Hi Michael, thanks for the response.
> How?
> My recollection is that DW does not provide
> an instant initial feasible solution.
> It might be the case that the more pieces into which yu divvy up the problem,
> the more likely it is that the first subproblem solutions
> will correspond to a feasible for the whole problem.
> With 8 subproblems, you have to work at finding a feasible solution.
>
You're right that DW doesn't provide instant feasibility. I neglected to
mention my Phase I procedure and just started talking about DW as if it had a
feasible sol'n. But still, your question gives me a new place to bug hunt.
Maybe my Phase I isn't as robust as I'd hoped and a flaw is hidden by a large
number of subprobs and exposed by a small number of them.
I'll update the thread if I find anything new this week.
Thanks again.
_________________________________________________________________
Hotmail® has ever-growing storage! Don’t worry about storage limits.
http://windowslive.com/Tutorial/Hotmail/Storage?ocid=TXT_TAGLM_WL_HM_Tutorial_Storage1_052009
_______________________________________________
Help-glpk mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-glpk
