[SCIP] vbcfile 4.0.1

[Marc Pfetsch]

Dear Uwe and everybody who is interested,

for the logs: There was a bug that prevented SCIP from the output of an
"A" line if the solution was found within probing, e.g. during strong
branching. The corresponding fix will be available with the next SCIP
release.

Best

Marc


On 24/11/17 16:06, Gotzes, Uwe Dr. wrote:
>> The order should be "A ..." to indicate that a solution has been found.
>> As a consequence, the bounds are sometimes updated afterwards (but not always, i.e. if the solution does not improve the bound) and a message is printed ("L ..." or "U ...").
> 
> Does that mean an L should only be printed into the vbc-output if there occurred an A prior to it?
> 
> Example:
> A,A,L,A,L is a typical sequence of A and L (the other indicator letters are omitted), while
> A,A,L,L,A should not appear in the output?
> ___________________
> 
> Attached you find a problem (solved with scip-4-0-1 and standard settings) and it's vbc-output where in line 60 the indicator L appears without a line with the indicator A prior to it.
> 
> Is this intended, or is it a bug?
> 
> If it is intended, what is the meaning of it?
> 
> Thanks and best regards
> Uwe
> 
> 
> -----Ursprüngliche Nachricht-----
> Von: Scip [mailto:scip-bounces at zib.de] Im Auftrag von Marc Pfetsch
> Gesendet: Sonntag, 19. November 2017 17:36
> An: scip at zib.de
> Betreff: Re: [SCIP] vbcfile 4.0.1
> 
> 
> 
>> okay, as I understand it, it is no longer necessary to set visual/objextern = FALSE to get a correct vbc outpout. Furthermore lines with "L..." or "U..." always concern primal bounds.
> 
> No, objextern is unrelated to this. This parameter decides whether the internal value of the objective function is used (it may have been shifted and scaled) or the "external" value, which is the one that one would expect for the original problem.
> 
> The "L ..."  and "U ..." indeed only concern primal bounds. We thought about adding output for the dual bounds, but have not yet implemented this.
> 
> Best
> 
> Marc
> 
> 
> 
> 
>> -----Ursprüngliche Nachricht-----
>> Von: Scip [mailto:scip-bounces at zib.de] Im Auftrag von Marc Pfetsch
>> Gesendet: Samstag, 18. November 2017 17:29
>> An: scip at zib.de
>> Betreff: Re: [SCIP] vbcfile 4.0.1
>>
>>
>>
>> Dear Uwe,
>>
>> the output of vbcfiles is indeed different:
>>
>> 1. The output for maximization problems should be correct now (previously it produced wrong output).
>>
>> 2. There is a message for each new solution ("A ...").
>>
>>
>> The order should be "A ..." to indicate that a solution has been found.
>> As a consequence, the bounds are sometimes updated afterwards (but not always, i.e. if the solution does not improve the bound) and a message is printed ("L ..." or "U ...").
>>
>> Best
>>
>> Marc
>>
>>
>>
>> On 16/11/17 12:01, Gotzes, Uwe Dr. wrote:
>>> Hi,
>>>
>>>  
>>>
>>> I have observed, that the vbcfile is generated in a different manner 
>>> than before in scip 4.0.1.
>>>
>>>  
>>>
>>> Here is a small example
>>>
>>>  
>>>
>>> #TYPE: COMPLETE TREE
>>>
>>> #TIME: SET
>>>
>>> #BOUNDS: SET
>>>
>>> #INFORMATION: STANDARD
>>>
>>> #NODE_NUMBER: NONE
>>>
>>> 00:00:00.17 N 0 1 3
>>>
>>> 00:00:00.17 I 1 \inode:\t1
>>> (000000000075B280)\idepth:\t0\nvar:\t-\nbound:\t-10000000000000000000
>>> 0
>>> .000000
>>>
>>> 00:00:00.18 L 10.000000
>>>
>>> 00:00:00.18 A 1 \nfound better solution: 1.000000
>>>
>>> 00:00:00.18 P 1 14
>>>
>>> 00:00:00.18 L 1.000000
>>>
>>> 00:00:00.18 P 1 11
>>>
>>> 00:00:00.18 A 1 \nfound better solution: 0.000000
>>>
>>> 00:00:00.18 P 1 14
>>>
>>> 00:00:00.18 L 0.000000
>>>
>>>  
>>>
>>> The occurrence of L and A is new, correct? I assume that A indicates, 
>>> that a better upper bound was found. Hence I would expect a line with 
>>> the indicator U (upper bound) before a line with the indicator A 
>>> appears. Or do I get something wrong?
>>>
>>>  
>>>
>>> Best
>>>
>>> Uwe
>>>
>>>  
>>>
>>>
>>>
>>> _______________________________________________
>>> Scip mailing list
>>> Scip at zib.de
>>> https://listserv.zib.de/mailman/listinfo/scip
>>>
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> https://listserv.zib.de/mailman/listinfo/scip
>>
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> https://listserv.zib.de/mailman/listinfo/scip
>>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
> 
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
>