[Cbc] Different optimization results

[Juhani Miettunen]

Oops. I have misunderstood term +2 x x. That means +2 x +1 x, not x^2.

I could I pass quadratic constraints or quadratic objective to CBC in lp
format?

BR, Juhani



*From:* Juhani Miettunen <juhani.miettunen at piimega.fi>
*Sent:* 21. elokuutata 2019 15:05
*To:* 'cbc at list.coin-or.org' <cbc at list.coin-or.org>
*Subject:* Different optimization results



Hello,

I have one very simple example problem in lp format. I get two different
optimization results from different solvers.

CBC and Cplex give same result which is actually not optimal. More optimal
result is achieved using SCIP. Also CBC gives same result using AMPL model.

Lp model needed to be modified a little for CBC due to quadrative
constraints. Models attached.

One strange thing with CBC (and Cplex) is, that objective value is not
correct. Correct delta_squares value calculated from the optimization
result is 112350.85 and the reported value is 107553.69.

SCIP and CBC with AMPL give objective value 110442.85, which is more
optimal than the result above.

SCIP (and CBC with AMPL) result:

x$30                                                8
(obj:0)

x$50                                                1
(obj:0)

x$60                                               11
(obj:0)

delta_squares                               110442.85   (obj:1)



CBC (and Cplex) result:

delta_squares            107078.000000

x$60                         13.000000

x$40                          2.000000

x$30                          5.000000

and the real objective value is 112350.85

Any idea what is going on?

BR, Juhani
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20190821/03e268b9/attachment.html>