A multi-paradigm tool for large neighborhood search

Abstract

Studies in Computational Intelligence Volume 434, 2013, Pages 389-414 A multi-paradigm tool for large neighborhood search (Article) Cipriano, R.a , Di Gaspero, L.b , Dovier, A.a a Dipartimento di Matematica e Informatica, Università Degli Studi di Udine, via delle Scienze 208, I-33100 Udine, Italy b Dipartimento di Ingegneria Elettrica, Gestionale EMeccanica, Università Degli Studi di Udine, via delle Scienze 208, I-33100 Udine, Italy View references (21) Abstract We present a general tool for encoding and solving optimization problems. Problems can be modeled using several paradigms and/or languages such as: Prolog, MiniZinc, and GECODE. Other paradigms can be included. Solution search is performed by a hybrid solver that exploits the potentiality of the Constraint Programming environment GECODE and of the Local Search framework EasyLocal++ for Large Neighborhood Search. The user can modify a set of parameters for guiding the hybrid search. In order to test the tool, we show the development phase of hybrid solvers on some benchmark problems. Moreover, we compare these solvers with other approaches, namely a pure Local Search, a pure constraint programming search, and with a state-of-the-art solver for constraint-based Local Search.