The initializer for this test was set to generate 1000 dark matter particles for the baryonic self-similar (Bertschinger 1985) test. This change simply sets that to 0, as it should be. This is required to get the correct solution as shown in the new enzo method paper. This change is really a no-brainer.
With this new change I'm still unable to reproduce the answer that nicely matches the analytic solution for this problem that's in the draft enzo method paper:
This was generated using 619d9a9. Do you see different results on the same changeset hash?
Thanks Nathan -- that was fast. In fact, the test also needs Mike Kuhlen's PR#120, which I had forgotten was still outstanding (which is my fault). Can you confirm that if you use both this PR and PR#120 that you get the correct solution?
Yes, that does produce identical results compared to what's currently in the paper:
Given that Mike's fix seems to be necessary to recover the analytic solution here for this test problem, I guess it should probably be merged in.