Error in isentropic Euler vortex initialization

[andrewwinters5000]

The initial condition for the isentropic vortex of the compressible Euler equations has a bug. The computation of the perturbed rho quantity uses "inverse division" [what the heck Julia?!?]

  rho = rho * (1+dtemp)^(1 \ (equations.gamma-1))

instead of normal division

  rho = rho * (1+dtemp)^(1 / (equations.gamma-1))

This error is present in all euler_vortex elixirs in the examples/tree_2d_dgsem and examples/special_elixirs.

Also, the amplitude, background pressure, etc. make for a very weakly perturbed vortex as compared to the initialization found here.

[ranocha]

Backslash comes from Matlab and friends for solving linear systems... Want to propose another setup and make a PR?

[ranocha]

But it's definitely great that you found this!

[sloede]

But it's definitely great that you found this!

I second that! Maybe this explains why it was always so fickle to use as a test setup. Can you spot the difference qualitatively when comparing contour plots or is it more subtle?

[andrewwinters5000]

But it's definitely great that you found this!

I second that! Maybe this explains why it was always so fickle to use as a test setup. Can you spot the difference qualitatively when comparing contour plots or is it more subtle?

The current initialization or switching to use division look qualitatively the same (i.e. like a vortex). What should be adjusted as well as is the size of the iniamplitude and/or the pressure p. These values influence how large the du values are as well as the Mach number. @gregorgassner and I cooked up a decent alternative version of this initial condition. We can (maybe) use it for all these elixirs as well.

[andrewwinters5000]

Backslash comes from Matlab and friends for solving linear systems... Want to propose another setup and make a PR?

Yeah, once I carve out some time the initial condition and the testing values can get updated accordingly.