Mahesh, NONMEM's SIDIG option is a convergence criterion. See http://en.wikipedia.org/wiki/Convergence_tests for some discussion of what convergence means.
Stuart Beal decided to use a criterion that all parameters should have >= SIGDIG significant digits. Another criterion could have been an OBJ value that changed by less than FOBJ where FOBJ might be 0.001. I would have preferred the latter criterion because it seems like NONMEM wastes a lot of time crunching through iterations with no reported change in OBJ. I dont believe the parameter estimates are really any better after this activity. Other similar programs (e.g. WinBugs, Monolix) don't have a convergence criterion but ask the user to decide by looking to see if they have found a fuzzy caterpillar rather then a wiggly worm (strange but true!). Picking a SIGDIG of 3 is just traditional - like using a Type 1 (alpha) error of 0.05 for hypothesis testing. For convenience, most people use SIGDIG of 3 so they don't have to spend time explaining why they might prefer 2 or 4. My own preference is 6 because I happened to come across a simple 2 cpt PK problem with NONMEM V that got the right answer (which was known a priori) with SIGDIG=6 but converged prematurely with SIGDIG=3. A SIGDIG of 5 was almost as good as 6 but 4 was wrong but not as bad as 3. NONMEM VI does better than NONMEM V but still terminates prematurely with SIGDIG=3. My conclusion from this one example is that SIGDIG=3 is not a reliable convergence criterion even with NONMEM VI. Details can be found at http://www.health.auckland.ac.nz/pharmacology/staff/nholford/misc/sigdig/ Comparing SIGDIG with significant digits in a parameter estimate (as you did for volume) isn't really the right thing to do. The imprecision of the estimate (e.g. from bootstrap confidence intervals) is much greater than implied by SIGDIG for almost any estimation problem. If we care about a parameter value then it is its bias and imprecision that are important -- not the value of SIGDIG for that parameter. Nick Mahesh wrote: > There have been some elegant references posted to the usersnet in response to > this question. However, one question has > generally gone unanswered. Navin tells us that NONMEM says MINIMIZATION > TERMINATED DUE TO ROUNDING ERRORS > (ERROR=134); then he change the SIGDIGITS to a lower value (this is generally > NSIG=2) and MINIMIZATION > SUCCESSUFUL shows up along with the STANDARD ERROR OF ESTIMATE. This is > actually one of the recommended tips in > Dr. Bonates book for Error 134. Dr. Bonate further explains If the rounding > error is a variance component then this is > usually an acceptable solution. > > <?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" /> > > Kindly note that almost always the parameter estimates are identical between > the run that had minimization terminated and the > job that ran successfully with NSIG=2. Could someone kindly teach us what > makes NSIG=3 so important. As an example, is > a volume estimate of 96.3 L that much better than 93 L. -- Nick Holford, Dept Pharmacology & Clinical Pharmacology University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealand [EMAIL PROTECTED] tel:+64(9)373-7599x86730 fax:+64(9)373-7090 www.health.auckland.ac.nz/pharmacology/staff/nholford
