RE: $OMEGA blocks and log-likelihood profiling
From: "Gastonguay, Marc" marc.gastonguay@snet.net
Subject: RE:[NMusers] $OMEGA blocks and log-likelihood profiling
Date: Wed, June 2, 2004 4:56 pm
No argument here, either. When the primary problem is over-parameterization,
constraining ETA estimates probably won't help much. In other cases,
however, constraining individual ETA estimates can be very helpful with
convergence in NONMEM. I have one example where constraining ETAs (within
broad, realistic bounds) reduced the bootstrap run convergence failure rate
from near 50% to less than 1%.
Marc