Separating search and strategy in solver cooperations

Abstract

In the constraint programming community, solver cooperation is now an established paradigm that aims at using several solvers to improve some kind of limitations or inefficiency imposed by the use of a unique solver; solver cooperation applications range over several fields such as heterogeneous domain solving, heterogeneous constraint forms over the same domain, or distributed problem solving. Meanwhile, search-languages have emphasised the need to clearly separate the different steps during a constraint problem resolution. In a similar direction, this paper presents a paradigm that enables the user to properly separate computation strategies from the search phases in solver cooperations. © Springer-Verlag Berlin Heidelberg 2003.