Here is the Comparison against Ramses:
DM Only Comparison:
Initially this was the power spectrum comparison:
Surprisingly RAMSES has ~30% more power then Cholla on small scales. For reference, Cholla has ~25% more power than Enzo on similar scales.
On the Ramses paper they say: Using the potential, the acceleration is computed on the mesh using the 5-points finite difference approximation of the gradient.
I changed the way the gravitational field is computed from the potential, originally I was using 3 point finite differences for the derivatives for the gradient :
I changed for the 5 point finite differences:
Now the results are much better:
Differences are <~1%.
I still have the temperature issue in Ramses: