Rimphony is a tool to calculate the eight coefficients needed to do numerical radiative transfer of polarized synchrotron emission. Unlike some comparable tools, it supports particle distributions that are anisotropic in pitch angle, and can calculate Faraday rotation and conversion coefficients. And, of course, it’s fully open source!
Rimphony began life as an experimental reimplementation of the Symphony synchrotron coefficient code in the Rust language. Symphony comes out of Charles Gammie's group and is primarily the work of Alex Pandya. Its algorithms are described in Pandya et al (2016) and Leung et al (2011).
Rimphony has adapted the computational framework of Symphony and adds new routines to calculate Faraday rotation and conversion coefficients using the formalism developed by Heyvaerts et al (2013; DOI:10.1093/mnras/stt135). Some of the Symphony implementation details have also been altered to increase speed and reliability in challenging corners of parameter space. While rimphony is not yet described in a refereed publication, based on the results of its built-in validation test suite and benchmarking, I (PKGW) feel comfortable recommending it for use in research projects.
As a side note, my overall takeaway is that Rust is an excellent language for this kind of project. In my opinion, the code of Rimphony is substantially more readable and extensible than that of Symphony, and the infrastructure of the Rust language make compilation, testing, benchmarking, documentation, and distribution all straightforward.
You need to have the Rust development tools installed. Follow these easy instructions if you don’t.
You also need to have GSL and its development files installed. This library is available on most OS’s and software packaging systems.
Once Rust is available, hopefully all that is needed is to run
cargo build
and then
cargo test
to run the test programs. The command
cargo run --example one-powerlaw-direct
will compile and run a test program.
To actually use this code for an application, you would need to write a
program that depended on the Rimphony “crate” (a Rust term), used the relevant
API calls, and did something with the results. For instance, the example
program crank-out-pitchypl computes and prints coefficient values for random
input parameters. I use these as training data for a neural network that can
quickly compute approximate synchrotron coefficients for arbitrary input
parameters — the neurosynchro
project.
The layout of the files in this repository follows standard Rust practices.
- The
gsl-syssubdirectory contains a subproject that implements some minimal bindings to the GSL numerics library. - The
leung-besselsubdirectory contains a subproject that provides access to the fast Bessel function evaluator described in Appendix B of Leung et al (2011). - The
srcsubdirectory contains the main code. - The
testssubdirectory contains some simple tests. - The
benchessubdirectory contains some minimal benchmarking code. - The
examplessubdirectory contains a few example programs that exercise the library.
Alex Pandya and Mani Chandra are the main authors of Symphony. Symphony builds on Harmony, primarily written by Po Kin Leung. This Rust reimplementation is by Peter K. G. Williams. If using this code in research work, cite the papers mentioned above.
This code is copyright Peter Williams and collaborators, and is licensed under version 3 of the GPL, like Symphony.