In my package 'glmmML' I'm using old C code and linpack in the
optimizing procedure. Specifically, one part of the code looks like this:
F77_CALL(dpoco)(*hessian, &bdim, &bdim, &rcond, work, info);
if (*info == 0){
F77_CALL(dpodi)(*hessian, &bdim, &bdim, det, &job);
........
This usually works OK, but with an ill-conditioned data set (from a user
of glmmML) it happened that the hessian was all nan. However, dpoco
returned *info = 0 (no error!) and then the call to dpodi hanged R!
I googled for C and nan and found a work-around: Change 'if ...' to
if (*info == 0 & (hessian[0][0] == hessian[0][0])){
which works as a test of hessian[0][0] (not) being NaN.
I'm using the .C interface for calling C code.
Any thoughts on how to best handle the situation? Is this a bug in
dpoco? Is there a simple way to test for any NaNs in a vector?
Göran Broström
PS. Yes, I will switch to .Call! Soon...
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
