Is there a somehow a possibility of using "optimize" (or anything else) to do a constrained minimization, i.e. where the variables of the function to be minimized are also subject to a constraint?
At present the only way to do this would be to implement your own penalty method using Optimize. The penalty method adds something to the merit function that increases the return value when it is in violation of the constraint. It needs to be smooth so that optimize can follow the gradient into fair territory. The value added as a penalty needs to be large relative to the non-penalized values to make sure that the minimum is found in the correct region. Sometimes it is necessary to iterate, gradually increasing the penalty to refine the solution.
Naturally, what I just wrote applies to situations where "good" is at low values of the merit function...
my first thought was that I could use FindRoots to force the constraints onto the values but I guess Optimize would not be able to find a solution.
Just curious: are there attempts to have a solver for such problems in IGOR 7?
Naturally, what I just wrote applies to situations where "good" is at low values of the merit function...
John Weeks
WaveMetrics, Inc.
support@wavemetrics.com
June 8, 2012 at 11:46 am - Permalink
Just curious: are there attempts to have a solver for such problems in IGOR 7?
June 11, 2012 at 08:13 am - Permalink
June 11, 2012 at 02:08 pm - Permalink
I'm not sure what you have in mind. It's not clear to me how you would use FindRoots in this situation.
We are still a long way from having a working Igor 7.
John Weeks
WaveMetrics, Inc.
support@wavemetrics.com
June 11, 2012 at 02:34 pm - Permalink