resolve

Documentation: http://ift.pages.mpcdf.de/resolve

Resolve aims to be a general radio aperature synthesis algorithm. It is based on Bayesian principles and formulated in the language of information field theory. Its features include single-frequency imaging with either only a diffuse or a diffuse+point-like sky model as prior, single-channel antenna-based calibration with a regularization in temporal domain and w-stacking.

Resolve is in beta stage: You are more than welcome to test it and help to make it applicable. In the likely case that you encounter bugs, please contact me via email.

Installation

• Install nifty8, ducc0, matplotlib, scipy (see Dockerfile)
• Optional dependencies are:
  • For reading measurement sets, install python-casacore.
  • For reading and writing FITS files: astropy.
  • Some operators support jax.

Related publications

• The variable shadow of M87* (arXiv).
• Unified radio interferometric calibration and imaging with joint uncertainty quantification (doi, arXiv).
• Radio imaging with information field theory (doi, arXiv).

How to run the demos

Basic imaging with automatic weighting

• Download the data and unpack it.
• Change the path under [data] in cygnusa.cfg to the path where the data is located.
• Run

resolve cygnusa.cfg

or, if you have mpi4py installed:

mpirun -np <ntasks> resolve cygnusa.cfg

which should speed up the computation. The number of threads used per task can be set via -j. The number threads multiplied by the number of MPI tasks should not exceed the number CPU cores available on the machine.

Multi-frequency imaging

• Download the data and unpack it.