As of December 6, 2003, this page is still under development. Further details will be added soon.
This outline is a tentative guide, and will be updated as the course progresses. Time will be allocated as appropriate for student presentations and discussion.
Refer to the course description for the topic and overall goals of this course. Generally, the emphasis this semester is on implementation and application of the techniques we studied last semester.
Note: Some of the material referenced below may be copyrighted, and hence is not generally available by clicking on it. Please email me, email@example.com, and I will email you either a Postscript or a PDF copy of the appropriate excerpts, according to the ``fair use" provision of the copyright law.
page for the course
Bibliography for the course in PDF format
|Topic no.||Description||Explanation / References / Projects|
||See the book Rigorous Global Search, Continuous Problems (Kearfott, Kluwer, 1996), and also see the GlobSol components overload/codlvars.f90 and overload/overload.f90 (module CODELIST_CREATION). To generate code lists, use the scripts and programs fmake, makegrad, and optimize_codelist. To exhibit code lists in a readable form, use prntcodl. (See me about using GlobSol on the Sparc's and about possibly installing it, and the necessary compiler, on personal machines.)|
||See LP_interface/(run_find_LP_solution find_lp_solution Neumaier_validate_LP).f90
in GlobSol. Also check out the routine DSPLP in the Slatec library.
(Look at /mathpkgs/SLATEC/dsplp.f on our Ultra system.)
||See create_LP_problem/(LP_operations test_LP_operations).f90 in
GlobSol. The basic structure and most basic linear operations are
already implemented in module LP_OPERATIONS. Projects include:
||We should review what we have done and identify what we need to do.|
||(tentative) A personal visit from George Corliss (week of March 8)|
|| The idea here is to go beyond the testing for correct programming
(which presumably we have done), and to test the value of the implementation
in solving actual problems. Possible sources of test problems include
|7.||Study of an alternate general environment: The COCONUT project||The COCONUT project was a collaboration to develop and open-source,
extensible environment for global optimization. We can install this
C++ environment, and we can contribute to it. References may include: