[Lemon-user] LEMON min-cost solver using DIMACS format

[Kovács Péter]

Dear All,

As Alpar said, the minimum cost flow implementations in LEMON were 
thoroughly compared to other available solvers. You can find a 
presentation about it here:
http://people.inf.elte.hu/kpeter/download/mcf_presentation_icai_2010.pdf
The last section presents the experimental results.

In short, our network simplex implementation is much better than ZIB MCF 
code, and our cost scaling algorithm is comparable or sometimes faster 
than Golberg's CS2 and the LEDA library.

The solver of CLP was not checked, but we will probably test it soon. If 
you happen to make such comparisons, we are pleased to hear the results 
here.

Best regards,
Peter


2010.02.16. 5:59 keltezéssel, Alpár Jüttner írta:
 > I don't know about a comparison with COIN/CLP, but it has thoroughly
 > tested with the best available MCF solvers and it is comparable or many
 > cases better than those.
 >
 > Peter may give more details on this.
 >
 > Regards,
 > Alpar
 >
 >
 >
 > On Mon, 2010-02-15 at 15:01 -0500, Matthew Galati wrote:
 >> Can someone show me an example of reading in a DIMACS standard
 >> min-cost NF format and solving with LEMON's network simplex?
 >>    http://people.brunel.ac.uk/~mastjjb/jeb/orlib/netflowinfo.html
 >>
 >> Does anyone have any benchmark studies comparing the MCF solver in
 >> LEMON versus any others? or even versus using standard dual simplex
 >> from COIN/CLP?
 >>
 >> Thanks in advance!
 >> Matt G
 >
 > _______________________________________________
 > Lemon-user mailing list
 > Lemon-user at lemon.cs.elte.hu
 > http://lemon.cs.elte.hu/mailman/listinfo/lemon-user