The NEOS Server offers Bonmin (Basic Opensource
Nonlinear Mixed INteger programming)
for the solution of mixedinteger nonlinearly constrained optimization problems in AMPL format. Bonmin is also
available open source under the
Common Public License.
Bonmin was developed by a large team of researchers at IBM and
Carnegie Mellon:
 Pierre Bonami
 Lorenz Biegler
 Andrew Conn
 Gerard Cornuejols
 Ignacio Grossmann
 Carl Laird
 Jon Lee
 Andrea Lodi
 Francois Margot
 Nicolas Sawaya
 Andreas Waechter
Information about Bonmin can be found on the Bonmin homepage on
the COINOR website. Bonmin is a hybrid
between two classical algorithms for
mixedinteger nonlinear programming: an outerapproximationbased
branchandcutbased algorithm and a pure branchandbound algorithm.
The user can set options to declare which version of the algorithm
should be employed through NEOS.
Using the NEOS Server with Bonmin
The user must submit a model in AMPL
format to solve a mixedinteger nonlinearly constrained optimization problem.
Examples of models in AMPL format are available from the
MacMINLP
library.
The mixedinteger nonlinear programming problem must be specified by a
model file with the options of a data file and a commands file.
If the commands file is specified, it must contain the AMPL solve
command. The commands file must not contain the AMPL model
or data commands. The model and data files are renamed internally
by NEOS.
Bonmin has a variety of options that can be set to alter the behavior of
the algorithm. The Bonmin User's Manual describes all the solver options available to
the user. There are two ways to set options for Bonmin:
