[Coin-discuss] OsiXpr

[Matthew Saltzman]

On Fri, 21 Jan 2005, Jesper Hansen wrote:

>
>
> We are trying the new version of OsiXpr, but we have some
> problems that we would ask you about. The first is that even though we
> set the variables to be integer and run with branchAndBound we still get
> continues variables from the optimizer.
>
> Well, to illustrate it I just wanted to create a small test problem and
> started out with the LP examples in the COIN tutorial from Lehigh. There
> is a few extra lines where the variables are set to integers and
> branchAndBound is called. The solution returned is 2,0,0.25 instead of
> 2,0,1, which is the integer solution. From the attached output it seems
> that Xpress finds the correct solution, but that a different solution is
> retreived from Osi.
>
> Any help is highly appreciated. Thanks.
>
> Jesper

Jesper-

Sorry, I should have replied sooner to your previous message.  The new 
OsiXpr is much improved over the old one, but still needs some work.  I 
plan to get to it shortly, and I'll look at this problem.  Thanks for the 
example--it will be useful.

-- 
 		Matthew Saltzman

Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs