RE: Simulation problem
From:"GIRARD PASCAL" PASCAL.GIRARD@adm.univ-lyon1.fr
Subject: RE: [NMusers] Simulation problem
Date: Mon, November 22, 2004 3:47 pm
Manoj,
In addition to what Nick writes, NONMEM will use the simulation
parameters you give in $THETA, $OMEGA and $SIGMA and reuse it as
initial estimates for the fit. So you will probably need to give
some boundaries for the fitting step even if they are ignored for
the simulation step. If you want to have different initial estimates
than the ones used for simulation you will need to have extra initial
THETAs. Then, in your $PRED or $PK you will need extra code to test
whether NONMEM is running the simulation or the fitting step (I think
you do it by testing ICALL, but am not 100% sure) then initialize using
the correct THETA. Same can be done for OMEGAs and SIGMAs with extra
ETAs and EPSILONs.
Pascal
_______________________________________________________