[Cbc] I think we found a bug in cbc: How can we report it and get fixed quickly ?

[Haroldo Gambini Santos]

Hi Thomas,

I generated a LP (attached) where columns apparently related to ym 
variables (if I understand the .col files correctly) appear also in the 
BOUNDS section of the LP as being positive.

It generated a different solution, apparently without negative values 
for ym variables.

Maybe the AMPL translated model is not properly getting the bounds. What 
if you explicitly include

  ym*   >= 0

constraints into your AMPL model and test the results ?

Cheers

On 30-09-2015 11:49, ovovis wrote:
> I generated the MPS file out of AMPL. If I understand this process 
> correctly then the variables will be abbreviated and the .col and .row 
> files, that I sent you as well, are meant to translate the mps 
> variables into the original names. Then you should have already all 
> information necessary to reproduce the original names.
>
> But maybe I misunderstood your question.
>
> Please let me know.
>
> Thomas
>
>> On Sep 30, 2015, at 4:12 PM, Haroldo Gambini Santos 
>> <haroldo.santos at gmail.com <mailto:haroldo.santos at gmail.com>> wrote:
>>
>> Hi Thomas,
>>
>> I converted your MPS to LP to take a look (don't have AMPL here).
>> Strangely no "bounds" section is generated.
>>
>> The MPS conversion also lost the original names, can yon generate 
>> including the original names so that the specific result for variable 
>> ym is visible ?
>>
>> Cheers
>>
>> On 30-09-2015 10:17, ovovis wrote:
>>> Hi Haraldo,
>>>
>>> have you had a chance to look at the lp yet ?
>>>
>>> Would be great to hear from you. Let me know if I can be of any 
>>> further help.
>>>
>>> Thomas
>>>
>>>> On Sep 28, 2015, at 3:44 PM, Haroldo Gambini Santos 
>>>> <haroldo.santos at gmail.com <mailto:haroldo.santos at gmail.com>> wrote:
>>>>
>>>> Hi Thomas,
>>>>
>>>> Can you send the LP where the problem occurs ?
>>>>
>>>> Cheers
>>>>
>>>> On 28-09-2015 01:37, ovovis wrote:
>>>>> Dear cbc team,
>>>>>
>>>>> my name is Thomas Enge from Berlin Germany. We are a small (2 
>>>>> persons) start up company (see also www.ovovis.de and then About 
>>>>> us). We use AMPL + cbc internally to provide a portfolio 
>>>>> optimization service to energy traders. So far we only ran tests 
>>>>> with our potential customers when we detected that cbc violates 
>>>>> the domain definition of variables in the solution for specific 
>>>>> problem settings. Our variables are defined as >= 0, but the 
>>>>> result also produces negative numbers for these variables. This is 
>>>>> the case for version 2.9.4 - 2.9.6.
>>>>>
>>>>> My question is how can we address this issue best to you ? And is 
>>>>> there a way to receive a quick fix or do we need to wait for the 
>>>>> next stable release ? The latter is very important for us to 
>>>>> understand since it effects our go-live date.
>>>>>
>>>>> Thanks in advance for your help
>>>>>
>>>>> Thomas Enge
>>>>>
>>>>>
>>>>> OvvvvvvvvvvvvvvvvvvvOvis
>>>>>
>>>>> ovovis GmbH
>>>>> Dr. Thomas Enge
>>>>> Stargarder Str. 57
>>>>> D-10437 Berlin
>>>>>
>>>>> M: +49 1515 910 5118
>>>>> F: +49 30 538 10 188
>>>>> E: tenge at ovovis.de <mailto:tenge at ovovis.de>
>>>>> W: www.ovovis.de <http://www.ovovis.de/>
>>>>> --------------------------------------
>>>>> Amtsgericht Charlottenburg
>>>>> HRB 116742B
>>>>> Vat-ID: DE262642185
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Cbc mailing list
>>>>> Cbc at list.coin-or.org
>>>>> http://list.coin-or.org/mailman/listinfo/cbc
>>>>
>>>> -- 
>>>> ==================================================
>>>> Haroldo Gambini Santos
>>>> D.Sc, Computer Science
>>>> Universidade Federal de Ouro Preto
>>>> http://www.decom.ufop.br/haroldo/
>>>
>>
>> -- 
>> ==================================================
>> Haroldo Gambini Santos
>> D.Sc, Computer Science
>> Universidade Federal de Ouro Preto
>> http://www.decom.ufop.br/haroldo/
>

-- 
==================================================
Haroldo Gambini Santos
D.Sc, Computer Science
Universidade Federal de Ouro Preto
http://www.decom.ufop.br/haroldo/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20150930/c8b0ca96/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: scheduledevrel.lp.tar.gz
Type: application/x-gzip
Size: 475000 bytes
Desc: not available
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20150930/c8b0ca96/attachment-0001.bin>