[Scip] Cutoff node being solved again?

何河 hehe.rebecca at gmail.com
Sun Oct 19 05:35:51 CEST 2014


Ah... Sorry for the "spamming".. I just found the number after "Processing
node" does not refer to the node number. My bad. The cutoff node is *not*
solved again. However, I'd appreciate if anyone could tell me why it's
repeatedly solving some part? Thanks!!

He

On Sat, Oct 18, 2014 at 11:15 PM, 何河 <hehe.rebecca at gmail.com> wrote:

> Hi there,
>
> Based on some heuristics, I called SCIPcutoffNode after a node was
> selected. However I found it was solved again soon. For example, the debug
> message shows:
>
> [src/scip/solve.c:3593] debug: Processing node 1 in depth 0, 0 siblings
> [src/scip/solve.c:4508] debug: Processing of node 1 in depth 0 finished. 0
> siblings, 2 children, 0 leaves left
> [src/scip/solve.c:3593] debug: Processing node 2 in depth 1, 0 siblings
> [src/scip/solve.c:4508] debug: Processing of node 2 in depth 1 finished. 0
> siblings, 2 children, 0 leaves left
> [src/scip/solve.c:3593] debug: Processing node 3 in depth 2, 0 siblings
> [src/scip/solve.c:4508] debug: Processing of node 3 in depth 2 finished. 0
> siblings, 0 children, 0 leaves left
> [src/scip/solve.c:3593] debug: Processing node 1 in depth 0, 0 siblings
> [src/scip/solve.c:4508] debug: Processing of node 1 in depth 0 finished. 0
> siblings, 2 children, 0 leaves left
> [src/scip/solve.c:3593] debug: Processing node 2 in depth 1, 0 siblings
> [src/scip/solve.c:4508] debug: Processing of node 2 in depth 1 finished. 0
> siblings, 2 children, 0 leaves left
> [src/scip/solve.c:3593] debug: Processing node 3 in depth 2, 0 siblings
> [src/scip/solve.c:4508] debug: Processing of node 3 in depth 2 finished. 0
> siblings, 2 children, 0 leaves left
>
> The log I got doesn't seem to indicate there's a restart.. Why node 1, 2,
> 3 is solved repeatedly? Thank you very much!!
>
> He
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20141018/07c09ea8/attachment.html>


More information about the Scip mailing list