From:"Stephen Duffull"   
Subject:[NMusers] BOV and BSV  
Date:Mon, March 11, 2002 7:35 pm  


Just a quick question about estimating BSV and BOV.

I have some data where I can get estimates of BOV but the estimates of BSV
for some of the parameters becomes essentially zero (< 1e-10) - ie all
parameter variability is thought to arise from BOV.  If I delete the ETAs
from the code of those parameters for which NONMEM cannot estimate BSV then
the code does not minimise successfully - indeed seems to 'go off into
space'; however, if I fix the OMEGA elements to zero then it works fine.  Is
it necessary for NONMEM to have BSV components for all parameters that it
wishes to estimate BOV for?  I cannot think of a reason why this would be


PS  I am not necessarily assuming that BSV is zero for those parameters for
which BSV cannot be estimated.
Stephen Duffull
School of Pharmacy
University of Queensland
Brisbane 4072
Tel +61 7 3365 8808
Fax +61 7 3365 1688