June 30, 2014

On June 15, we discovered a problem with the locations of galaxies in the shear simulations that made verifying any shear submissions impossible, and a separate bug in the PSF convolution code for the cluster simulations. We are regenerating the simulations and aim to have them back online by July 10. We apologize for the inconvenience!

October 31, 2014


The DECam simulations have been re-submitted, as they were projected (and also convolved) onto the wrong pixel scale).


Release March 31, 2014, closing January 1, 2015 (NEW)

For the first release, we seek to isolate the algorithms that extract shears from galaxy images. In order to accurately probe the ability of algorithms to extract shear from the galaxy shapes, we use galaxy morphologies drawn from deep Hubble Space telescope images as our simulated galaxies. (See for a description of the input galaxy images and the process of creation of the images. However, to isolate the effect of shear extraction, the simulations do not include the following effects:

In future releases we will relax each of these assumptions one at a time to allow testing of cluster shear algorithms for each effect separately. For the current release, we produce two independent simulation sets.

Simulation Set 1:  Grid galaxy simulations

For the first simulation set, we produce grids of galaxies approximately regularly spaced galaxies.   The individual galaxies are separated to prevent any blending, allowing for the shear extraction to be tested cleanly.  Within an image, each galaxy is distorted by a reduced shear g as if it were in the lensing deflection field of a cluster.  All galaxies in the grid image have the identical reduced shear (magnitude and direction of the applied shear), but different images differ both in direction and magnitude of the applied shear.  

The images (both individual images and gzipped tarfiles) are available at the following locations:

For each image, the task is to measure the shape of every galaxy and to produce a catalog of galaxy shapes (and thus an estimate of the reduced shear) in that image.

Simulation Set 2:  Simple spherically symmetric clusters.

For the second simulation set, we produce simulations of simple clusters, to test the ability to extract shears that are not constant across the images.  Each image is distorted by a single, spherically symmetric NFW profile cluster centered on the image center.  The cluster and background galaxy redshifts are held constant (zcl=0.3, zgalaxy=1.5), so that the angular diameter distance ratios are constant for all sources involved.  The simulations span a 2Mpc x 2Mpc region around each cluster, and therefore span a very large range of reduced shears.   The simulated images are produced in the high S/N regime, but the source density is reduced relative to a realistic case to reduce the effect of isophote blending on the measured shapes.  The images are produced with a range of M200 values (and a corresponding NFW concentration parameter, as we do not mean to test the ability to independently constrain mass and concentration). 

The images (both individual images and gzipped tarfiles) are available at the following locations:

For each image, the task is to produce the shape of every galaxy and produce a catalog of galaxy shapes (and thus an estimate of either the mean reduced shear or the shear profile).

Submission methods

Submission may be made by e-mailing catalog files to  Catalog files should have the same name as the simulated images,  except with the .cat extension rather than .fits (i.e. would correspond to the catalog derived from grid1.fits).  We will provide weekly feedback results to submitters.  There is no limit to the number of resubmissions allowed (the goal is to improve algorithms, after all!)

Criteria for Evaluation

Evaluation is based on the mean offset and rms scatter between the reported and input reduced shears.  We will rank submissions both by overall match and by the slope of the error as a function of reduced shear, as this is a useful indicator of the range of validity of the algorithms.