Runtime parameters

Runtime parameters are read from param.dat each time the executable is launched — you can change them between runs without recompiling. Each problem ships with a sample param.dat under prob/PROB/param.dat; copy it into your output directory before running.

The parameter file format is straightforward: each line begins with a type tag ([dbl], [int], or [str]) that tells the parser which data type to read, followed by the parameter name and its value. Lines beginning with # are ignored.

The torus problem (prob/torus/param.dat) is used as the running example throughout this page.

Time control

[dbl] tf     = 1000.0
[dbl] dt     = 1.0e-06
[dbl] cour   = 0.9

tf is the simulation end time in code units (gravitational units $GM/c^3$ for black hole problems). dt sets the initial timestep; the code will adapt it quickly, so this value mainly prevents an enormous first step — 1e-6 is a safe conservative starting point. cour is the Courant–Friedrichs–Lewy (CFL) number: the timestep is set to cour times the minimum light-crossing time across all zones. Values between 0.5 and 0.9 are typical; smaller values are more stable but take more steps to reach tf.

Output cadences

[dbl] DTd  = 5.0
[dbl] DTf  = 10.0
[dbl] DTl  = 0.5
[int] DTr  = 10000
[int] DTp  = 100

These control how often different types of output are written. DTd, DTf, and DTl are measured in code time units; DTr and DTp are measured in timesteps.

DTd — interval between fluid-state dumps written to dumps/.
DTf — legacy "full" dump interval, retained for historical compatibility and unused in current runs.
DTl — interval between entries in log.out (for example, radial fluxes and maximum divergence error).
DTr — interval between restart files, written every N timesteps.
DTp — interval between performance reports printed to stdout, written every N timesteps.

For black hole accretion problems in EHT production runs, a typical value is DTd = 5.0.

Problem-specific

In addition to the core parameters above, each problem registers its own parameters in set_problem_params() inside prob/PROB/problem.c. These are read from the same param.dat — there is no separate file — so what appears in param.dat depends entirely on the problem. For example, mhdmodes2d adds only a single parameter nmode (an integer selecting which linear MHD eigenmode to initialise), whereas the torus problem adds rin, rmax, u_jitter, beta, mad_type, explained below.

[dbl] rin      = 6.0
[dbl] rmax     = 12.0
[dbl] u_jitter = 4.0e-02
[dbl] beta     = 100.0
[int] mad_type = 0

rin and rmax define the Fishbone-Moncrief torus: rin is the inner edge of the torus (in $r_g\equiv GM/c^2$) and rmax is the radius of the pressure maximum (also in $r_g). It is a solution to the relativistic Euler equations in a stationary, axisymmetric background for an isentropic, axisymmetric, purely azimuthal flow. The solutions are parameterized by a constant angular momentum mass density, $\ell\equiv u_{\phi}u^{t}$. The angular momentum of the torus is set analytically from rmax. Typically we set rin and rmax to 10 and 20 for SANE simulations, and 20 and 41 for MAD simulations.

u_jitter adds a small random perturbation to the initial internal energy field. This perturbs the equilibrium torus, seeds instabilities such as the magnetorotational instability (MRI), and helps trigger accretion.

beta is the target plasma $\beta = P_{\rm gas} / P_{\rm mag}$ used to normalise the initial magnetic field. beta = 100 is the standard choice and it implies that the initial field is dynamically weak, which is standard for most GRMHD torus setups.

mad_type specifies the initial magnetic field configuration. 0 selects the standard and normal evolution (SANE), while 1 selects a magnetically arrested disk (MAD) setup. Although the steady-state magnetic flux is not trivially determined by the initial conditions, these configurations reflect setups identified in previous work as producing either SANE or MAD flows.

Fluid parameters

[dbl] gam = 1.333333

gam is the adiabatic index $\gamma$ of the gas. For a relativistic gas, $\gamma = 4/3 \approx 1.333$; for a non-relativistic monatomic gas, $\gamma = 5/3 \approx 1.666667$. The latter is a good choice for single-temperature GRMHD.

Coordinates and domain

[dbl] a           = 0.9375
[dbl] Rout        = 50.0
[dbl] hslope = 0.3
[dbl] mks_smooth  = 0.5
[dbl] poly_xt     = 0.82
[dbl] poly_alpha  = 14.0

a is the dimensionless black hole spin parameter, $a = J c / G M^2$, ranging from 0 (Schwarzschild) to 1 (maximally spinning Kerr).

Rout sets the outer radial boundary of the grid in gravitational radii ($r_g = GM/c^2$). The inner boundary is calculated based on the outer boundary and the resolution in the radial direction so that there are five zones inside the event horizon.

hslope controls the degree of coordinate compression toward the midplane in MKS coordinates — smaller values concentrate more zones near $\theta = \pi/2$; 0.3 is a common choice for torus problems. mks_smooth, poly_xt, and poly_alpha are FMKS coordinate shape parameters (only used when DEREFINE_POLES = 1).

Electron thermodynamics

[dbl] game    = 1.333333
[dbl] gamp    = 1.666667
[dbl] fel0    = 0.01
[dbl] tptemin = 0.001
[dbl] tptemax = 1000.0

These parameters are only read when ELECTRONS = 1 is set in parameters.h. game and gamp are the adiabatic indices for electrons and ions respectively; the standard choices are $\gamma_e = 4/3$ (relativistic electrons) and $\gamma_p = 5/3$ (non-relativistic ions). fel0 is the initial fraction of dissipated energy assigned to electrons; it also sets the electron entropy in the initial condition. tptemin and tptemax clamp the ion-to-electron temperature ratio $T_p/T_e$ to a physically reasonable range.