[Couenne] Objective fn value

[Pietro Belotti]

Akshay,

thanks very much. This might have to do with an old issue of Couenne.

Pietro

--
Pietro Belotti
Dept. of Mathematical Sciences
Clemson University
email: pbelott at clemson.edu
phone: 864-656-6765
web:   myweb.clemson.edu/~pbelott

On Tue, 30 Aug 2011, Akshay Gupte wrote:

> Pietro,
> I think I may have found another possible cause of this bug. After playing 
> around with Couenne options in <couenne.opt> file, I noticed that whenever I 
> removed
> lp_solver cplex
>
> the values displayed within AMPL were correct.
>
> Regards,
> Akshay Gupte
> Ph.D. student
> Operations Research
> Georgia Tech
>
>
> On 8/30/2011 12:35 AM, Pietro Belotti wrote:
>> Thanks. Unfortunately I cannot reproduce the problem: if I run the .run 
>> file attached, I get the output in the .txt attachment, and both display 
>> instructions show an objective function of -877, which seems correct. What 
>> command do you use to display the objective's value?
>> 
>> Regards,
>> Pietro
>> 
>> -- 
>> Pietro Belotti
>> Dept. of Mathematical Sciences
>> Clemson University
>> email: pbelott at clemson.edu
>> phone: 864-656-6765
>> web:   myweb.clemson.edu/~pbelott
>> 
>> On Mon, 29 Aug 2011, Akshay Gupte wrote:
>> 
>>> Pietro,
>>> I am attaching a sample model and data file. When I try to display the 
>>> objective value via AMPL, I get a return value of zero.
>>> 
>>> Regards,
>>> Akshay Gupte
>>> Ph.D. student
>>> Operations Research
>>> Georgia Tech
>>> 
>>> 
>>> On 8/25/2011 9:58 PM, Pietro Belotti wrote:
>>>> Akshay,
>>>> 
>>>> it seems this might be related to a constant value of the objective 
>>>> function, i.e., after some AMPL preprocessing the problem is of the form 
>>>> min {c: x in X}, with c a constant.
>>>> 
>>>> I'll submit a ticket. I was able to create a problem that replicates the 
>>>> behavior you describe, but can you attach a model where this happens?
>>>> 
>>>> Thanks,
>>>> Pietro
>>>> 
>>>> -- 
>>>> Pietro Belotti
>>>> Dept. of Mathematical Sciences
>>>> Clemson University
>>>> email: pbelott at clemson.edu
>>>> phone: 864-656-6765
>>>> web:   myweb.clemson.edu/~pbelott
>>>> 
>>>> On Wed, 24 Aug 2011, Akshay Gupte wrote:
>>>> 
>>>>> Hi
>>>>> I am using the stable 0.4 version. I noticed that after Couenne finished 
>>>>> it's solve, the optimal (or best found) value displayed by Couenne is 
>>>>> completely different from the value using the foll. command in Ampl
>>>>> 
>>>>> display obj.val;
>>>>> 
>>>>> This happened for many instances that I was trying. In some cases, the 
>>>>> value displayed from within Ampl was simply zero. Is it possible that 
>>>>> the link with Ampl is broken somewhere ?
>>>>> 
>>>>> Akshay Gupte
>>>>> 
>>>>> Sent from my iPhone
>>>>> 
>>>>> _______________________________________________
>>>>> Couenne mailing list
>>>>> Couenne at list.coin-or.org
>>>>> http://list.coin-or.org/mailman/listinfo/couenne
>>> 
>