[Coin-discuss] OsiSolverInterface use and in particular OsiClpSolverInterface
JP Fasano
jpfasano at us.ibm.com
Sat Jan 4 12:02:02 EST 2003
Vivian,
Your testcase has been added to OsiSolverInterfaceTest.cpp.
Thanks for submitting it.
The following solverIntefaces work fine with the test: clp, vol, soplex,
osl.
These solverInterfaces seem to have problems with it: dylp, glpk.
I don't know about xpress or cplex solverInterfaces
JP Fasano
jpfasano at us.ibm.com
(914)945-1324 (tie line 862-1324)
COIN www.coin-or.org
Vivian De Smedt
<vivian at vdesmedt.com> To: coin-discuss at oss.software.ibm.com
Sent by: cc:
coin-discuss-admin at www- Subject: Re: [Coin-discuss] OsiSolverInterface use and in particular
124.ibm.com OsiClpSolverInterface
12/16/2002 02:58 PM
Please respond to
coin-discuss
Dear Osi,
To reply to FP Fasno mail asking complementary information about my
problem.
In my experience of OsiSolverInterface based on Cplex and Glpk interface
implementation lead me to define problem following the next recipe:
- starting with an empty model
- adding columns with no content
- adding rows with coefficient
- set the sense
In attachement you will find a program that work only if the INIT flag is
defined.
It equivalent work fine for Cplex, Glpk and Vol algorhithme (that last one
version should replace infinite bound of variable by finite one but that's
an explicit limitation of the solver).
Hoping that it help to understand what I mean in my previous mail.
Thank to every one for your reply and contributions.
Vivian De Smedt.
#### main.cpp has been removed from this note on January 04, 2003 by J P
Fasano
More information about the Coin-discuss
mailing list