[Bonmin] Negative cpu secs?

[Andreas Waechter]

Hi Rodrigo,

This is indeed strange.  I have never seen that either, except when Ipopt 
has been running for a very long time, because then it can happen (I 
think) that there is an overflow in the numbers in which the CPU time is 
represented.

Does that happen at the beginning of the optimization?  Which version are 
you using?  If you have an example that can be used to reproduce this 
behavior, maybe you could create a ticket for Bonmin (or Ipopt) and attach 
the files for reproducing the issue?

Thanks

Andreas

On Thu, 24 May 2007, Pierre Bonami wrote:

> Hi Rodrigo,
> I had never seen this before. Do you get it only with Bonmin or also with 
> Ipopt?
>
> Best,
> Pierre
> On May 6, 2007, at 12:32 AM, Rodrigo López Negrete de la Fuente wrote:
>
>> Hi all,
>> 
>> I've been noticing that, sometimes, ipopt returns negative time values when 
>> using bonmin. I'm not sure where the error could be, but for example I 
>> sometimes get:
>> 
>> [ sniped output ...]
>> Number of Lagrangian Hessian evaluations             = 135
>> Total CPU secs in IPOPT (w/o function evaluations)   =   -267.789
>> Total CPU secs in NLP function evaluations           =    548.166
>> 
>> EXIT: Optimal Solution Found.
>> IpOp0006I Ipopt return (for resolve): status  0, iter count  135, time 
>> 280.378
>> IpOp0009I     3        OPT         3095.02        135      280.378
>> 
>> I wish to properly report the time that the solver took, but I'm not sure 
>> if I should add the absolute values or if there really is a problem in the 
>> timing functions. Has anybody else seen this type of behavior? Any ideas?
>> 
>> Thanks,
>>   Rodrigo
>> 
>> 
>> -- 
>> Rodrigo López Negrete
>> http://muon.blogdns.org/~rush/
>> _______________________________________________
>> Bonmin mailing list
>> Bonmin at list.coin-or.org
>> http://list.coin-or.org/mailman/listinfo/bonmin
>
>