CE

Computer time

  • We need to perform 2 simulations
  • We may only be able to complete ~1 simulation using Anvil
  • DOE allocation?

Reduction of ambient pressure and density

Method 1 from last post: Starting from t=0 with low ambient pressure and density

  • Continued Run 026 (ideal gas) and Run 027 (MESA EOS), from last blog post
  • Run 026 took about ~1.75 hours per frame up to 19 frames
  • Run 027 took about ~2.5 hours per frame up to 34 frames
    • crashed shortly after frame 34
      • might have been caused by extra refinement (up to level 3) outside of desired level 4 refinement sphere
        • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run027_fr34_grid.png
        • I tried defining a shape like the level 4 refinement sphere but with "DEREFINE_OUTSIDE"
        • I also tried setting Refinement%Tolerance=-1d0
        • Both of these changes had no effect
        • Finally, in physics.data, changed refineVariableFactor to 0d0,0d0,0d0,0d0,0d0,0d0,0d0,0d0,0d0,0d0,0d0 from 1d0,1d0,1d0,1d0,1d0,1d0,1d0,1d0,1d0,1d0,1d0, to force derefinement outside of refinement object 1, and evolved from frame 34 to 35.
          • Prevents from crashing and speeds up simulation from 2.5 hours to 1.5 hours per frame
            • Is this method fine? If so, should I
              • continue from frame 35?
              • restart from frame 19 (when unwanted refinement was either absent or marginally present)?
              • restart from beginning of simulation?
  • Checked separation curve up to 19 days and it seems okay:
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/p_mult_282_277_026_027.png
  • Visualization up to 19 days looks okay but artifacts present, especially in Run 027 — harmless?
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run026_fr19_rho.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run282_fr19_rho.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run026_fr19_P.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run282_fr19_P.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run027_fr19_rho.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run277_fr19_rho.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run027_fr19_P.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run277_fr19_P.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run027_fr19_T.png
    • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run277_fr19_T.png

Method 3 from last post: Restarting from frame 47 of old runs and modifying ambient

  • Found that setting iSelfGravity=2 (non-momentum-conserving self-gravity) was the key to getting rid of the artifacts
    • Also speeds up simulation such that it is about half as fast as the original simulation at frame 47 to 47.1
    • However, when I try changing back to iSelfGravity=1 (momentum-conserving self-gravity) artifacts at boundary of level 1 grid appear (though artifacts at boundary of level 0 grid that I was seeing before — when iSelfGravity was left unchanged from 1 during the ambient modification — do not appear)
      • Figures show density, pressure and temperature, comparing frame 47 (left panel; original Run282, iSelfGravity=1), 47.1 (middle panel; low ambient, iSelfGravity=2) and 47.2 (right panel; iSelfGravity=1):
        • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run282_028_030_rho.png
        • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run282_028_030_P.png
        • https://www.pas.rochester.edu/~lchamandy/ambient_tests/run282_028_030_T.png
      • Is there a way around this?
      • If not, then how bad would it be to use iSelfGravity=2 for the remainder of the simulation?

Discussion

  • Method 1 working well now? Artifacts present in ambient but seem harmless
  • Decision about refineVariableFactor

Comments

No comments.