new GalSim version: v0.5

[Rachel Mandelbaum]

We are pleased to announce a new tagged version of GalSim, v0.5, which 
can be obtained from our GitHub page, 
https://github.com/GalSim-developers/GalSim/ 
or for Mac users, from macports (see detailed 
instructions in 
https://github.com/GalSim-developers/GalSim/blob/master/INSTALL.md ). 
Below is a list of changes from the previous version, reproduced from 
CHANGELOG.md in the version that was tag (link directly to file: https://github.com/GalSim-developers/GalSim/blob/9e0d7b5e87fbaf50062b32a704ef0d439bf5c7f9/CHANGELOG.md ):

• Added document describing the operations being carried out by the lensing engine when it draws shears according to a user-specified power spectrum. (Issue #248)

• Added a simple prescription for adding diffraction spikes, due to secondary mirror / instrument support struts obscuring the pupil, to the OpticalPSF. (Issue #302)

• Added the ability to draw lensing shears and convergences self-consistently from the same input shear power spectrum. (Issue #304)

• Added a utility that can take an input set of shears on a grid, and reconstruct the convergence. (Issue #304)

• Sped up the HSM module. (Issue #340)

• Added the ability to modify parameters that control the precise rendering of GSObjects using the new GSParams class. (Issue #343, #426)

• Added Shapelet class (sub-class of GSObject) for describing shapelet profiles. (Issue #350)

• Made various speed improvements related to drawing images, both in real and Fourier space. (Issue #350)

• Changed obj.draw() to return the added_flux in addition to the image in parallel to existing behavior of drawShoot. (Issue #350)

• Added des module that add some DES-specific types and paves the way for adding similar modules for other telescopes/surveys. Specifically, there are classes for the two ways that DES measures PSFs: DES_Shapelet and DES_PSFEx, demoed in examples/des.py and examples/des.yaml. (Issue #350)

• Enabled InputCatalog to read FITS catalogs. (Issue #350)

• Added FitsHeader class and config option. (Issue #350)

• Added the ability to read/write to a specific HDU rather than assuming the first hdu should be used. (Issue #350)

• The ImageCorrFunc has been superseded by the CorrelatedNoise, which like the GaussianNoise, PoissonNoise etc. classes inherits from the BaseNoise. The class contains all the correlation information represented by the ImageCorrFunc, as well as the random number generator required to apply noise. (Issue #352)

• Similarly the get_COSMOS_CorrFunc() is replaced by the getCOSMOSNoise() function, which now initializes a Noise model with a stored random number generator. (Issue #352)

• Bug fixed in the generation of correlated noise fields; formerly these erroneously had two-fold rotational symmetry. (Issue #352)

• The correlated noise classes now have an applyWhiteningTo() method. The purpose of this function is to add noise to images that contain correlated noise; the power spectrum of the added noise is specifically designed to result in white (uncorrelated) noise in the final image. (Issue #352)

• Added the ability to modify algorithmic parameter settings for the moments and shape measurement routines using the new HSMParams class. (Issue #365)

• Deprecated AtmosphericPSF and Ellipse classes. (Issue #372) (They will both eventually be replaced by significantly different functionality, so they should not be used.)

• Changed HSM routines to require an explicit galsim.hsm. prefix and edited some of the function names. e.g.galsim.EstimateShearHSM -> galsim.hsm.EstimateShear. (Issue #372)

• Added kmin_factor and kmax_factor parameters to PowerSpectrum buildGrid function. (Issue #377)

• Changed the default centering convention for even-sized images to be in the actual center, rather than 1/2 pixel up and to the right of the center. This behavior can be turned off with a use_true_center=False keyword in draw or drawShoot. (Issue #380)

• Fixed bug in draw routine that led to spurious shears when an object is shifted. Probably also whenever the profile is not radially symmetric in fact. (Issue #380)

• Added index_convention option in config to allow for images with (0,0) as the origin rather than the usual (1,1). (Issue #380)

• Changed the name of the center item for the Scattered image type to image_pos, and added a new sky_pos which can instead specify the location of the object in sky coordinates (typically arcsec) relative to the image center. (Issue #380)

• Added LINKFLAGS to the list of SCons options to pass flags to the linker. (Issue #380)

• Added a new script, galsim/pse.py, that contains a PowerSpectrumEstimator class that can be used to estimate the shear power spectrum from a set of shears defined on a grid. The main functionality of PowerSpectrumEstimator actually does not require an installed version of GalSim, just Python 2.6 or 2.7 and NumPy. (Issue #382)

• Added ability to truncate Sersic profiles with optional trunc parameter. (Issue #388)

• Fixed some bugs in the InterpolatedImage class, one related to padding of images with noise fields and one related to real-space convolution. (Issues #389, #432)

• Added trefoil to optical aberration. (Issue #390)

• Added utilities galsim.utilities.compare_dft_vs_photon*. (Issue #401)

• Fix bugs in obj.drawK() function. (Issue #407)

• Improved speed and accuracy of non-truncated Moffat. (Issue #407)

• Fix some errors in code and documentation of the fits module related to writing to an HDUList. (Issue #417)

• Add new function galsim.fits.writeFile. (Issue #417)

• Fixed some issues with image arithmetic (failure to check/respect shape and scale), added the ability to raise all elements in an image to a float/int power. (Issue #419)

• Added ability to specify Sersic profiles by the scale radius. (Issue #420)

• Fixed some bugs in the Sersic class that were leading to low level ringing features in the images when drawn with FFTs. In addition, now Sersic with n=0.5 or 1 agree with Gaussian and Exponential at the expected accuracy tolerance. (Issue #426)