bandersen441 / rtpipe

Pipeline for real-time fast transient detection in radio interferometric data

rtpipe

rtpipe (as in 'real-time pipeline') is a Python package for searching visibility data from radio interferometer data for fast (subsecond) transients. In contrast to traditional time-domain techniques used on large single-dish telescopes or "phased" interferometers, fast-sampled visibilities can precisely localize sources anywhere in the entire field of view. This package supersedes tpipe and uses a paradigm that defines how to break long (large) data into smaller, independent pieces with a single treatment (flagging, image gridding, calibration, etc.).

Requirements

• Python 2.7
• Standard scientific Python stuff: numpy, scipy, matplotlib, multiprocessing, bokeh
• pwkit (for access to CASA libraries)
• Cython (tested with 0.19.1)
• sdmpy (for reading SDM format data)
• pyFFTW (accelerated FFTs; tested with 0.92)

Install

conda install numpy scipy pwkit casa-tools casa-python casa-data jupyter bokeh cython matplotlib
pip install rtpipe

pwkit is needed to access CASA libraries and so the anaconda installer is thus required. Alternatively, one can build pwkit and environment variables set like so to see libraries of an existing CASA installation:

setenv CASA_HOME /home/casa/packages/RHEL6/stable/current
setenv PYTHONPATH ${PYTHONPATH}:${CASA_HOME}/lib/python2.7
setenv LD_LIBRARY_PATH ${LD_LIBRARY_PATH}:${CASA_HOME}/lib
setenv CASAPATH "${CASA_HOME} linux local ${HOST}"

The trick here is being sure that the python interpreter is binary-compatible with that used to build CASA.

Contributors

• Casey Law
• Peter Williams (CASA libraries and general advice)
• Paul Demorest (sdmpy)