I attempted to run the single-process AstroBEAR application today. Having had difficulties with my previous attempt to run AstroBEAR, I switched from orda12 to orda10. The principle goal of this test was to try running the program on a different blade; unfortunately it ended up introducing a new variable to the test, as orda10 had no user processes running on it (orda12 had one other big user process running when I tried running AstroBEAR).

This test progressed smoothly up until the 14th frame, when AstroBEAR crashed with a CFL # on the order of 10⁸.

I spoke with Tim, who indicated that he was having a similar problem with one of his simulations. Following his suggestion, I reduced the suggested CFL to 0.3 and dropped the refinement level to 1. The simulation ran fine at this point.

Next I re-ran the simulation with the CFL increased to 0.4 (the refinement level stayed at 1). The simulation ran fine.