Cactus Code Thorn IllinoisGRMHD Author(s) : Zachariah B. Etienne Note : IllinoisGRMHD is based on the GRMHD code of the Illinois Numerical Relativity group, written by Matt Duez, Yuk Tung Liu, and Branson Stephens (original version), and then developed primarily by Zachariah Etienne, Yuk Tung Liu, and Vasileios Paschalidis. Maintainer(s) : Zachariah B. Etienne Special thanks: The authors would like to thank Leo Werneck for proofreading the code, as well as revising comments and adding comments to it. License : GPLv2 for HARM Conservative-to-Primitive solver (src/harm_*) GPLv2+ for all other code -------------------------------------------------------------------------- ========== 1. Purpose ========== IllinoisGRMHD solves the equations of General Relativistic MagnetoHydroDynamics (GRMHD) using a high-resolution shock capturing scheme. It is a rewrite of the Illinois Numerical Relativity (ILNR) group's GRMHD code, and generates results that agree to roundoff error with that original code. Its feature set coincides with the features of the ILNR group's recent code (ca. 2009--2014), which was used in their modeling of the following systems: 1) Magnetized circumbinary disk accretion onto binary black holes 2) Magnetized black hole--neutron star mergers 3) Magnetized Bondi flow, Bondi-Hoyle-Littleton accretion 4) White dwarf--neutron star mergers IllinoisGRMHD is particularly good at modeling GRMHD flows into black holes without the need for excision. Its HARM-based conservative-to-primitive solver has also been modified to check the physicality of conservative variables prior to primitive inversion, and move them into the physical range if they become unphysical. Currently IllinoisGRMHD consists of 1) the Piecewise Parabolic Method (PPM) for reconstruction, 2) the Harten, Lax, van Leer (HLL/HLLE) approximate Riemann solver, and 3) a modified HARM Conservative-to-Primitive solver. IllinoisGRMHD evolves the vector potential A_{\mu} (on staggered grids) instead of the magnetic fields (B^i) directly, to guarantee that the magnetic fields will remain divergenceless even at AMR boundaries. On uniform resolution grids, this vector potential formulation produces results equivalent to those generated using the standard, staggered flux-CT scheme. This scheme is based on that of Del Zanna (2003, see below OPTIONAL CITATION #1). ======================================================= 2. Instructions for committing patches to IllinoisGRMHD ======================================================= Before committing a patch to IllinoisGRMHD that might *conceivably* affect IllinoisGRMHD's roundoff-level agreement with previous versions, you must first perform a correctness test. Here are the instructions: Definitions: a) EXE/ is the folder containing the executable built from your new, patched version of IllinoisGRMHD b) TEST/ is arrangements/WVUThorns/IllinoisGRMHD/test/ c) ./ is arrangements/WVUThorns/IllinoisGRMHD/ Instructions: 1) Copy TEST/magnetizedTOV.par and ./code_validation.txt to EXE/. 2) Go to the EXE/ directory 3) Edit magnetizedTOV.par, replacing the line cactus::cctk_itlast = 2 with cactus::cctk_itlast = 240 4) Perform the run on at least 8 MPI processes. If your computer has 16 cores, then use the command: OMP_NUM_THREADS=2 mpirun -n 8 -x OMP_NUM_THREADS ./cactus_[configname] magnetizedTOV.par 5) Make sure the run completes without error. If there are errors, you must backtrace what is wrong before committing the patch. 6) The central density is extremely sensitive to beyond-truncation-error changes, so we'll focus on that quantity exclusively. Parse the output file as follows: awk '{if(NF==3 && NR%3==0) print $0}' magnetizedTOV/rho_b.maximum.asc > myoutput.txt 7) Next compare the results: paste code_validation.txt myoutput.txt | awk '{printf("%.3f %.3f\t%e\t%e\n",$1,$2,sqrt($4^2),sqrt((($3-$7)/$3)^2));}' Column definitions: Column 1: cctk_iteration Column 2: cctk_time Column 3: The relative difference between the trusted version of the code and itself, but with a 15th significant digit perturbation at t=0. Column 4: The relative difference between the trusted version of the code and your updated code. If Columns 3 and 4 are within 1.5 orders of magnitude through iteration 240, and the relative difference at iteration 240 is 3e-7 or less, then your change is accepted. Please push to the git repository. =================== REQUIRED CITATIONS: =================== 1) IllinoisGRMHD code announcement paper: Class. Quantum Grav. 32 (2015) 175009, (http://arxiv.org/abs/1501.07276) 2) Noble, S. C., Gammie, C. F., McKinney, J. C., \& Del Zanna, L. \ 2006, Astrophysical Journal, 641, 626. OPTIONAL CITATIONS: 1) Del Zanna, Bucciantini & Londrillo A&A 400, 397 (2003), Eq. (44)