Solver crashes during initialization when particle # above certain threshold

[stephendkang]

The solver crashes when too many particles are put into the system. To replicate this problem, one can construct the simplest possible simulation (e.g. Nvol_c = 1, Nvol_s = 0, Nvol_a = 0, type = homog) and increase the particle number to "Npart_c = 3000." The program crashes while displaying "0% ..ETA."

• Changing DAESOLVER settings (MaxNumSteps, IterIC, StepsIC, MemoryIncrements, etc) does not seem to affect this threshold.

• Physical memory resources of the machine seems unrelated to the problem. Memory usage can be made minimal but the initialization will still fail. (about 600 MB for 2500 particles)

• The threshold for crashing is platform dependent. On windows (daetools 1.9.0), the threshold is about 2453 (plus minus a few randomness each run) particles. On a linux machine the threshold seems to be above 4100, but eventually the same problem happens. This is possibly related to the C compiler used for these platforms rather than the platform itself.

• Will probably have to debug daetools to find the exact reason.

[d-cogswell]

Has anyone tried to compile daetools with debug flags enabled? It would be interesting to know what is causing the segmentation fault.

[d-cogswell]

Here are config files to reproduce this issue:
SimExample.zip