-
Notifications
You must be signed in to change notification settings - Fork 1
GRMHD Output Format
Fluid simulation output data are saved in HDF5 "dump" files according to the format specified on this page. This unified format is meant to enable interoperability between the various hydro+ and radiative transfer codes we provide.
In the specification below, we use "double" to mean the 8-byte wide floating-point type, "float" to mean the 4-byte wide floating-point type, "int" to mean the 4-byte wide floating point type, and "string" to mean fixed-size zero-terminated character arrays of size 20. Arrays are indicated by the use of trailing square brackets "[]" where the number of sets of brackets corresponds to the dimension of the array. All numerical values are stored in little-endian format.
Just interested in reconstructing four-vector and metric tensor quantities? Check here.
The data in our output files can be schematically separated into two parts: header information and fluid-state data. Because many fluid dumps typically correspond to the same fluid simulation, we further divide header information into data that does and does not change over the course of the simulation. Data that is constant over the entire simulation is stored in the /header group. Other dump-specific information is rooted at the document root /.
At minimum, header data includes the following information:
| Description | Path | Type | Notes |
|---|---|---|---|
| simulation time | /t |
double | in code units |
| fluid gamma | /header/gam |
double | |
| logical grid dimensions |
/header/n1, /header/n2, /header/n3
|
int | |
| metric name | /header/metric |
string | |
| names of primitive variables | /header/prim_names |
string[] | |
| number of primitive variables | /header/n_prim |
int | |
| simulation end time | /header/tf |
double | in code units |
| version string | /header/version |
string |
Optional entries into header data include:
| Description | Path | Type | Notes |
|---|---|---|---|
| use of electron temperature model | /header/has_electrons |
int | flag value 0 or 1 for true or false |
| use of radiation physics (sets units) | /header/has_radiation |
int | flag value 0 or 1 for true or false |
| notes | /header/notes |
string | |
| thermodynamic gamma of electrons | /header/gam_e |
double | if has_electrons |
| thermodynamic gamma of ions | /header/gam_p |
double | if has_electrons |
| length unit (code -> cgs) | /header/units/L_unit |
double | if has_radiation |
| mass unit (code -> cgs) | /header/units/M_unit |
double | if has_radiation |
| time unit (code -> cgs) | /header/units/T_unit |
double | if has_radiation |
Header data also includes geometric information about the grid within the /header/geom group. The logical grid description is given by the six double values described below:
| Description | Path | Type | Notes |
|---|---|---|---|
| logical grid left edge |
/header/geom/startx1, /header/geom/startx2, /header/geom/startx3
|
double | |
| logical zone width |
/header/geom/dx1, /header/geom/dx2, /header/geom/dx3
|
double |
Finally, the header also includes metric parameters for the metric identified in the /header/metric string variable. These parameters, along with an understanding of the specified metric form, enable conversion between logical coordinates and (especially) Kerr-Schild coordinates. Metric variables are rooted at /header/geom/$METRIC where $MERTIC is defined by /header/metric.
For a mathematical description of how to use metric parameters, click here.
The bulk of the fluid data is stored as primitive variables (in code units) in a single four-dimensional array /prims of data type float. This array should be indexed by position and then variable number, where variable number corresponds to the desired primitive as listed in the "names of primitive variables" array in the header. For MHD+ simulations, the first eight variables will always be
RHO, UU, U1, U2, U3, B1, B2, B3
so to access the density (rho) of the cell at zone i, j, k, you should do
/prims[i][j][k][0]
Any indices 7 < i < NVAR correspond to extra variables moving passively with the fluid. Currently in iharm3d these are all entropy values, first for the total fluid (k_tot = /prims[i][j][k][8]) and then for electrons, under different heating models. These are written at index 9, 10, 11, 12 as KEL0, KEL1, KEL2, KEL3 correspond to KAWAZURA, WERNER, ROWAN, and SHARMA respectively. See iharm3d source file electrons.c for details. Note that the electron ordering is changed from v3.5 & 3.6, which placed a single electron entropy KEL at index 8 and KTOT at index 9.
The contravariant components of the four-current ja, calculated according to Fab;b = 4 π ja, are stored in a four-dimensional array /jcon of data type float and should be indexed according to position and then component index a as
/jcon[i][j][k][a]
/header/cour float
/fel0 float
/gam float
/tf float
/gam_e float # if has_electrons
/gam_p float # if has_electrons
/tptemax float # if has_electrons
/tptemin float # if has_electrons
/header/gridfile string
/metric string
/reconstruction string
/version string
/prim_names string[NVARS]
/header/n1 int
/n2 int
/n3 int
/n_prim int
/n_prims_passive int
/has_electrons int
/has_radiation int
/header/units/L_unit float # if has_radiation
/M_unit float # if has_radiation
/T_unit float # if has_radiation
/RHO_unit float # if has_radiation
/U_unit float # if has_radiation
/B_unit float # if has_radiation
/Ne_unit float # if has_radiation
/Thetae_unit float # if has_radiation
/header/geom/dx1 float
/dx2 float
/dx3 float
/startx1 float
/startx2 float
/startx3 float
/n_dim int
# if /header/metric == "mks", then
/header/geom/mks/r_eh
/r_in
/r_out
/a
/hslope
# if /header/metric == "mmks", then
/header/geom/mmks/r_eh
/r_in
/r_out
/a
/hslope
/mks_smooth
/poly_alpha
/poly_xt
/dt float
/dump_cadence float
/full_dump_cadence float
/t float
/is_full_dump int
/n_dump int
/n_step int
/jcon float[N1][N2][N3][4]
/prims float[N1][N2][N3][NVARS]
# nothing beyond this point is guaranteed
/fail int[N1][N2][N3]
/divB float[N1][N2][N3]
/gamma float[N1][N2][N3]
/extras ...