Adam,
Getting the variance of MLE estimator when the true parameter is on the
boundary is a very difficult problem. It is known that the standard bootstrap
does not work. There are some sub-sampling approaches (Springer book:
Politis, Romano, Wolff), but I am not an expert on this.
Dear Ravi,
Thank you so much for the help. I switched to using the optimx function
but I continue to use the spg method (for the most part) because I found
that only spg consistently converges give different datasets. I also
decided to use AIC rather that a likelihood ratio test.
I have a
Adam Zeilinger zeil0006 at umn.edu writes:
Dear R Help,
I have two nested negative log-likelihood functions that I am optimizing
with the spg function [BB package]. I would like to perform model
selection on these two objective functions using AIC (and possibly
anova() too).
Adam,
See the attached R code that solves your problem and beyond. One important
issue is that you are enforcing constraints only indirectly. You need to make
sure that P1, P2, and P3 (which are functions of original parameters and time)
are all between 0 and 1. It is not enough to impose
Dear R Help,
I have two nested negative log-likelihood functions that I am optimizing
with the spg function [BB package]. I would like to perform model
selection on these two objective functions using AIC (and possibly
anova() too). However, the spg() function returns a list and I need a
5 matches
Mail list logo