithm given by Kojima, Shindoh and Hara
for semidefinite linear complementarity programs. The main features
of the SDPA are:
- The SDPA solves larger problems faster than the PINPAL. This is
mainly due to the computer languages used in the SDPA and the
PINPAL; the SDPA is written in C++, while the PINPAL in Mathematica.
- The SDPA utilizes the Meschach to increase the numerical stability.
- The SDPA provides some information on infeasibility of a semidefinite
program to be solved.
- The SDPA handles not only block diagonal matrices but also sparse
This is a software package which can be downloaded via anonymous ftp from
The directory is "pub/OpRes/software/SDPA".
Clicking on the postscript link below will get you the user manual.
Research Report on Information Sciences B-308,
Tokyo Institute of Technology, December, 1995.