RE: [Help-glpk] DIMACS, mincost and GLPSOL

[Andrew Makhorin]

> Your example on Page 35 is even better.

That is what I meant, not that one on p.32. Sorry.

> 
> Roughly, the out-of-kilter ran about 5 times faster than the 
> simplex for the same problem, and took about 2/3rd's the memory.

I'd like to note that today the out-of-kilter algorithm is not a 
best one for solving mincost. For example, RELAX-IV developed by
Prof. Bertsekas is much faster (even faster that the network
simplex). It is free software, so I plan to translate it to C and
include in glpk as a better alternative.