[Coin-discuss] Bonmin and Optimization Services

[Kipp Martin]

Hi:

The Optimization Services (OS) project, 
https://projects.coin-or.org/OS,  is now configured to work with the 
COIN-OR Bonmin nonlinear integer solver.  This feature is available in 
the trunk version of OS.  It has been tested on various flavors of 
Linux, Windows using the Microsoft cl compiler (with Intel ifort to 
build the necessary Fortran libraries), Windows Cygwin (gcc), Windows 
MinGW, and Mac OS X.

There are four ways to use the Bonmin solver with OS.

1) Solve problems locally using the OSSolverService executable.

For example, to solve a problem in the OS XML format with the 
OSSolverService executable

OSSolverService  -osil ../data/osilFiles/bonminEx1.osil  -solver bonmin

The OSSolverService can also read Ampl nl files:

OSSolverService  -nl ../data/amplFiles/bonminEx1.nl -solver bonmin


2) Build the model instance using the OSInstance API.  The OSInstance 
API provides an interface with methods to build a general nonlinear 
integer program in-memory.  See the OS User's Manual and the example 
problem in

OS/examples/instanceGenerator

Once an in-memory osinstance object is available it is trivial to invoke 
Bonmin.

DefaultSolver *solver = new BonminSolver();
solver->osinstance = osinstance;
solver->buildSolverInstance();
solver->solve();


3) Call a remote solver server running Bonmin.  Examples of how to call 
a remote solver service using OS Web services is given in 
OS/examples/osRemoteTest.


4)  Use either the AMPL or GAMS modeling language. If AMPL is used, 
building the OS project provides an executable OSAmplClient that appears 
to AMPL as a solver. The OSAmplClient can be invoked directly inside 
AMPL to solve a problem with Bonmin.  It is also possible to build the 
model using the GAMS modeling language and then use GAMSlinks (see 
https://projects.coin-or.org/GAMSlinks) to call the OS supported solvers 
either locally or remotely.

The COIN-OR OS project now directly supports the following solvers:

Bonmin
Cbc
Clp
Cplex
DyLP
Glpk
Ipopt
Lindo
SYMPHONY
Vol

In addition, the Mosek ApS optimization solver and the Frontline Systems 
Solver Platform SDK currently support OSiL for problem instance 
representation of mixed integer linear programs.

See the User's Manual in the OS/doc folder and at

http://www.coin-or.org/OS/doc/osUsersManual_1.1.pdf

for more information on how to use the OS project.


Thank you,

-- 
Kipp Martin
Visiting Professor
University of Cincinnati
College of Business
Dept. of QAOM
525 Lindner Hall
Cincinnati, OH  45221-0130

513-556-6834

kipp.martin at chicagogsb.edu
http://homepages.uc.edu/~martinrk/
https://projects.coin-or.org/OS
http://www.coin-or.org