M.E.S.S. (Matrix Equation Sparse Solver)
Amongst other things, M.E.S.S. can solve Lyapunov and Riccati equations, and perform model reduction of systems in state space and structured differential algebraic form. M.E.S.S. has been implemented in MATLAB as well as C, with bindings also via MEX and to Python. M.E.S.S. is therefore not restricted to the solution of "academic toy problems". Several measures have been taken to enhance the computational performance of MESS routines in both implementations. To put this into the right perspective, Lyapunov equations of order 20 000 were solved by MESS within less than a minute on a regular laptop computer. On a 64bit computeserver algebraic Riccati equations of order 250 000 can be solved in well below an hour and solutions to Lyapunov equations for 3d multiphysics applications with roughly 500 000 DOFs have been computed in only a few hours. When using standard (dense) methods, supercomputers are needed to solve problems of this size in reasonable time.
The final release of the C version of M.E.S.S. is still work in progress, but the MATLAB and Octave version is available below. For citations to the Software please see the CITATION.md file in the top level directory of the corresponding download or installation.
Matlab and Octave
- M-M.E.S.S.-1.0.1 Matlab Toolbox file
(recommended for Matlab R2014a and above)
- M-M.E.S.S.-1.0.1 Octave package
(recommended for Octave 4 and above)
- M-M.E.S.S.-1.0.1 Zip Archive
- M-M.E.S.S. public git access
- Additional NSE model data Zip Archive (800 MB)
The Python package pyMESS will be released as part of the C-version. It uses the C library for fast execution and can be used together with the numpy and scipy packages.
Information about the current development are available at: https://gitlab.mpi-magdeburg.mpg.de/mess/mmess-releases
Alternatively, please send an email to the contact persons on the right if you have any questions about the current release plans and development.