There is a problem which sometime occurs with FLUKA geometries. As far as I understand, it is due to too complex cells. Flair does not complain on geometry, but FLUKA does.
How to reproduce it (in the current flukaFix branch):
They say it's due to a hard-coded limit on the 'tx' array size, and the solution is to simplify the complex cells by splitting them to more simple cells. I currently do not see an elegant way to get around this problem without recompiling FLUKA or manually simplifying those cells in the CombLayer code just for FLUKA (which I would like to avoid).
If I simplify geometry by, say, building BilbaoWheel only, then FLUKA runs fine.
There is a problem which sometime occurs with FLUKA geometries. As far as I understand, it is due to too complex cells. Flair does not complain on geometry, but FLUKA does.
How to reproduce it (in the current
flukaFixbranch):./ess -r -fluka AAmv AA1.x AA.inp$FLUTIL/rfluka -N0 -M1 AA.inpThe run will crash and in the beginning of fluka_*/AA001.log there will be a message like this:
At line 55 of file rpnorm.f Fortran runtime error: Index '538879075' of dimension 1 of array 'tx' above upper bound of 100000The FLUKA forum says it's due to too complex cell definitions. See discussion in these threads: http://www.fluka.org/web_archive/earchive/new-fluka-discuss/5268.html http://www.fluka.org/web_archive/earchive/prova/1444.htmlThey say it's due to a hard-coded limit on the 'tx' array size, and the solution is to simplify the complex cells by splitting them to more simple cells. I currently do not see an elegant way to get around this problem without recompiling FLUKA or manually simplifying those cells in the CombLayer code just for FLUKA (which I would like to avoid).
If I simplify geometry by, say, building BilbaoWheel only, then FLUKA runs fine.