Re: The commercial software on Constraints Programming

On Mar 17, 3:22 pm, "Neng-Fa Zhou" <n...@xxxxxxx> wrote:
21 solvers entered the second CSP solvers competition. You can check the
results and the sources of the solvers at:

http://www.cril.univ-artois.fr/CPAI06/

Cheers,
--Neng-Fa

Original question was about COMMERCIAL solvers...

A.L.

I wonder how well would the commercial solvers (ILOG, Sicstus, IF/Prolog) do
had they entered the competition. For example, could they pass the test
without giving any wrong answers and could they compete with the top rated
solvers in the competition (VALCSP in 2-ARY-EXT, buggy_2_5_s in 2-ARY-INT,
BPrologCSPSolver in GLOBAL and N-ARY-INT, and Abscon in N-ARY-EXT).

Cheers,
Neng-Fa

Hello,

I think these instances are not realistic. Most of them are small (for
quasi group completion for instance).
I don't understand why people from industry should compare their tools
on unrealistic instances.

I am the author of the alldiff constraint and of the fastest algorithm
for n-ary constraints, so works which try to implement in the best way
my work are interesting. I am just surprised to see people from
academia spending times on the implementation of algorithms developed
by people from industry :-)
I am also surprised that nobody compares the results with ILOG Solver
for instance.

I have also published several papers with several benchmarks (sports
scheduling, quasigroup completion...), so by using the instances I
used
and then by testing the solvers on them you will have your answer.

Best Regards

jcr


.

Relevant Pages