MOSEK

 Top  Previous  Next

MOSEK is a solver for linear, mixed-integer linear, and convex nonlinear mathematical optimization problems developed by Erling D. Andersen and Knud D. Andersen of MOSEK ApS as described at http://www.mosek.com/.  MOSEK contains several optimization approaches designed to solve large-scale sparse problems.  The current optimizers include:

•Interior-point optimizer for all continuous problems
•Conic interior-point optimizer for conic quadratic problems
•Simplex optimizer for linear problems
•Mixed-integer optimizer based on a branch and cut technology

MOSEK can solve LP, MIP, QCP, RQCP, MIQCP, RMIP, NLP, DNLP, and RMINLP model types.  The NLP components it works on ordinarily should be highly convex as it can get stuck on non convex problems.  The solver manual is Mosek.pdf.

 

A bare bones, free version of MOSEK is available and is called COINMOSEK.

.