Skip to content

Latest commit

 

History

History
64 lines (47 loc) · 2.5 KB

CHANGELOG.rst

File metadata and controls

64 lines (47 loc) · 2.5 KB

Changes from version 1.3 to 1.4

Output file changes

  • Stars that were rejected as outliers are now included in the output file, but flagged with flag_psf=1. (#153)

API Changes

  • Changed the list of property names that a PSF uses for interpolation to psf.interp_property_names. (#150)
  • Deprecated some redundant or potentially unclear type names. (#151) * GSObjectModel -> GSObject * GPInterp -> GP or GaussianProcess * kNNInterp -> KNN or KNearestNeighbors * Star -> StarImages
  • Deprecated the include_reserve=True option for outlier rejection. Now that all objects are preserved in the output file, reserve stars that would have been rejected are marked as such, so you can choose whether or not to use them for any diagnostic tests. (#153)
  • Deprecated copy_image=False option to psf.drawStar and psf.drawStarList. (#155)

Performance improvements

  • Updated the DES-specific parameters for the Optical PSF model to be more accurate. (#138)
  • Improved the flux/centroid-finding step of each iteration to be slightly more accurate. (#154)

New features

  • Added mirror_figure options for the Optical PSF model. (#138)
  • Added trust_pos option in the input field. This tells Piff not to adjust the star positions when fitting the PSF. (#154)
  • Added fit_flux option for GSObject and PixelGrid models. (#155)
  • Added init option for all models. (#155)
  • Added SumPSF PSF type. (#157)
  • Added ConvolvePSF PSF type. (#162)
  • Added dof as a column in the stars section of the output file. (#165)

Bug fixes

  • Fixed a number of minor errors and inefficiencies in the Optical class. (#156)

Changes from version 1.4.0 to 1.4.1

  • Added an image of the average star and model in the StarImages output plot. (#167)
  • Changed the default behavior of the StarImages plot to include the average star and model. To recover the old version without these images, use include_ave = False. (#167)
  • Fixed a bug where reflux errors could cause the fit not to converge even though the offending stars had already been flagged and removed from consideration.
  • Fixed an error in TwoDHist stats that T_model was incorrectly using the star data.
  • Fixed an error that caused the candidate stars not to display correctly in SizeMag plot.
  • Added an option to show large T candidate stars in SizeMag plot that would otherwise be off the top of the plot.