Polarized covariant radiative transfer in C, for imaging black hole accretion systems such as those being imaged by the Event Horizon Telescope. ipole
is described in Moscibrodzka and Gammie 2018.
This file details how to use ipole
and includes a note about when it might not be be appropriate for your project.
The only prerequisites are the GNU Scientific Library (GSL) and HDF5. Specifically, the executable h5cc
should be in your PATH
.
On Illinois EHT/BH cluster, this means loading the modules
$ module load gnu hdf5
On Linux machines, if you have root access, these can be installed to the system with
$ sudo apt install libhdf5-dev libgsl-dev
or
$ sudo dnf install hdf5-devel hdf5-static gsl-devel
The packages are straightforward to compile from source as well.
The native macOS version of clang
does not support OpenMP parallelization, so on macOS with brew
, install
$ brew install llvm gsl hdf5
these should all be binary packages, no compiling should be required.
Then just build by running
$ make
with or without any of the usual make
options, e.g., -j
. You can
make ipole
into any destination directory by going there and then
specifying the location of the makefile with -f
. Note that
ipole
does not support building in directories that contain spaces.
If you are using your own version of HDF5, you can specify where to find it with
$ make CC=/path/to/h5cc
... or compile with parallel HDF5:
$ make CC=/path/to/h5pcc
A particular fluid model or data format can be specified with
MODEL=model_name
. For nearly all GRMHD fluid data, the iharm
model
should be used -- this is the default if no model is specified. After building
a model, be sure to run make clean
before building a different one.
The optional argument NOTES=
can be any string containing no spaces.
This string will be "baked-in" to the executable and printed out whenever the
program is run.
To facilitate runs over many files with similar parameters,
ipole
can read parameter files, specified with
-par path/to/file.par
.
Parameter files should be lists of key value pairs, one per line.
# Example parameter
thetacam 90.0
Empty lines and lines beginning with the #
character are ignored. A list of supported parameters for each model can be found in the example.par
file in its directory. Expected keys that do not appear in the parameter file will be set to a backward-compatible default value (see, e.g., par.c
for more information). Unexpected keys are silently ignored for forward compatibility.
The same parameters can also be specified on the command line by using double-dash syntax,
$ ./ipole -par default.par --key1=value1 --key2=value2
$ ./ipole --freqcgs=230.e9 --MBH=6.2e9 --M_unit=1.e25 --thetacam=17 --dump=/path/to/dump.h5 --outfile=image.h5
In all cases, ipole
reads parameters (and parameter files) in the order they are specified, overwriting when encountering a repeated parameter name. Thus for most use cases the parameter file should be specified first.
There are two command-line specific flags:
-unpol
: Only compute using approximate "unpolarized" transfer equation. This speeds up computation and is useful for (e.g.) fitting Munit.
-quench
: Don't write any files to disk.
These options do not affect the rest of the arguments.
This version of ipole can output a file containing everything it used to compute a particular pixel. This includes the coordinate locations of each step along the geodesic, as well as several scalars at each point (see the format doc for specifics).
To activate tracing for a pixel (i,j), measured from the bottom left with i rightward along the x axis and j upward along y, run ipole with
$ ./ipole -par parameters.par --trace=1 --trace_i=i --trace_j=j
an alternative option trace_stride
is available to trace every N pixels
across the entire image.
The ipole output format for images is documented here. Converters are available to translate output into FITS images, or into the old 7-column text format (i,j,F,I,Q,U,V).
The ipole output format for traces is documented here
ipole
is not a general-purpose imaging code. It has been designed, tested for,
and used for modeling of particular sources. It may crash or
produce incorrect results in other use cases. We strongly recommend you test
the code in an appropriate regime before you use it.
The ipole algorithm is described in Moscibrodzka and Gammie 2018. The original version can be found here. Each radiative transfer step is based on an analytic solution to the polarized transfer equation assuming constant coefficients. This is intended to produce sensible results even when absorption or Faraday rotation and conversion are large in a single step. A comparison of polarized relativistic radiative transport schemes can be found in Prather (2021, in prep).
If you are imaging GRMHD simulations, you will find guidance on how a GRMHD simulation and analysis pipeline fits together in the PATOKA pipeline paper, Wong (2021, in prep).
ipole
treats synchrotron emission and absorption but not bremsstrahlung or Compton
scattering. The transfer coefficients (emissivities, absorptivities, and rotativities)
are drawn from analytic fits presented
in Dexter 2016 , Pandya
2016, Pandya
2018, and
Marszewski 2021, with a summary in the latter. All
coefficients assume that the electron distribution function is isotropic and that the frequency
is large compared to the plasma frequency and cyclotron frequency. ipole implements fits for
several electron distribution, including a thermal (Maxwell-Juttner) distribution, a power-law
distribution, and a kappa distribution.
We welcome contributions and pull requests from the community implementing new features or
issue fixes. Not every new application fits the ipole
scope, however, and we reserve
the right to choose which features to support.
You are welcome to file issues in this repository to ask questions of general interest and
report when ipole
doesn't work for you. Please keep in mind that while bugs related to
ipole
's primary purpose will likely be given attention quickly, new uses and extensions
can only be supported on a best-effort basis.
ipole is an open source, community code that originated in the AFD group at Illinois and in the Astrophysics group at Radboud. Unfortunately we are not funded to provide support for all users, so you may need to do some detective work yourself. We will prioritize fixing problems that are manifestly bugs, rather than problems encountered in extending ipole into new parameter regimes.
Funding for the development of ipole at Illinois was provided by the National Science Foundation under multiple grants, including NSF AST 17-16327, NSF OISE 17-43747, NSF AST 20-07936, and NSF AST 20-34306.