[Scip] how to free a problem preserving the problem data

Stefan Heinz heinz at zib.de
Wed Feb 13 20:21:53 MET 2013


Hi Andreas,

it is actually a little bit more complicated since SCIPfreeProb() also 
calls the callback method of the probdata to free it. Therefore, I 
suggested the following:

First you collect the probdata pointer
probdata = SCIPgetProbData(scip)
http://scip.zib.de/doc/html/scip_8h.shtml#aa49ecadcab6a4085933352f7a7e0750e

Then you set the probdata pointer within SCIP to NULL
SCIPsetProbdata(scip, NULL)
http://scip.zib.de/doc/html/scip_8h.shtml#ae3e190835aa7114e9557f3c44d6c0fbe

Then you call the free method.
SCIPfreeProb(scip)
http://scip.zib.de/doc/html/scip_8h.shtml#a9c7bee6cab2932a89e63bfb62053d72b

Now you place the probdata back.
SCIPsetProbdata(scip, probdata)
http://scip.zib.de/doc/html/scip_8h.shtml#ae3e190835aa7114e9557f3c44d6c0fbe

After doing this all variables and constraints are deleted and you 
probdata is still alive.

Best Stefan

On 02/12/2013 05:34 PM, Ambros Gleixner wrote:
> Hi Andreas,
>
> I think what you need is SCIPfreeProb(), see
>
> http://scip.zib.de/doc/html/scip_8h.shtml#a9c7bee6cab2932a89e63bfb62053d72b
>
> Best regards,
> Ambros
>
>
> Am 12.02.2013 15:35, schrieb Andreas Gebauer:
>> Hi all,
>>
>> I just wondered what the best way may be in order to free a SCIP problem completely (including variables and constraints) except for the problem data.
>> We try to implement a rolling solution process with a rather extensive initialization (using some self-written readers) that feeds the ProbData.
>> After reaching a (hopefully good) solution, we want to delete all variables and constraints (as they are too specific in order to use them in the next step) and free the solving process. After that, new variables and constraints are added using the problem data.
>> SCIPfreeTransform does not delete variables and constraints while SCIPfree frees all ProbData.
>> I've also experimented with SCIPcopy functionality as well as with copying probdata but I got nowhere.
>>
>> Thanks and best regards
>> Andreas
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> http://listserv.zib.de/mailman/listinfo/scip
>>



More information about the Scip mailing list