Version 2.2 released

[Mike Jarvis]

Hi all.  We have just released GalSim version 2.2.0.  You can now view the change log online.  I've also copied it below for those who would rather just read it in email.

Peace,

Mike

On behalf of all the GalSim developers

Changes from v2.1 to v2.2

The most visible change in this version is likely the new documentation.. Other important changes are updates to the WFIRST module to conform to Cycle 7 specifications and some efficiency improvements in the atmospheric PSF modeling.

See below for a complete listing of all features, bug fixes, etc. The numbers in parentheses refer to the GitHub issues where the change was made. cf. https://github.com/GalSim-developers/GalSim/milestone/19?closed=1

Deprecated Features

• Deprecated the nominally private class galsim.correlatednoise._BaseCorrelatedNoise. If you were using it for any purpose, you should now use galsim.BaseCorrelatedNoise. (#160)
• Deprecated the class RandomWalk in favor of RandomKnots. (#977)
• Deprecated the tol parameter of the various Interpolant classes. Users should use thekvalue_accuracy parameter of gsparams instead. (#1038)

API Changes

• Removed functionality to store and re-load WFIRST PSFs, and to obtain multiple WFIRST PSFs simultaneously. (#919)
• Changed the function signature of StampBuilder.addNoise. If you have a custom module that uses that, you should remove the skip parameter (which had always been False). (#1048)

Changes to Shared Files

• Added option to set the galsim.meta_data.share_dir via environment variable GALSIM_SHARE_DIR. (#1014)
• Changed hosting of the RealGalaxy COSMOS catalog to Zenodo, which seems to have solved some flakiness in the previous hosting at the University of Manchester. (#1033)

Config Updates

• Added some more cumstomization hooks in the StampBuilder class. (#1048)
• Added quick_skip option to skip an object before doing any work. (#1048)
• Added obj_rng=False option to use the same rng for all objects in image. (#1048)
• Added rng_index_key option to use a different rng just for particular values. (#1048)
• Fixed @ strings to work with input objects as well as normal variables. (#1048)
• Fixed various minor bugs when doing complicated things with index_key and rng_num. (#1048)

Documentation Updates

• The documentation is now rendered in Sphinx, rather than Doxygen, which looks much nicer. The new docs are accessible at http://galsim-developers.github.io/GalSim/. (#160)

New Features

• Added FitsHeader.extend method. Also, read_header option to galsim.fits.read. (#877)
• Updated WFIRST WCS and PSF routines to use Cycle 7 specifications for detector configurations, pupil planes, and aberrations. In particular, there is now a different pupil plane image for shorter- and longer-wavelength bands. (#919, #1019)
• Enabled Zernikes up to 22 (previously 11) in WFIRST PSFs, and added dependence on position within the SCA. (#919)
• Improved efficiency of drawing RandomKnots objects when transformed. (#977)
• Added WFIRST fermi persistence model. (#992)
• Added r0_500 argument to VonKarman. (#1005)
• Improved ability of AtmosphericScreen to use shared memory in multiprocessing context. (#1006)
• Use OpenMP when appropriate in SiliconSensor.accumulate (#1008, #1046)
• Added array versions of BaseWCS.toWorld and BaseWCS.toImage. (#1026)
• Exposed some methods of Interpolant classes that had only been in the C++ layer. (#1038)
• Added Zernike polynomial +, -, and * operators. (#1047)
• Added Zernike polynomial properties .laplacian and .hessian. (#1047)
• Added center option to the GSObject.drawImage method. (#1053)

Bug Fixes

• Fixed a couple places where negative fluxes were not working correctly. (#472)
• Fixed FITS I/O to write out comments of header items properly. (#877)
• Fixed error in the serialization of RandomKnots instances. (#977)
• Fixed error in PhaseScreenPSF when aberrations has len=1. (#1006, #1029)
• Fixed treatment of quantization in WFIRST detector effects. (#1016)
• Fixed error in BaseWCS.makeSkyImage when crossing ra=0 line for some WCS classes. (#1030)
• Fixed slight error in the realized flux of some profiles when using photon shooting. The bug was most apparent for Kolmogorov and VonKarman, where the realized flux could be too small by about 1.e-3. (#1036)
• Fixed error in Sersic class when n is very, very close to 0.5. (#1041)