PENSDP is a computer program for solving optimization problems with linear objective and linear matrix inequality constraints (linear semidefinite programming problems). It is aimed at large-scale dense and sparse SDP problems. Current version 2.2.
PENSDP was developed by Michal Kočvara and Michael Stingl.
- precompiled libraries for Windows 64, MacOS 64 (Intel and ARM) and Linux 64bit in directory bin
- User's guide for installation and use in directory doc. Please read this first.
- stand-alone program with input file in the sparse SDPA input format
- MATLAB function with PEN or YALMIP interface (SDPA-to-PEN interface available)
- function called from a C or Fortran program with data communication through parameters
- aimed at large scale problems
- efficient treatment of different sparsity patterns in problem data
- detecting infeasibility (MATLAB version)
- Hessian-free version, resulting in significant memory savings
- Iterative solution of the Newton system (optional)
- Stopping criteria based on the DIMACS error measures, resulting in more reliable computations
- Hybrid mode for the solution of the Newton system
- control theory
- relaxations of integer programming problems
- structural optimization
- optimization of eigenvalues
- chemical engineering
M. Kočvara and M. Stingl. On the solution of large-scale SDP problems by the modified barrier method using iterative solvers. Mathematical Programming Series B, 109(2-3):413-444, 2007.