Welcome to the Cbc home page
Note that these project webpages are based on Wiki, which allows webusers to modify the content to correct typos, add information, or share their experience and tips with other users. You are welcome to contribute to these project webpages. To edit these pages or submit a ticket you must first register and login.
Introduction
Cbc (Coin-or branch and cut) is an open-source mixed integer programming solver written in C++. It can be used as a callable library or using a stand-alone executable. It can be called through AMPL (natively), GAMS (using the links provided by the Optimization Services and GAMSlinks projects), MPL (through the CoinMP project), AIMMS (through the AIMMSlinks project), PuLP, or CMPL.
Cbc links to a number of other COIN projects for additional functionality, including:
- Clp (the default solver for LP relaxations)
- Cgl (for cut generation)
- CoinUtils (for reading input files and various utilities)
For more information on supported platforms, links to dependent projects, current version, and more, click here
Cbc is developed by John Forrest, now retired from IBM Research. The project is currently managed by John Forrest, Ted Ralphs, and the rest of the Cbc team.
News
- The latest stable version of Cbc is
.
- The latest release version of Cbc is
.
- Click here to see the current change log.
December 26, 2013
- Cbc 2.8.8 has been released.
November 23, 2013
- Cbc 2.8.7 has been released.
November 14, 2013
- Cbc 2.8.6 has been released.
October 15, 2013
- Cbc 2.8.5 has been released.
September 13, 2013
- Cbc 2.8.4 has been released.
July 16, 2013
June 14, 2013
- Cbc 2.8.2 has been released.
April 26, 2013
April 10, 2013
- Cbc 2.8.0 has been released.
- Introduced new secondaryStatus 8 to indicate that solving stopped due to an iteration limit.
- Solution pool is now accessible via the command line and the CbcMain* interface.
- New mipstart option to read an initial feasible solution from a file. Only values for discrete variables need to be provided.
- Added Proximity Search heuristic by Fischetti and Monaci (off by default):
The simplest way to switch it on using stand-alone version is "-proximity on".
Proximity Search is the new "No-Neighborhood Search" 0-1 MIP refinement heuristic recently proposed by Fischetti and Monaci (2012). The idea is to define a sub-MIP without additional constraints but with a modified objective function intended to attract the search in the proximity of the incumbent. The approach works well for 0-1 MIPs whose solution landscape is not too irregular (meaning the there is reasonable probability of finding an improved solution by flipping a small number of binary variables), in particular when it is applied to the first heuristic solutions found at the root node. - An implementation of Zero-Half-Cuts by Alberto Caprara is now available.
By default, these cuts are off. To use add to your command line -zerohalfCuts root (or other options) or just -zero. So far, they may help only on a small subset of problems and may need some tuning.
The implementation of these cuts is described in
G. Andreello, A. Caprara, and M. Fischetti
"Embedding Cuts in a Branch and Cut Framework: a Computational Study with {0,1/2}-Cuts"
INFORMS Journal on Computing 19(2), 229-238, 2007
http://dx.doi.org/10.1287/ijoc.1050.0162 - An alternative implementation of a reduce and split cut generator by Giacomo Nannicini is now available.
By default, these cuts are off. To use add to your command line -reduce2AndSplitCuts root (or other options).
The implementation of these cuts is described in
G. Cornuejols and G. Nannicini
"Practical strategies for generating rank-1 split cuts in mixed-integer linear programming"
Mathematical Programming Computation 3(4), 281-318, 2011
http://dx.doi.org/10.1007/s12532-011-0028-6 - An alternative robust implementation of a Gomory cut generator by Giacomo Nannicini is now available.
By default, these cuts are off. To use add to your command line -GMI root (or other options).
The implementation of these cuts is described in
G. Cornuejols, F. Margot, and G. Nannicini
"On the safety of Gomory cut generators"
http://faculty.sutd.edu.sg/~nannicini/index.php?page=publications - To encourage the use of some of the more exotic/expensive cut generators a parameter -slowcutpasses has been added.
The idea is that the code does these cuts just a few times - less than the more usual cuts. The default is 10. The cut generators identified by "may be slow" at present are just Lift and project and ReduceAndSplit (both versions). - Allow initialization of random seed by user. Pseudo-random numbers are used in Cbc and Clp. In Clp they are used to break ties in degenerate problems, while in Cbc heuristics such as the Feasibility Pump use them to decide whether to round up or down. So if a different pseudo-random seed is given to Clp then you may get a different continuous optimum and so different cuts and heuristic solutions. This can be switched on by setting randomSeed for Clp and/or randomCbcSeed for Cbc. The special value of 0 tells code to use time of day for initial seed.
- Building on this idea, Andrea Lodi, Matteo Fischetti, Michele Monaci, Domenico Salvagnin, Yuji Shinano, and Andrea Tramontani suggest that this idea be improved by running at the root node with multiple copies of solver, each with its own different seed and then passing in the solutions and cuts so that the main solver has a richer set of solutions and possibly stronger cuts. This is switched on by setting -multipleRootPasses. These can also be done in parallel.
- Few changes to presolve for special variables and badly scaled problems (in CoinUtils).
- New option -extraVariables <number> which switches on a trivial re-formulation that introduces extra integer variables to group together variables with same cost.
- For some problems, cut generators and general branching work better if the problem would be infeasible if the cost is too high. If the new option -constraintFromCutoff is set, the objective function is added as a constraint which rhs is set to the current cutoff value (objective value of best known solution).
- Click here to see changes from release 2.7.8
November 21, 2012
- Cbc 2.7.8 has been released.
June 7, 2012
- Cbc 2.7.7 has been released.
February 10, 2012
- Cbc 2.7.6 has been released.
November 3, 2011
- Cbc 2.7.5 has been released.
October 22, 2011
October 16, 2011
- Cbc 2.7.3 has been released.
September 20, 2011
- Cbc 2.7.2 has been released.
- Allow row/column names for GMPL models
- Added CbcModel::haveMultiThreadSupport() to indicate whether Cbc library has been compiled with multithread support
- Added CbcModel::waitingForMiniBranchAndBound() to indicate whether sub-MIP heuristic is currently running
- Cbc shell should work with readline if configured with --enable-gnu-packages
- Support for compressed input files (.gz, .bz2) is now enabled by default
- Fix problems with relative gap tolerance > 100% and further bugs
- Fixes for MSVC++ Version 9 files
- Minor fixes in buildsystem; update to BuildTools 0.7.1
- Click here to see changes from release 2.7.1
July 22, 2011
- Cbc 2.7.1 has been released.
July 18, 2011
- Cbc 2.7.0 has been released.
- License has been changed to the EPL.
- Support for MSVC++ version 10 added.
- Support for BuildTools version 0.7 to incorporate recent enhancements, including proper library versioning in Linux, prohibiting installation of private headers, etc.
- Updating externals to new stable versions of dependent projects.
- Improvements to heuristics.
- New options for cut generation.
- Improved reporting of results.
- Improvements to documentation.
- Minor bug fixes.
January 4, 2011
Background/Download
Cbc is written in C++ and is released as open source code under the Eclipse Public License (EPL). It is available from the COIN-OR initiative. The code has been written by primarily by John J. Forrest.
You can obtain the Cbc source code either via subversion or in form of nightly generated tarballs. The recommended method is to use subversion because it makes it easier to obtain updates. The following commands may be used to obtain and build Cbc from the source code using subversion:
- svn co https://projects.coin-or.org/svn/Cbc/stable/2.8 coin-Cbc
- cd coin-Cbc
- ./configure -C
- make
- make test
- make install
Step 1 issues the subversion command to obtain the source code. One should check page https://projects.coin-or.org/Cbc/browser/stable to find out the latest stable version number. Alternatively one can obtain the source code from the tarball directory.
Step 3 runs a configure script that generates the make file.
Step 4 builds the Cbc library and executable program.
Step 5 builds and runs the Cbc unit test program.
Step 6 Installs libraries, executables and header files in directories coin-Cbc/lib, coin-Cbc/bin and coin-Cbc/include.
The BuildTools project has additional details on downloading, building, and installing.
The Binary project provides a downloadable binary distribution of Cbc.
Cbc is available in Debian and Ubuntu.
Doxygen Documentation
If you have Doxygen available, you can build the html documentation by typing
make doxydoc
in the directory coin-Cbc. Then open the file coin-Cbc/doxydoc/html/index.html with a browser. Note that this creates the documentation for the Cbc package. If you prefer to generate the documentation only for a subset of these projects, you can edit the file coin-Cbc/doxydoc/doxygen.conf to exclude directories (using the EXCLUDE variable, for example).
If Doxygen is not available, you can use the link to the Cbc html documentation listed below.
Documentation
- See the current README and INSTALL files to get started quickly.
- User's Guide (single page format)
- Cbc html documentation
- FAQ (Frequently Asked Questions)
- Lessons from the trenches
- Source code examples
- Success Stories
- Cbc Papers
- Cbc undocumented features.
- Setting up a Visual Studio project to use CBC
Project Links
