welcome: please sign in
location: Diff for "Participants/EZCSP"
Differences between revisions 10 and 11
Revision 10 as of 2011-05-17 10:34:20
Size: 2165
Comment:
Revision 11 as of 2011-05-17 10:55:14
Size: 2161
Comment:
Deletions are marked like this. Additions are marked like this.
Line 12: Line 12:
Marcello Balduccini <<BR>> Selim Erdogan <<BR>> Yulia Lierler  * Marcello Balduccini
* Selim Erdogan
* Yulia Lierler
Line 15: Line 17:
All the underlying solvers were used with default settings. The following table shows which underlying solver(s) were used for each benchmark.
<<BR>>
All the underlying solvers were used with default settings. The following table shows which underlying solver(s) were used for each benchmark. <<BR>>
Line 24: Line 24:

EZCSP

Association

Eastman Kodak Company
University of Kentucky

Description

Our team was interested in evaluating and comparing cutting-edge ASP and hybrid languages and solvers on challenging, industrial-sized domains. As the “core” solver we used ezcsp, which would run underlying solvers as needed.
Ezcsp is a language and solver featuring a lightweight integration of Answer Set Programming (ASP) and Constraint Programming (CP). The integration is such that different ASP and CP solvers can be freely selected according to the features of the domain. Another distinctive feature of ezcsp over similar solvers is that it supports the use of global constraints, if they are available in the CP solver selected.
Depending on the benchmark, as underlying ASP solver we used clasp, iclingo, ASPM (M.Balduccini, “A General Method To Solve Complex Problems By Combining Multiple Answer Set Programs”, in ICLP09 Workshop on Answer Set Programming and Other Computing Paradigms (ASPOCP09), 2009). For every encoding where CP constraints were used, we compared the performance of B-Prolog and Sicstus Prolog, which was straightforward because of the interchangeability of solvers in ezcsp. We ended up selecting B-Prolog in each case because of its more consistent performance.

Participants

  • Marcello Balduccini
  • Selim Erdogan
  • Yulia Lierler

System Settings

All the underlying solvers were used with default settings. The following table shows which underlying solver(s) were used for each benchmark.

Disjunctive Scheduling
Packing Problem
Magic Square
Weight-Assignment Tree
Incremental Scheduling

Clasp + B-Prolog

Reverse Folding

We use two encodings: the first fast but incomplete and the second slower but complete.
Encoding 1: Clasp
Encoding 2: Clasp + B-Prolog

Tomography

ASPM + Clasp

Airport Pickup

ASPM + Clasp + iClingo

Hydraulic Leaking Hydraulic Planning

iClingo

-- All Others --

Clasp

ASP Competition 2011: Participants/EZCSP (last edited 2011-05-23 07:29:24 by CarmenSantoro)