Members
Overall Objectives
Research Program
Application Domains
Software and Platforms
New Results
Bilateral Contracts and Grants with Industry
Partnerships and Cooperations
Dissemination
Bibliography
XML PDF e-pub
PDF e-Pub


Section: Software and Platforms

CHOCO

Participants : Nicolas Beldiceanu, Alexis de Clerq, Jean-Guillaume Fages [main developer] , Narendra Jussien [correspondant] , Arnaud Letort, Xavier Lorca [correspondant] , Thierry Petit, Charles Prud'Homme [main developer] , Remi Douence.

CHOCO is a Java discrete constraints library integrating within a same system explanations, soft constraints and global constraints (90000 lines of source code). This year developments were focussing on the following aspects:

  1. Since September 2011, we are working on a new version of the CHOCO solver. This implies a total refactoring of the source code in order to make it simpler to use and maintain. We introduce a new propagation engine framework that directly handle state-of-the-art techniques, such as advisors, propagator groups, activity-based search and explanations, to ensure a good level of efficiency, and plug a MiniZinc modeling language parser. An alpha release will be available by the beginning of 2013.

  2. In the context of the new version of the CHOCO solver we design an adaptive propagation engine to enhance performance as well as a solver independent language to write strategies for controlling the new adaptive propagation engine. The adaptive propagation engine can both deal with variable-oriented propagation engines and constraint-oriented propagation engines. It is usually accepted that there is no best approach in general and modern constraint solvers therefore implement only one.

  3. New scalable global constraints were provides both in the context of graph constraints (with also graph variables) and in the context of scheduling constraints. These constraints respectively allow to handle sparse graphs with up to 10000 vertices, and resource scheduling problems with up to one million tasks.

  4. A new global constraint called focus for concentrating high cost values motivated by several concrete examples, such as resource constrained scheduling problems with machine rentals, was introduced.

  5. The work on providing probability-based constraints to get light propagation filtering algorithm has been pursued. A particular focus has been put on calculating the probabilistic indicator for the bound-consistency propagator of an alldifferent constraint.

  6. A stable version of Choco, tagged 3.1.0, is available since September 2nd 2013. This version won two silver medals in the MiniZinc Challenge 2013 . It has been downloaded more than 4000 times between September and December 2013.

The link to the system and documentation is http://choco.emn.fr .