Discussion:
Matlab vs. CPLEX
(too old to reply)
tiger
2005-01-16 23:22:01 UTC
Permalink
I am new in the OR area. I have to deal with a linear programming
problem. The researchers in my field used CPLEX to solve the problem. I
wonder if I can use MATLAB to do the same thing ( I already installed
the operation research toolbox) because I cannot access to the CPLEX.
Can anyone help me to see if I can still use the MATLAB to solve the
linear programming problem? Are there any significant difference
between MATLAB and CPLEX in terms of their limitations or other stuff?
Thanks.
hansm
2005-01-17 00:06:01 UTC
Permalink
Hi,
one quite important point: you did not say if the LP is continuous or
mixex-integer. LINPROG in the optimization toolbox is OK for the former
but you can use free codes with Matlab interface, see the LP section of
http://plato.asu.edu/guide.html
for example, QSopt. For MIP or MILP it is a little harder to find
competitive alternatives, but see again the same source. Finally, you
can use NEOS
http://neos.mcs.anll.gov/
For comparisons of codes see the benchmark section of the first link.
Hans Mittelmann
Kent Paul Dolan
2005-01-17 22:00:07 UTC
Permalink
Finally, you can use NEOS
http://neos.mcs.anll.gov/
You meant:

http://neos.mcs.anl.gov

[My daughter used to run the supercomputer
site there, including NEOS, is how I know.]

xanthian.
--
Posted via Mailgate.ORG Server - http://www.Mailgate.ORG
tiger
2005-01-18 21:45:12 UTC
Permalink
Hi,
My problem is 0/1(binary) linear programming. Can I use MATLAB or NEOS
like you said?
Thank you very much.
hansm
2005-01-19 00:41:42 UTC
Permalink
Hi,
good that you know specify your problem more. The 0/1 case cannot
really be solved any more easily than the pure integer case and most
software even allows mixed integer (also real variables).
You can use, for example, GLPK with its GLPKmex Matlab interface. On
NEOS you can use in principle all solvers in the Mixed integer LP
section. Note that NEOS-GLPK also allows Matlab input.

Hans Mittelmann
tiger
2005-01-20 16:58:48 UTC
Permalink
HI Hans,
Thank you so much for provide me this information. I visited that
website of NEOS and found it is really great ! The GLPK also provides
Matlab interface. I think I will try to learn how to use it and use it
for my dissertation. I really really appreciate your help.
edadk
2005-01-17 10:27:38 UTC
Permalink
Post by tiger
I am new in the OR area. I have to deal with a linear programming
problem. The researchers in my field used CPLEX to solve the problem. I
wonder if I can use MATLAB to do the same thing ( I already installed
the operation research toolbox) because I cannot access to the CPLEX.
Can anyone help me to see if I can still use the MATLAB to solve the
linear programming problem? Are there any significant difference
between MATLAB and CPLEX in terms of their limitations or other stuff?
Thanks.
I think MATLAB is excellent to build prototype models and exepriment.
Things are not as efficient as if you code your application in C but in
many cases MATLAB is way good enough. [A modelling language like AMPL
is sometimes better than MATLAB.]

This assumes you call a good optimizer from MATLAB of course. I like
the
MOSEK optimization toolbox. But I am biased because I developed it. You
can try MOSEK for free vy downloading it from
http://www.mosek.com/download.html


Erling
tiger
2005-01-18 21:55:32 UTC
Permalink
HI Erling,
Thank you very much for this information. My problem is binary (0/1)
linear programming. I checked this website and I found that "The mixed
integer optimizer is a separately licensed option". I am a doctoral
student and would like to know how I can use MOSEK with the mixed
integer optimizer for free or a student discount.

Thanks!
Irv Lustig
2005-01-28 22:17:07 UTC
Permalink
You might want to check out TOMLAB at http://www.tomlab.biz/ which
provides connectors from MATLAB to CPLEX (and other solvers). You then
can use the power of MATLAB with the power of CPLEX, and get software
that is commercially supported.

-Irv Lustig, PhD
***@ilog.com
Manager, Technical Services, ILOG Direct
http://www.ilog.com/

Loading...