[FLASH-BUGS] refinement problem

[Daniel Seifried]

Dear FLASH team,

I found a strange behavior of FLASH4 during the initial refinement of
the grid, and I don't know whether this is a bug or not.

I set up a simple test setup to demonstrate the problem. The setup
consists of a centrally concentrated density profile with uniform temperature. The code now gives different initial grid
structures depending on whether I use e.g. only the density for the
second derivative refinement or e.g. the density plus the pressure or the temperature. This seems to me as a bug, since for a uniform
temperature, the pressure is just a scaled version of the density. Moreover, it gives different grid structures if I
exchange the first and the second refinement variable with each other (same executable).

I used the test setup (attached) with the most recent release of FLASH4,
which I downloaded today. I did not make any changes to the code except
the attached files of the setup.

I've also attached the different log files for a first look. All of the
four simulations have been performed with the identical executable +
flash.par file (except the changes in the refinement criterion) and
clearly show different grid structures. The grid structure is also not
symmetric (w.r.t. to the center) as it actually should be for
such a centrally concentrated profile.

The first checkpoint file, which is written out, looks fine except that the grid structure
show these inconsistencies, but all quantities have the values they should have.

I would be very grateful if you could help me with that problem.

Kind regards,

Daniel Seifried