WSClean Version 1.3

Released 2014-08-05

In this version, it is now possible to select the new & supposedly fast multi-scale algorithm I’ve been working on. It works well without tweaking on some MWA data of Vela & Puppis A, but other telescopes / sets might require tweaking of the parameters (with -multiscale-threshold-bias and -multiscale-scale-bias). For those using Cotton-Schwab cleaning that includes XY/YX polarizations, I urge upgrading to this version for some of the bug fixes.

Changes:

  • Feature: multi-scale cleaning can now be selected by using wsclean -multiscale. This is a new, supposedly fast algorithm for multi-scale cleaning and has shown good initial results. It’s still somewhat experimental, and might require tweaking the threshold bias and scale bias parameters (see the help).

  • Feature: wsclean -version will now report version info

  • Bugfix: Fixing big bug in prediction for XY and YX – sign error in imaginary image values before prediction, on baselines with w>=0. This could potentially cause noisier/incorrectly deconvolved XY/YX images.

  • Bugfix: Fixing writing of wrong imaginary residual image – the actual saved image in XYi-residual was the real value of XY

  • Bugfix: Throw exception when cleaning XY/YX polarizations without joined pol

  • Bugfix: Fixing help on -joinpolarizations and -pol

  • Bugfix: Fixing copy constructor of fitsreader – causing segfault on complex image prediction

  • Bugfix: Fixing error in -predict mode with XY/YX polarizations

  • Bugfix: Fixing issue with -predict: wouldn’t use Kaiser-Bessel gridding by default

  • Bugfix: Fixing problem in -predict with uninitialized reader causing possibly undefined behaviour

  • Bugfix: Making sure that uvw samples outside the field of view do not force more w-layers

  • Bugfix: Fixing possible crash when imaging beyond horizon

  • Bugfix: Extra safety check on input parameters in WSInversion