[Cbc] CBC 2.9 vs. 2.8, presolving
John Forrest
john.forrest at fastercoin.com
Mon Feb 2 06:37:21 EST 2015
Christoph,
Happens with 2.8.12 (for me). It is a bad dual solve. Problem goes
away if anything changed e.g. scaling used.
I have added some code to switch back to primal if things look as if
they are going very badly.
John
On 02/02/15 08:14, Christoph Cullmann wrote:
> Hi,
>
> using the example:
>
> http://www.absint.com/dla/lpsolve/value_9041131.lp.gz
>
> and a solver like below that mail, now the presolving no longer
> terminates in any reasonable time (e.g. > 20 minutes here instead of seconds with CBC 2.8).
>
> Is this a regression or just some API misuse?
>
> Greetings
> Christoph
>
>
More information about the Cbc
mailing list