The MOSEK optimization software is designed to solve large-scale mathematical optimization problems. MOSEK main features are listed below. For additional questions, contact our support or browse the online documentation.
Problem types MOSEK can solve
- Conic quadratic.
- Semi-definite (Positive semi-definite matrix variables).
- Quadratic and quadratically constrained.
- General convex nonlinear.
- Mixed integer linear, conic and quadratic.
- Problem size limited only by the available memory.
- Primal and dual simplex optimizers for linear programming.
- Highly efficient pre-solver for reducing problem size before optimization.
- Branch&bound&cut algorithm for mixed integer problems.
Strengths and features of MOSEK
- The strongest point of MOSEK is its state-of-the-art interior-point optimizer for continuous linear, quadratic and conic problems.
- Exploits hardware i.e. SSE2 instructions available in recent Intel CPUs.
- The optimizer is parallelized and capable of exploiting multiple CPUs/cores.
- The optimizer is run-to-run deterministic.
- Reads and writes industry standard formats such as the MPS, CBF and LP formats.
- Includes tools for infeasibility diagnosis, repair and sensitivity analysis for linear problems.
- Ships with an optimization server for remote optimization.
- Optimizer API: C, Java, .NET, Python.
- Fusion API: C++, Java, .NET, Python, MATLAB.
- MOSEK optimization toolbox for MATLAB.
- R package.
- Command line interface.
- Other third party commercial and open-source tools and products have interfaces to MOSEK.
- OSX, Linux and Windows running on Intel X86 based CPUs. See the download section for details.
Upcoming changes appearing in version 9
- MOSEK can handle the exponential and power cones.
- The general convex optimizer is removed. We recommend that convex problems are reformulated on conic form which is almost always possible.
- The so called gpopt and scopt interface are removed. Those problems can always be reformulated as conic problems.
- The Fusion API for MATLAB is deprecated.
- MOSEK can not solve nonconvex problems. Only convex problems including one or more integer constrained variables.
- MOSEK has no sequential quadratic optimizer because it is not competitive with the algorithms implemented in MOSEK.
Try our Remote Optimization Server!
The Optimization Server (OptServer) is a MOSEK service for executing optimization tasks on a remote machine, including job scheduling, user management and other features. See documentation for details.