#1796: Include IllinoisGRMHD into the Toolkit, as an Arrangement ------------------------------------+--------------------------------------- Reporter: zachetie@… | Owner: Zachariah Etienne Type: enhancement | Status: new Priority: major | Milestone: ET_2015_11 Component: EinsteinToolkit thorn | Version: development version Resolution: | Keywords: GRMHD IllinoisGRMHD ------------------------------------+---------------------------------------
Comment (by Zach Etienne):
A test case will be added as required, despite the fact that time limitations mean it will be mostly useless to me (see below; I'd love to be proven wrong). I figure that at least the users might benefit from an example par file.
After any update to IllinoisGRMHD that might conceivably introduce beyond- roundoff-level differences (due to a bug), I manually perform full *correctness* testing on a physically-motivated scenario prior to pushing my changes. This involves a careful analysis of the growth of roundoff error versus a trusted version. I also compare the trusted version with itself, but with a 15th significant digit initial perturbation. If the growth in roundoff in the new version of IllinoisGRMHD does not match the growth in roundoff in the trusted version, it is concluded that IllinoisGRMHD has a bug.
As you might imagine, this testing requires many-timestep runs with AMR grid movement, and I don't care if the test takes a couple hours; my science requires that the tool be reliable and consistent with trusted versions beyond a doubt. Perhaps we should have a longer discussion about the possibility of including more robust correctness testing in ET.