Zephod,

Indeed, this looks like an instance of cycling. However you don't
provide enough information to give you specific answers. In particular
you don't say whether you are solving a MIP or an LP and whether you
are using the standalone solver or the API. What you can try in any
case is to use the experimental long-step ratio test that was
introduced in version 4.60 (see the NEWS file) which should help with
the numerical stability of the solving process.

Assuming you are using the API (you mentioned #define statements in
your original mail) you can use something like:
smcp.r_test = GLP_RT_FLIP;
for an LP or
iocp.flip = GLP_ON;
for a MIP.

Other things to try are changing the preprocessing and scaling options.

Do let us know whether any of this helped with your problem.


Best Regards,

Chris Matrakidis

_______________________________________________
Help-glpk mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/help-glpk

Reply via email to