Dear Colin, Not sure if this will help you but one possibility is to incorporate the constraints into the function itself. For example if your variable x must be larger than some constant C, i.e. x > C, then you can replace every instance of x in your function with log(exp(C) + exp(x)). The latter is automatically bounded below by C and grows linearly for x large. You will have to recompute your gradients if you're using them but that shouldn't be a problem. You're then free to change C as you please.
All the best, James On 7 June 2012 17:00, <[email protected]> wrote: > Send NLopt-discuss mailing list submissions to > [email protected] > > To subscribe or unsubscribe via the World Wide Web, visit > http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss > or, via email, send a message with subject or body 'help' to > [email protected] > > You can reach the person managing the list at > [email protected] > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of NLopt-discuss digest..." > > > Today's Topics: > > 1. Moving Upper Bounds and Lower Bounds Dynamiclly > (McGonagill, Colin) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Wed, 6 Jun 2012 16:57:20 +0000 > From: "McGonagill, Colin" <[email protected]> > To: "[email protected]" > <[email protected]> > Subject: [NLopt-discuss] Moving Upper Bounds and Lower Bounds > Dynamiclly > Message-ID: > <[email protected]> > Content-Type: text/plain; charset="us-ascii" > > Is it possible (or wise) to change the Upperbounds and Lowerbound > constraints from within the function dynamically? I only need one variable > to move about 10%. I have a problem that has hard limits (upperbounds and > lowerbounds) that change dynamically as the system optimizes. I am finding > that if my function leaves an applicable space it will crash. Normally I > would use inequality constraints to constrain the dynamic space but as I > said I cannot have a guess that goes outside the boundary conditions. I > also do not want to use a penalty function or something of that nature > because the convergence is sometimes wrong or it will lock itself into a > hole. I normally only use COBYLA. > > Thanks, > > Colin McGonagill > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://ab-initio.mit.edu/pipermail/nlopt-discuss/attachments/20120606/35267b9c/attachment-0001.html > > > > ------------------------------ > > _______________________________________________ > NLopt-discuss mailing list > [email protected] > http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss > > > End of NLopt-discuss Digest, Vol 35, Issue 1 > ******************************************** > >
_______________________________________________ NLopt-discuss mailing list [email protected] http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
