I've made a note in the subroutine that in some machines NJOY-2012 processes some angular distributions in JEFF-3.2 in such way that the highest value of the PDF of the scattering cosine differs from 1.0. I've been doing some extensive validations with JEFF-3.2 libraries processed with NJOY-2012, so this is probably a numerical problem in NJOY. In any case, if the data is not completely screwed up, it would be good if Serpent could use it.
Try replacing line 380 in processmudistributions.c:
Code: Select all
Die(FUNCTION_NAME, "Error in angular distribution (mt %ld)", mt);
with:
Code: Select all
Warn(FUNCTION_NAME, "Error in angular distribution (mt %ld %E %E)", mt, RDB[l1 - NP], RDB[l1 - 1] - 1.0);
If this is related to the same problem, you should get a warning instead of an error. Could post the results?