::
CSIP Solver Home Page
CSIP
is a MATLAB code to solve Convex Semi-Infinite Programming problems.
Numerical results were obtained with a subset of test problems available at the
SIPAMPL database. The subset of problems can be obtained from
here
(the .mod files are the AMPL
models for the problems).
The
implemented solver is available at the download page (it includes the test
problems in AMPL).
:: Installing
-
Simple download the ZIP file from the Download Page and extract it to a
selected location using your
favorite unzip utility.
::
Using
-
The algorithm proposal and
details are described in the paper:
Le Thi Hoai An, Mohand Ouanes, and A.I.F. Vaz, A
branch and bound approach for convex semi-infinite programming,
submitted.
-
After extracting the ZIP file your are left with a main directory named
csip and a private directory.
-
The private directory
provides the SIPAMPL-MATLAB interface functions and the AMPL test problems
database (under the subdirectory Problems). This directory is only
requested if the SIPAMPL-MATLAB interface is to be used. We provide a MEX
file for MATLAB needed to read the .nl files (gsipampl.mexw64). This
MEX file is for a Microsoft Windows operative system running a 64 bits
MATLAB version.
-
To reproduce the numerical
results, available in the previous cited paper, just run the MATLAB script
RunAll.m. Results are added to a file named Solutions.txt.
This requests the use of the SIPAMPL-MATLAB interface. Please note that the
.nl files are the ones being used by the interface.
-
For illustrative reasons we
provide the run_leon10.m MATLAB script that solves the leon10
problem.
::
Go to Download Page
|