crpurcell / rm-tools

Stand-alone scripts to analyse Faraday spectra in 1- and 3-D.

RM-Tools: Software to analyse Faraday rotation

Python scripts to perform RM-synthesis, RM-clean and QU-fitting onpolarised radio spectra.

IMPORTANT: the actively developed RM-tools directory has moved. RM-tools can now be found at

https://github.com/CIRADA-Tools/RM

Further development of RM-tools will take place in that repository, so all feature requests, bug reports, and other communication should be done there. This repository will no longer be updated. The main branch contains the last update of the Python 2 version of RM-tools, while the CIRADA branch contains the Python 3 version as of May 2019.

RMtools_1D ... 1D utilities to analyse Faraday spectra. RMtools_3D ... 1D utilities to analyse Faraday spectral cubes. RMutils ... Core functions used by the analysis utilities.

IMPORTANT NOTE: The master and develop branches of RM-tools are no longer being actively maintained. The cirada branch has been updated to Python 3; the master branch has been left for Python 2 compatability but will not receive updates except for critial bugs (at our discretion). The most recent version of the cirada branch can be found at the new repository linked above.

Questions, bug reports, and feature requests can be sent to Cameron Van Eck, cameron.van.eck (at) dunlap.utoronto.ca.

More information on the Canadian Initiative for Radio Astronomy Data Analysis (CIRADA) can be found at cirada.org.

rm-tools's Issues

Issue with 2D arrays and RM-CLEAN

For 2D arrays (e.g. HEALPIX), I think there is a bug currently present in RM-CLEAN. Specifically, in util_RM.py -> do_rmclean_hogbom, lines 504 and 505. Currently, they read:

elif nDims==2:
        dirtyFDF = np.reshape(list(dirtyFDF.shape[:2])+[1])
        RMSFArr = np.reshape(list(RMSFArr.shape[:2])+[1])

This is syntactically incorrect for np.reshape. I think they should be:

elif nDims==2:
        dirtyFDF = np.reshape(dirtyFDF, list(dirtyFDF.shape[:2])+[1])
        RMSFArr = np.reshape(RMSFArr, list(RMSFArr.shape[:2])+[1])

Following this, in the main loop (i.e. lines 546+) there is an issue with the indexing of 2D arrays. I found that for a 2D array of shape (N_phi_Chan, N_pix), as produced by RM-Synthesis, the indexing variables xi and yi were not in the correct order for lines 586-588, specifically:

# Restore the CC * a Gaussian to the cleaned FDF
  cleanFDF[:, yi, xi] += \
  gauss1D(CC, phiPeak, fwhmRMSFArr[yi, xi])(phiArr_radm2)

That is, for fwhmRMSFArr the order was flipped. I was able to temporarily sidestep this issue by replacing this line with:

# Restore the CC * a Gaussian to the cleaned FDF
  cleanFDF[:, yi, xi] += \
  gauss1D(CC, phiPeak, fwhmRMSFArr.T[yi, xi])(phiArr_radm2)

But, this seems like it might be missing the underlying issue.