This repository contains example code magnetic-retrival-example.py for retrieving the magnetic field geometry of ultracool dwarfs (or any magnetised body) from detections of radio emission as demonstrated in Kavanagh et al. (2024). The model assumes that the emission originates from active field lines within a dipolar magnetic field that rotates rigidly with the object, driven by the electron cyclotron maser instability. The nested sampling algorithm Ultranest is used to explore the likelihood space, which provides the posterior distributions for each model parameter.
The example code is set up to read in the file data.csv, which contains columns of the rotation phase and Stokes V flux density at two different frequency bands (including their errors). Note that the frequency bands of 950-1150 MHz and 1300-1500 MHz are hard-coded in the script. A joint likelihood is computed for these two bands, although this can be modified easily for a single observing band. The code can be parallelised easily, and is scaleable for high-performance computing resources via e.g. mpiexec (note that this requires the package h5py). From our testing, splitting the task over more than 32 cores results in a slow down in performance.
Here is an example of the magnetic geometry inferred for the ultracool dwarf WISE J062309.94-045624.6 (J0623):
