home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.genetic
- Path: sparky!uunet!scifi!acheron!philabs!linus!agate!stanford.edu!nntp.Stanford.EDU!peetah
- From: peetah@leland.Stanford.EDU (Peter Gage)
- Subject: Constraint-handling in GA function optimization
- Message-ID: <1993Jan27.235948.6423@leland.Stanford.EDU>
- Followup-To: peetah@leland.stanford.edu
- Sender: peetah@leland.stanford.edu
- Organization: DSG, Stanford University, CA 94305, USA
- Date: Wed, 27 Jan 93 23:59:48 GMT
- Lines: 24
-
- Recently there was a request for information on appropriate
- penalty functions to use for constrained function optimization
- with genetic algorithms. (Unfortunately I discarded the original
- message.) I would like to extend the scope of this inquiry to
- include all methods for handling constraints. It seems that you
- lose information by shoving constraint violations and objective
- into a single scalar merit function, and I am interested in
- methods which treat constraints separately.
-
- I have noticed the mention of constraint-handling 'by repair',
- but it's not clear to me exactly what this means. Is this
- method used to make a viable genome (e.g. check that every
- city is represented in travelling salesman problem) so that
- fitness can be evaluated, or is it applied after fitness has
- been evaluated so that future fitness can be improved?
-
- Are there other methods for handling constraints 'out there'?
-
- Please e-mail to me, and I will summarise for the group.
-
- Peter Gage
-
-
-
-