The basic strategy:

  1. Sorin 1D (confirmed his results), bearclaw
  2. Sorin 1D module extended to 2D successfully, bearclaw
  3. Sorin/Kris 2D module, bearclaw
  4. Sorin/Kris 2D module with perturbation, bearclaw
  5. Kris 2D module, astrobear
  6. Kris 2D module with perturbation, astrobear

The result of step #3: comparison versus the results from steps 1 and 2, at t~0.3ns (tfinal is 1 ns).

Blue: Sorin 1D
Green: Sorin 1D -> 2D extension
Red: Sorin/Kris 2d module

The only difference between green and red is the data which are read in. The red interpolates onto a my=800 grid data taken from Riccardo's original my=2005 data files, whereas the green uses a my=800 data file generated by Sorin. At frame 0, the pressure below the interface is uniformly less in the data which are interpolated than those which were generated.

I'm assuming this is the source of the discrepancy. Otherwise, they all appear in good agreement (certainly compared to previous iterations), so Lan can try using Sorin's code & see what happens.

The 1536 strongly-cooling shocked clump is still going: (click for big)

Here's a movie (assuming it gets done in time)


Here's a comparison at frame 71 (of 150): frame71.gif


Comparison of vorticities: