This project is under development and, althought useful, must be reviewed to be ready for anyone to use.
METSlib is an OO (Object Oriented) metaheuristics optimization framework in C++ designed to make easy implementing or adapting models. The model is modular: all the implemented search algorithms can be applied to the same model.
Once the model is implemented one can test some Tabu Search Strategies or Simulated Annealing or other local search based search strategies.
METSlib is Open Source and is covered by the GNU GPL license or, at your option, by the Common Public License 1.0. Commercial licensing or custom development can be considered by the author upon request.
METSlib implements the basics of some metaheuristics algorithm:
* Random Restart Local Search
* Variable Neighborhood Search
* Iterated Local Search
* Simulated Annealing (with linear, exponential and custom cooling
schedule)
* Tabu Search.
In its current state, the main focus of the library is on Tabu Search, other algorithm may have some rough edges.
To use any algorithm you must implement an objective function, a neighborhood (move manager) and some moves.
For each algorithm you are free to use some of the already implemented termination criteria and/os cooling schedule (for SA) and/or tabu lists and aspiration criteria (for TS), or you can implement you own specialized versions.
The object oriented structure was carefully thought out to be highly reusable and the algorithm where reviewed by more than one person and more than one time.
This OO library was inspired by the OTS library released by the Coin-OR project.
You may want to read https://github.coin/coin-or/metslib
- https://coin-or.github.io/metslib/docs/stable/0.5/html/
A METSlib Tutorial is available here:
You can also look in the distributed examples in project https://github.com/coin-or/metslib-examples.
This library was used in different projects by different developers and proved quite useful to implement a new model (or to adapt an old one) without re-inventing the wheel each time you need to change the algorithm or the neighborhood. At the same time it's young so I'll be happy to help you while implementing the nuts and bolts of your own model since documentation is still a work in progress.
Feel free to write an email to mirko.maischberger@gmail.com.
Join the mailing list http://list.coin-or.org/mailman/listinfo/metslib and start contributing
I'd like to thank all the people at the Global Optimization Laboratory (http://gol.dsi.unifi.it) for continuous support and the many useful discussions, in particular Fabio Schoen, Bernardetta Addis and Paola Cappanera.
The automake template was borrowed from some projects by Murray Cumming murrayc@usa.net and Cedric Gustin gustin@pcpm.ucl.ac.be
- Mirko Maischberger, "COIN-OR METSlib, a Metaheuristics Framework
in Modern C++" April 2011
BibTex Entry:
@misc{metslib2011, author={Mirko Maischberger}, title={{COIN-OR} {METSlib}, a Metaheuristics Framework in Modern C++}, month={April}, year={2011}, howpublished = {https://coin-or.github.io/metslib/docs/stable/0.5/metslib-tr.pdf}, }
- Mirko Maischberger