All challenge data now available!

Servers

We are pleased to announce the release of the ‘variable PSF’ and ‘full’ simulations, which are the last 8 branches of the challenge dataset, as well as updated 'multiepoch’ simulations. As usual, you can download them from the challenge data download page.

A few other notable updates:

  1. We have made a new version of the multiepoch simulations due to this issue. The new version is the same format as before, but you will have to re-download that data in order to participate in those branches of the challenge. We are sorry for the inconvenience, but the new data are now available!
    If you’re using our presubmission script to get your results for multiepoch/variable shear in the proper format for submission, please make sure to pull the latest version from our code repository, since the new simulations have new subfield offsets.
  2. On our public code repository,
    we have just released an example script for estimating PSFs from the star fields in the variable PSF branches.
  3. As stated in the handbook and on the leader board site, the normalization of our metrics used to evaluate methods was based on an assumption about the typical S/N of shear estimation, so we thought there might be a need to renormalize if the typical S/N for real methods turns out to be different from our assumption. Early submissions indicate that our estimate of S/N was overly optimistic, so we do indeed plan to carry out a metric renormalization, which for the case of constant shear is nonlinear but strictly rank-preserving.

    The variable shear case is in fact more complicated and we will send a separate message with some plots explaining what is going to happen there. This should take place within the next 2 weeks, and will be described in a modified version of the handbook that we will post to arxiv around that time.

Thanks for participating, and we look forward to seeing many of you at the mid-challenge meeting in less than a month!