RE: $OMEGA blocks and log-likelihood profiling

From: Leonid Gibiansky lgibiansky@emmes.com Subject: RE:[NMusers] $OMEGA blocks and log-likelihood profiling Date: Tue, June 8, 2004 5:45 pm I think we need to discuss what is "failed" run. The run may fail for variety of reasons with different error messages. The most "safe" reason is insufficient accuracy: failure to achieve required precision. I am sure everyone observed this phenomenon when the run stops at the precision just below the requested one. If you request 3 digits you get 2.9, if you request 5 you get 4.5. I believe, these runs should be treated as "converged", not failed. Indeed, what is the difference if the run attained numerical precision of 3 significant digits with SIGDIGITS=3 (what is called good convergence) or with SIGDIGITS=5 (which would be treated as failure)? I would guess that failed runs in Nick example are of this variety. On the other hand, if the run fails with undefined OF or undefined precision, I would be hesitant to accept it even in the bootstrap setting. Leonid