Constraint-based Very Large-Scale Neighborhood search

摘要

Very Large-Scale Neighborhood (VLSN) search is the idea of using neigh-borhoods of exponential size to find high-quality solutions to complex optimization problems efficiently. However, so far, VLSN algorithms are essentially described and implemented in terms of low-level implementation concepts, preventing code reuse and extensibility which are trademarks of constraint-programming systems. This paper aims at remedying this limitation and proposes a constraint-based VLSN (CB-VLSN) framework to describe VLSNs declaratively and compositionally. Its main innovations are the concepts of cycle-consistent MoveGraphs and compositional moves which make it possible to specify an application in terms of constraints and objectives and to derive a dedicated VLSN algorithm automatically. The constraint-based VLSN framework has been prototyped in Comet and its efficiency is shown to be comparable to dedicated implementations.