Re. fmincon vs lsqnonlin
in [Matlab]
|
From: Jason Park on 20 Mar 2010 00:19 Having read the thread, I got a couple of more questions in addition to that related to my case that I am working on. 1) I am using LSQNONLIN at the moment, and my code reached convergence despite the fact that some of the parameter estimates actually are out of upper and lower bounds pre-specified. Why is this happening? 2) what is the benefit of using LSQNONLIN as compared to FMINCON? is it more efficient? and can FMINCON deliver somewhat more accurate estimates than LSQNONLIN? what is the best for fitting a complicated model/function to the data with a couple of loops embedded within? 3) what other suggestions do you have for my case if you'd like to comment for faster convergence? It is taking me more than hours or sometimes days to even get errors at the moment... I am pretty much a beginner for MATLAB, and any constructive criticism and suggestions are more than welcome. If what I have in my code would help you understand this better, I can probably post it following a request if there comes any. Thanks heaps for reading this!! Jason
|
Pages: 1 Prev: Re. fmincon vs lsqnonlin Next: Can i run my simulink model from my GUI |