Hi all,

I had to call in through Google Voice today, as I have just arrived in Barcelona for the Sant Cugat Forum on Astrophysics. My connection was not great, so I report on my progress here.

I fixed a number of compiler warnings in IllinoisGRMHD over the past week, to minimize the chances of any compiling issues. The only remaining warnings are completely harmless (e.g., variable FOO may be used uninitialized). I have verified that IllinoisGRMHD compiles without errors using GCC (4.7.3), Intel (13.1.3 20130617), and BlueWaters' default PGI compiler. In addition, I ran Valgrind tests with poisoning disabled, and zero memory issues were found in my IllinoisGRMHD test case (evolving random initial data). I also backtraced and fixed a discrepancy between the ET IllinoisGRMHD code and our old, trusted GRMHD code version.

I am working with Roland to verify that the staggered prolongation/restriction operators work as expected on AMR grids with 2+ levels of refinement on the coarse grid. He is concerned of potential issues with restriction communicating data across two or more levels. Thus far, my tests have focused only on 2-level AMR grids (i.e., one level of refinement on the coarsest grid). The new tests consist of Nth order prolongation and restriction of Nth order polynomial data on staggered grids, testing all four staggerings. I anticipate completing those tests before the next telecon.

At the moment, I am confident that there will be no issues in compiling the code on other machines, so would it be okay if we proceeded now to adding IllinoisGRMHD as an experimental thorn, as planned? That is to say, make IllinoisGRMHD part of the default ET download for this upcoming release, but commented out of the default ThornList, with a comment along the lines of: