DAOOPT: Distributed AND/OR Optimization. An implementation of sequential as well as distributed AND/OR Branch-and-Bound and its Breadth-Rotating AND/OR Branch-and-Bound enhancement for combinatorial optimization problems expressed as MPE (max-product) queries over graphical models like Bayes and Markov networks. Also implements the following: full context-based caching. mini-buckets for heuristic generation. minfill heuristic to find variable orderings. limited discrepancy search to quickly find initial solution. stochastic local search to quickly find initial solution (via GLS+ code by Frank Hutter). stochastic iterative greedy variable ordering (code by Kalev Kask). join graph cost-shifting techniques (MPLP, JGLP, MBE-MM) (code by Alex Ihler).
Keywords for this software
References in zbMATH (referenced in 3 articles )
Showing results 1 to 3 of 3.
- Hurley, Barry; O’Sullivan, Barry; Allouche, David; Katsirelos, George; Schiex, Thomas; Zytnicki, Matthias; de Givry, Simon: Multi-language evaluation of exact solvers in graphical model discrete optimization (2016)
- Otten, Lars; Dechter, Rina: Anytime AND/OR depth-first search for combinatorial optimization (2012)
- Marinescu, Radu; Dechter, Rina: AND/OR branch-and-bound search for combinatorial optimization in graphical models (2009)