KTH /
Engineering Science
/
Mathematics
/
Optimization and Systems Theory
SF2822 Applied Nonlinear Optimization, 7.5hp
This course will be given in period 4, spring 2011. (Please note that
the course has been moved from period 2 to period 4.)
Important! If you plan to take this course and do not have the
textbook by Griva, Nash and Sofer, you should order it well in advance. Read
more here.
Examiner and instructor:
Anders Forsgren
Exercise leader and project leader:
Anders Möller
The GAMS exercise sesstion on March 30 will be held in computer rooms
Butter, Spelhallen and
Sporthallen. See map
and information.
- Course information
- Theory questions for the final exam (pdf)
- GAMS documentation
- GAMS introductory slides
- Introductory exercises to GAMS
- GAMS files to introductory exercise
- M-files that may be used for exercise 2B:
-
qp.m. Suggested "skeleton" to QP-solver. (Two
versions, with and without slack variables. Make the choice that suits
your chosen method.)
-
sqp.m. Suggested "skeleton" to SQP-solver.
-
ex512.m. Data for problem 5.12 from the exercise booklet, in a format that suits the above SQP skeleton.
-
hexagon.m. Data for the problem of
maximizing the area of a hexagon when no two corners are more than one
unit of lenght apart. See the m-file for the problem formulation.
(This is probably a rather hard problem. You are doing well if you are
able to solve this problem.)
-
sample.m. M-file skeleton illustrating the
syntax for solving the two problems above.
- Matlab-files that may be used for drawing the feasible region of
a two-dimensional problem with linear constraints: polyplot.m and ineqplot.m.
-
GAMS at the F students' Ubuntu computers.
-
Type "module add gams" or add it to a suitable login file.
-
Use an editor, for example emacs, to create/modify model files (".gms") and reading output files (".lst").
-
Put the model files in your home catalog. Run GAMS from that catalog, e.g. "gams trans1".
-
Please note that there is a whole library of example files at GAMS subdirectory "modlib".
-
GAMS in MMT's computer rooms.
-
GAMS 2.50, which is Windows based, can be run in Bergs computer rooms
B21, m110 och m122. It is under "Programs" and it is called
GAMSIDE. However, GAMS must be run from DOS. Create a DOS window by
clicking up and right in GAMSIDE. Go to your home catalog and run the
GAMS file from there. The list file may then be opened in
GAMSIDE. (The problem is due to GAMS requiring write access on a
system disk, which is not allowed in this network.)
-
When running from the DOS prompt, you are recommended to add the
following line in your GAMS file:
options nlp=snopt;
or
options nlp=knitro;
(Then you choose the solvers SNOPT or KNITRO, which are suitable.)
-
GAMS on your own computer.
-
The demo version of GAMS (which we use) can be downloaded from the GAMS website.
-
To solve larger problems than what the demo version of GAMS can handle.
-
There is an option to solve optimization problems over the internet by NEOS
(http://www.neos-server.org/). Here you can send for example
GAMS files and obtain access to various solvers without the size limitations
of the demo version. This is a very useful tool. We recommend
using the solvers
SNOPT
or KNITRO
for solving NLP models.
-
NEOS is usually reliable, but please note that you will not be allowed
to hand in late because of possibly waiting for answer from
NEOS. Please do not wait to the last moment with submitting jobs.
- Exams
- Final exam December 15 2007 (pdf)
Solutions to final exam December 15 2007 (pdf)
- Final exam June 5 2008 (pdf)
Solutions to final exam June 5 2008 (pdf)
- Final exam December 20 2008 (pdf)
Solutions to final exam December 20 2008 (pdf)
- Final exam June 10 2009 (pdf)
Solutions to final exam June 10 2009 (pdf)
- Final exam December 17 2009 (pdf)
Solutions to final exam December 17 2009 (pdf)
- Final exam June 9 2010 (pdf)
Solutions to final exam June 9 2010 (pdf)
- Final exam May 28 2011 (pdf)
Solutions to final exam May 28 2011 (pdf)
- Final exam August 23 2011 (pdf)
Solutions to final exam August 23 2011 (pdf)
|
