Problems with MRW=True

[PeterSchilke]

Hi Kees,

we are trying to run a bunch of RADMC-3d models as input for training a neural network. These are radial models of dense cores with high luminosity sources. We are running the branch fixoutofcellbetter, since we had problems with out of cell error messages.
For one of our highest density models, RADMC-3d with MRW=False run since 12 days now. With MRW=True it runs 1 s and then crashes with the following error message:

...
  Reading dust densities...
 Dust mass total =    224.75473889676726       Msun
 In Monte Carlo initialization: found that scattering_mode==0, so putting kappa_scat=0.
 Computing emissivity database...
 Computing total input luminosities...
 Modified Random Walk mode is switched ON
  
 Starting the thermal Monte Carlo simulation....
 Thread Nr           0 of          16 threads in total
 Thread Nr           1 of          16 threads in total
 Thread Nr          12 of          16 threads in total
 Thread Nr           6 of          16 threads in total
 Thread Nr           7 of          16 threads in total
 Thread Nr           3 of          16 threads in total
 Thread Nr          14 of          16 threads in total
 Thread Nr          10 of          16 threads in total
 Thread Nr           8 of          16 threads in total
 Thread Nr           5 of          16 threads in total
 Thread Nr           4 of          16 threads in total
 Thread Nr          13 of          16 threads in total
 Thread Nr          15 of          16 threads in total
 Thread Nr           9 of          16 threads in total
 Thread Nr          11 of          16 threads in total
 Thread Nr           2 of          16 threads in total

stderr: Note: The following floating-point exceptions are signalling: IEEE_OVERFLOW_FLAG
STOP 5931

Any ideas? I am of course happy to provide all the input files if needed.

 Thanks for any help,

      Peter