Issues at AMR Level Boundaries with General Equation of State

[hwhitehead]

When implementing a general equation of state, I notice anomalies forming on the interface between MeshBlocks at different refinement levels. Such irregularities are not present for the adiabatic or isothermal cases.

See this animation for an example of this anomaly in a the logarithmic density map of a shear flow. There are no custom source functions included in this system, only a linear velocity shear in the x direction for a homogenous density background. Overdensities arise unnaturally at the refinement edge and propogate through the system with the shear.

I originally thought these were a render error, but I believe now that there is some failure in hydro communication across the two levels. I have experiemented with changing the reconstruction, and noticed a warning for higher order reconstructions:

"The combination of general EOS and xorder > 2 has documented bugs in some cases."

Is it possible that these two issues are related? The anomaly persists for xorder = [1, 2, 3]. I was unable to find reference to this "documented bug" in the Issues section, so direction here would be appreciated.

Has anyone else come across this issue before?

[hwhitehead]

Bumping this thread as I am returning to this project and the issue persists. Are there any special consideraitons that must be made when using a general equation of state with AMR?