[SCIP] about providing the solution at SCIP_STAGE_PROBLEM

Feng Gao fgao784 at gmail.com
Fri Apr 29 11:57:11 CEST 2016


Hi Jacob,

Thank you for your direction. I tried the debug mode. However before I 
tried in debug mode, I revised the code. Now the error disappear. I 
cannot remember what I did. But I think it relates to the initialization 
of Bool* stored.

In debug mode, now I got a assertion "SCIPcreateCons: Assertion `name != 
((void *)0)' failed." However I checked all the places I call 
SCIPcreateCons(...), the name are all assigned a specific char* or 
char[]. I do not know why this assertion appears. Certainly, it is 
another question which is not suitable to discuss under this topic. But 
any help with this problem is also appreciated.

Best,

Feng

On 04/29/2016 03:03 AM, Jakob Witzig wrote:
> Hi Feng Gao,
>
> please compile SCIP in debug mode with 'make OPT=dbg' and run your 
> code again. In debug mode all assertions are enabled and you can see 
> which assertion fail before the segfault happens. If no assertion 
> fails, please have a look into the code where the segfault comes from. 
> For this you can use a debugger like gdb: 'gdb --args <your scip call>'.
>
> Cheers,
> Jakob
>
> Am 29.04.2016 um 04:18 schrieb Feng Gao:
>> Hi All,
>>
>> I am trying to provide the solution at SCIP_STAGE_PROBLEM, i.e., before
>> I call SCIPsolve(scip).
>> The problem is after I call SCIPcreateSol(...) and SCIPsetSolVal(...),
>> the program gives "Segmentation fault" when I call SCIPaddSol(...),
>> SCIPcheckSol(...), SCIPcheckSolOrig(...), SCIPtrySol(...), or
>> SCIPtrySolFree(...).
>>
>> I am quite sure that the solution I provide is feasible. However, when I
>> call these functions, they all give me only "Segmentation fault" but
>> nothing else.
>>
>> What could go wrong when I set the solution? I would be very appreciated
>> if anyone could give me a clue.
>>
>> Best,
>>
>> Feng Gao
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> http://listserv.zib.de/mailman/listinfo/scip
>
>



More information about the Scip mailing list