Skip to content

Latest commit

 

History

History
77 lines (70 loc) · 4.07 KB

CHANGELOG.md

File metadata and controls

77 lines (70 loc) · 4.07 KB
Raw

Changes from v1.1 to v1.2

Note: v1.2 is the official version we are tying the GalSim paper to. (Nearly published -- we are responding to the referee report now!) A few of the changes herein were made to address perceived deficiencies we noticed as we were writing the paper.

New Features

  • Changed name of noise whitening routine from noise.applyWhiteningTo(image) to image.whitenNoise(noise), parallel to image.addNoise(noise); use of noise.applyWhiteningTo() is deprecated. (#529)
  • Added an option to impose N-fold symmetry (for user-selected even values of N>=4) on the noise in images with correlated noise rather than fully whiten the noise called image.symmetrizeNoise(noise, N). (#529)
  • Added magnitudes as a method to set the flux of SED objects. (#547)
  • Added SED.calculateDCRMomentShifts and SED.calculateChromaticSeeingRatio convenience functions for estimating chromatic PSF moment shifts. (#547)
  • Added new methods of the image class to simulate two detector effects: nonlinearity and reciprocity failure. (#552)
  • Renamed the GSParams parameter alias_threshold to folding_threshold, a clearer term for the profile image folding in real space that this GSParam controls. (#562)
  • Modified the internals of noise generation by correlated noise models to use Hermitian symmetry, for greater efficiency. (#563)
  • Extended to the rotate, shear, and transform methods of ChromaticObject the ability to take functions of wavelength for the arguments. (#581)
  • Added a module to describe charge deflection in CCD pixels (also known as the "brighter-fatter" effect) following model of Antilogus et al (2014). (#524)
  • Make it possible for OpticalPSF to model non-trivially complicated obscuration and/or struts by allowing it to take an optional image of the pupil plane. (#601)
  • Added nx, ny, and bounds keywords to drawImage() and drawKImage() methods. (#603)

Bug Fixes and Improvements

  • Modified BoundsI and PositionI initialization to ensure that integer elements in NumPy arrays with dtype==int are handled without error. (#486)
  • Changed the default seed used for Deviate objects when no seed is given to use /dev/urandom if it is available. If not, it reverts to the old behavior of using the current time to generate a seed value. (#537)
  • Changed SED and Bandpass methods that return a new SED or Bandpass to attempt to preserve the type of the calling object if it is a subclass of SED or Bandpass respectively. (#547)
  • Changed the the file_name argument to CorrelatedNoise.getCOSMOSNoise() to no longer be required. The normal file to use is now installed along with GalSim (in the directory PREFIX/share/galsim), so that file can be used by default. (#548)
  • Fixed the dtype= kwarg used when initializing Image instances to interpret the aliases int and float as the numpy.int32 and numpy.float64 data types, respectively. Previously the behavior was unpredictable and platform dependent. (#571)
  • Added workaround for bug in pyfits 3.0 that affected galsim.fits.read. (#572)
  • Fixed the Image constructor so that if it is passed a NumPy array with the opposite byteorder as the native one on the system, it does not return an Image with different contents. (#594)
  • Fixed bug that prevented calling LookupTables on non-square 2d arrays. (#599)
  • Updated the code to account for a planned change in NumPy that array==None will be an element-wise comparison rather than equivalent to array is None. (#604)
  • Fixed a bug where the dtype of an Image could change when resizing, which should not be the case. (#604)
  • Defined a hidden __version__ attribute according to PEP 8 standards. (#610)

Updates to config options

  • Moved noise whitening option from being an attribute of the RealGalaxy class, to being a part of the description of the noise. (#529)
  • Added RandomPoisson, RandomBinomial, RandomWeibull, RandomGamma, and RandomChi2 random number generators, corresponding to the random deviate classes in the python layer. (#537)