[Scip] Display

michael.winkler@zib.de michael.winkler at zib.de
Mon Oct 8 02:04:53 MEST 2012


Hi Jose,

>    1. First of all, I would like to know if it is possible to disable the
>    display of the optimal primal solution at the end of each run (while
>    keeping the statistics). My problem has many ten-thousands variables
> and
>    since I am just running a timed based comparison between the models, I
> am
>    only interested in the objective rather than in the individual value of
>    each variable. At the beginning it was not an issue, however, now that
> I am
>    gathering the results it is a bit annoying to have to go through
> thousands
>    of useless lines.

It looks like you are using SCIP with the command line option "-f". If you
don't want to see the whole solution vector, you should use the "-c" flag
and add all wanted/needed SCIP commands, which you can also enter in the
interactive shell of SCIP.

e.g. scip -c "read problem.lp optimize display statistics quit"

(You can also use abbreviations(, if they are unambiguous). The above
commmand shortened: scip -c "r problem.lp o d st q" .)
(Another possibility is to write a batch file with all SCIP commands and
launch SCIP with: scip -b batchfile.)

>    2. I am interested in extracting some of the information that is
>    provided in the statistics that are displayed at the end of the run. I
> am
>    particularly interested in: The execution time, the final gap, the
> number
>    of nodes, the dual bound at the root, the total time used by the
> pricer,
>    the number of times the pricer is called, the total time used by the
>    constraint handlers that I defined, the total number of columns
> generated,
>    among others. I would like to know if there is a way (hopefully one
> that do
>    not involve many lines of code) to extract those values from within the
>    execution and thus avoid parsing the output file.

Therefor, you need to write a script extracting your needed information(,
or use corresponding interface methods). If you have some experience with
awk-scripts you can take a look into the "check.awk" inside the "check"
sub-directory in the SCIP code.

>    3. Finally,l am not sure about the differences between the verbosity
>    levels. I do not see the difference between some of those (0,1, and 2
> look
>    exactly equal).

The verbosity level 0, disables all verbosity messages. The more important
the message the lower the verbosity level. Most of the times you will not
see any difference when switching between 1 to 3. Some syntax error
messages in some reader plugin of SCIP will be printed from level 2 on.
The problem statitics after reading the original problem stating the
amount and types of variables and constraints will be printing from level
3 on.

Best, Michael


More information about the Scip mailing list