[Coin-discuss] inconsistant behavior.

[Matthew Saltzman]

On Fri, 24 Jan 2003, John J Forrest wrote:

> Vivian,
>
> I don't understand when you say the first version is dual infeasible as x,y
> >=0 and <=5 (by equations).  I tried Clp on Linux and I got 5,0 and 3,4 as
> solutions which is correct.

I looked at this for a while and came to the conclusion that x and y are
intended to be free and the objectives are to be maximized.  Is that
correct, Vivian?

>
> John Forrest
>
>
>
>                       Vivian.DeSmedt at aisystems.be
>                       Sent by:                                     To:       coin-discuss at www-124.southbury.usf.ibm.com
>                       coin-discuss-admin at www-124.southbury.        cc:
>                       usf.ibm.com                                  Subject:  [Coin-discuss] inconsistant behavior.
>
>
>                       01/24/2003 11:14 AM
>                       Please respond to coin-discuss
>
>
>
>
>
> Dear Osi,
>
> I continue my experiment. I don't dare modifying the code yet not being
> sure what should be the correct behavior. In the mean time I just try to
> report the differences I observe between the solvers.
>
> In the following test suite and in particular in test 5 glpk is the only
> one to behave correctly.
>
> I think
>  - cplex fail to tell the the first verion of the problem is dual
> infeasible.
>  - clp fail to tell that the first version is dual infeasible and that no
> optimal solution was found
>  - clp fail to find a solution for the second version of the problem.
>
> Vivian De Smedt.
>
>
> ps:
>
> Here are a readable description of both problems:
>
> obj: 3x + y
> 2x + y < 10
> x + 3y < 15
>
> which is dual infeasible?
>
>
> obj: x + y
> 2x + y < 10
> x + 3y < 15
>
> which have optimal (3, 4)
>
>
> _______________________________________________
> Coin-discuss mailing list
> Coin-discuss at www-124.ibm.com
> http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-discuss
>
>
> _______________________________________________
> Coin-discuss mailing list
> Coin-discuss at www-124.ibm.com
> http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-discuss
>

-- 
		Matthew Saltzman

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