B3M35HYS, BE3M35HYS - B251: Software

Complementarity in modeling dynamical systems

When it comes to dynamical systems, tt appears that there is currently just one well-developed and maintained software for complementarity systems.

SICONOS – free and open source, C++ and Python. Motivated by non-smooth dynamical systems. It also contains some optimization functionality.

Complementarity in numerical optimization

When it comes to optimization, for solving an LCP problem in Matlab (using their Optimization Toolbox), there is not much more we can use than just the general nonlinear optimize fmincon.

But there are some third party solvers that do offer such functionality:

PATH solver: https://pages.cs.wisc.edu/~ferris/path.html (academic license).
in Tomlab (yet another commercial optimization toolbox for Matlab): https://tomopt.com/docs/quickguide/quickguide029.php
in Knitro (Artelys, commercial, can be interfaced from Matlab): https://www.artelys.com/docs/knitro/2_userGuide/complementarity.html
in Mosek (commercial, can be interfaced from Matlab): https://docs.mosek.com/10.0/capi/tutorial-djc-optimizer.html
in Yalmip: can be modelled through complements.

Finally, for those who have not yet explored Julia programming language (and ecosystem), note that support for complementarity problems is available in JuMP – the flagship package for modeling optimization problems in Julia: https://jump.dev/JuMP.jl/stable/manual/constraints/#Complementarity-constraints. The syntax is particularly neat, check it out.

Naposledy změněno: středa, 29. listopadu 2023, 10.13