whaeck
commented
3 years ago Using test 35 as a template and setting nunx=2 in that input file, PURR crashes on the two do loops (do ie=1,nunr) on lines 435 and 485. This is due to the fact that since nunx=2, only 2 tables have been calculated (for the first and the last unresolved energy point) while these output loops still iterate over every unresolved resonance energy and assume that every energy point has a probability table.
One solution is to change the two do loops in do while loops like the one on line 286 and increment the ie index by the previously calculated nstep. We could have used nunx instead of nunr in these loops, but we actually must iterate over the unresolved resonance energy grid since the heating cross sections that will be included in the probability table are calculated on the complete unresolved resonance energy grid, before we even calculate the probability tables.
Bob MacFarlane sent us a few updates to the PURR module. One of these updates relates to the nunx value in the input file.
nunx allows a user to ask for all unresolved cross sections at each unresolved energy (nunx=0) or just nunx values. It would appear that this option is actually ignored in NJOY2016 (going back through the source code of NJOY2012 and NJOY99, it would appear that this has always been the case) since nunx is never used in any of the energy point loops.
The value of nunx is used to calculate an integer step value so that the points being calculated are spread out over the unresolved resonance range. E.g. if nunx=2, then nstep=nunr/(nunx-1)=nunr, which means only the lower and upper unresolved energy point will be calculated.