A declarative framework for tree decomposition powered optimization and Boltzmann sampling
copied from cf-staging / infraredInfrared allows specifying problems declaratively, which are then automatically solved by tree decomposition based efficient algorithms. It is therefore well suited for rapid prototyping and development of methods that benefit from such techniques. Infrared's solvers can either optimize or perform Boltzmann sampling based on defined features. Problems are specified using a modeling interface in Python, which also supports the definition of specific constraints, functions, and features. Infrared implements a fast and space efficient C++ engine that evaluates the constraint networks.
