Combinatorial Problem Solving
The Combinatorial Problem Solving group
conducts its research along two related tracks:
Constraint programming over finite domains. We (Beldiceanu,
Carlsson) investigate solver architectures and filtering algorithms,
using SICStus Prolog and its CLP(FD) solver as
our research platform. The research is focused on global constraints,
with the ultimate goal of developing the next generation of constraint
programming tools. We are actively pursuing the following topics:
Case studies from industrial settings involving innovative use of
constraint programming provide invaluable feedback and inspiration for
our work. In TACIT, we applied the
technology to scheduling and production planning for a steel mill. In
one study, we investigated the combination of integer programming
and constraint programming techniques to configuration problems. In
another study in the same domain, we developed and applied the new
global constraint case, with which arbitrary relations can be
defined as constraints.
- A systematic classification of global constraints into
constraint families. (Draft).
- Using this classification for designing languages for
modeling, visualization, and expressing heuristics.
- Searching for essential principles from which one can derive
several constraint propagation algorithms.
- Development of efficient constraint algorithms that reuse or
adapt existing work from data structures, graph theory, and geometry.
We (Carlsson, Danielsson, Mildner) develop and maintain
SICStus Prolog and Quintus
Prolog which run on UNIX and Windows platforms. Both Prologs
have hundreds of licenses and offer full Prolog and powerful
SICStus Prolog is our platform for both constraint programming
research and Prolog implementation research. In a recent Master's Thesis
project, the benefits of different virtual machine instruction
encodings for SICStus Prolog were investigated.
LNCS papers are © Springer-Verlag.