Dear All, I'm trying to maximize a likelihood with respect one parameter using "optimize" on simulated data (without error component). I've iterated the maximization procedure 1000 times and I should always obtain the same estimate of the parameter (equal to the simulated one) but, instead, i obtain different results (the likelihood function shouldn't be flat). Does anybody has experience with the optimize procedure? Thanks in advance, Chiara Chiara Seghieri Dipartimento di Scienze Statistiche "G.Parenti" V.le Morgagni, 59 50134 Firenze, Italy tel.: +39 055 4237230 fax: +39 055 4223560
If indeed your error space is pitted with local minima then I suggest you
might try a method called
bootstrap restarting to avoid these spurious local minimas.
The paper was authored by Simon Wood and can be found here
http://www.stats.gla.ac.uk/~simon/simon/papers/bsfit.pdf
-----Original Message-----
From: Chiara Seghieri [mailto:seghieri@ds.unifi.it]
Sent: 29 March 2004 10:27
To: r-help@stat.math.ethz.ch
Subject: [R] Problems with "optimize"
Dear All,
I'm trying to maximize a likelihood with respect one parameter using
"optimize" on simulated data (without error component).
I've iterated the maximization procedure 1000 times and I should always
obtain the same estimate of the parameter (equal to the simulated one) but,
instead, i obtain different results (the likelihood function shouldn't be
flat). Does anybody has experience with the optimize procedure?
Thanks in advance,
Chiara
Chiara Seghieri
Dipartimento di Scienze Statistiche "G.Parenti"
V.le Morgagni, 59
50134 Firenze, Italy
tel.: +39 055 4237230
fax: +39 055 4223560
______________________________________________
R-help@stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html
KSS Ltd
Seventh Floor St James's Buildings 79 Oxford Street Manchester M1 6SS
England
Company Registration Number 2800886
Tel: +44 (0) 161 228 0040 Fax: +44 (0) 161 236 6305
mailto:kssg@kssg.com http://www.kssg.com
The information in this Internet email is confidential and m...{{dropped}}
Chiara Seghieri wrote:> Dear All, > > I'm trying to maximize a likelihood with respect one parameter using > "optimize" on simulated data (without error component). > I've iterated the maximization procedure 1000 times and I should always > obtain the same estimate of the parameter (equal to the simulated one) > but, instead, i obtain different results (the likelihood function > shouldn't be flat). Does anybody has experience with the optimize > procedure? > > Thanks in advance, > > ChiaraMaybe you are stuck in local minima? You have to expect different results, if the data has been simulated each time you have optimized, BTW. We cannot say anything concrete since we don't now how different your results are and we do not know anything about the real underlying problem. Uwe Ligges> > > Chiara Seghieri > Dipartimento di Scienze Statistiche "G.Parenti" > V.le Morgagni, 59 > 50134 Firenze, Italy > tel.: +39 055 4237230 > fax: +39 055 4223560 > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://www.stat.math.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide! > http://www.R-project.org/posting-guide.html