RE: Very big NONMEM
Bernard
>> I am actually performing simulations of a somewhat stiff system, with 6 or
>> more DVs over long time periods (weeks). Currently I can fit things in 500
>> records if I am careful with the selection of timepoints but I was wondering
>> if larger versions than "big" were possible. I have seen some discussion of
>> this subject in previous postings but these were with nmvi or earlier. I do
>> realise that NONMEM may not look like the best tool for this but I'd like to
>> try fitting real world data later.
I realise this question is about understanding the application of NONMEM to
these circumstances. However, why don't you simulate your model using
different software (S, R or MATLAB) - where these limitations don't exist? The
problem seems to be using software that is optimized for estimation to do
simulation ...
You can easily call NONMEM (for estimation) from S, R and MATLAB. Hence
simulating the complex data sets and then calling NONMEM for the estimation
process (once the data sets have been reduced to an appropriate size) is
relatively simple.
Regards
Steve
--
Professor Stephen Duffull
Chair of Clinical Pharmacy
School of Pharmacy
University of Otago
PO Box 56 Dunedin
New Zealand
E: [email protected]<mailto:[email protected]>
P: +64 3 479 5044
F: +64 3 479 7034
W: http://pharmacy.otago.ac.nz/profiles/stephenduffull
Design software: http://www.winpopt.com
Quoted reply history
From: [email protected] [mailto:[email protected]] On
Behalf Of Bernard Murray
Sent: Saturday, 9 October 2010 8:13 a.m.
To: [email protected]
Subject: RE: [NMusers] Very big NONMEM
Hello there,
Now that the subject of (unusually) large versions of NONMEM has been raised
again, I was wondering if it is possible to increase the number of observation
records per subject (NO) past the value of 500 provided in the standard "big"
version of SIZES with nm7.
I am working with NONMEM 7.1.2 with g95 (version 0.93!) on win32 (WinXP SP3)
If I increase NO and rebuild everything (as per User Guide III 2.9.2) Windows
complains ("Not a valid win32 application"). I have also tried increasing
LIM1, LADD and some of the other LIM buffers. I don't know if this is a
fundamental limitation or poor understanding on my part.
I am actually performing simulations of a somewhat stiff system, with 6 or more
DVs over long time periods (weeks). Currently I can fit things in 500 records
if I am careful with the selection of timepoints but I was wondering if larger
versions than "big" were possible. I have seen some discussion of this subject
in previous postings but these were with nmvi or earlier. I do realise that
NONMEM may not look like the best tool for this but I'd like to try fitting
real world data later.
Thanks for any suggestions.
All the very best,
Bernard
Bernard Murray, Ph.D.
Senior Research Scientist, Drug Metabolism
Gilead Sciences, Foster City CA