[Coin-discuss] strange behavior

[Bernardo Pagnoncelli]

Hello everyone,

My name is Bernardo and I've been using OSI and SYMPHONY.
I am solving a very simple mixed Integer test problem: it has two integer
variables (B1 and B2)
and two continuous variables (x1+x2). The coefficients are random: on every
iteration
I sample them from a uniform distribution. Fine.

I ran 5 trials and in the first, second and fourth I got the correct
answers.
In the third and fifth the I got wrong answers: the objective value was too
high (90004.4)
and all continuous variables () were equal to zero event though the
objective function
is x1 + x2. Furthermore, both binary variables were equal to 7e-05 in the
optimal solution.

Then I copy the fifth problem into a separate file and ran it separately.
The solver returned the correct answer!!! Both binary were zero
and the continuous are OK.

What can possibly be going on?

Thanks in advance,
Bernardo

-- 
Bernardo Kulnig Pagnoncelli
PhD student
Dep. of Mathematics, Pontifical Catholic University of Rio de Janeiro
(PUC-Rio)
Rua Marquês de São Vicente 225 Gávea Rio de Janeiro RJ Brazil CEP: 22453-900
Tel:             55  21  3527 - 1730
Fax:            55  21  3527 - 1282
e-mail: bernardokp at gmail.com bernardo at mat.puc-rio.br
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/coin-discuss/attachments/20081218/8c33930b/attachment.html>