This repo was created to track progress on 2PT validation (LSS 1.1.X tasks).
There are three pieces that make things up:
https://github.com/damonge/CoLoRe -- highly optimized C code that makes lognormal mocks.
https://github.com/slosar/fastcat -- public repo that contains a Catalog object for abstract storage of mock catalogs (used to be able to be use randomfield as backend, but this has (temporarily?) been pushed into attick)
https://github.com/DarkEnergyScienceCollaboration/2pt_validation -- this repo that contains various glues between CoLoRe and fastcat, etc.
Before anything else: if you're planning on using this repo, it's highly recommendable to read doc/Cookbook.md in detail rather than anything else!
To see how this works in action, look at the following ipython notebook,
inspired by Javier's hack last week. It loads a fastcat catalog and then uses treecorr to calculate correlation function. As you know, I'm not a fan of correlation functions, but it works great as an example. Note that fastcat objects are not braindead, for example, when loading the HDF5 file it will work out that window function is a simple declination cut and will set up and appropriate window object -- this can then be used to generate random catalogs. However, at the same time HDF5 format was written in a way that doesn't require you to go through the python route so you can use any language that you want. The format is described in this dir:
https://github.com/slosar/fastcat/tree/master/hdf5_format_doc
The 0.1 format is what was used for the 10 mocks last week and 0.2 is a minor update since people expressed wish that the mocks should really be unique, all the data mucking is done on the mock-creation level (i.e. I already add gaussian PZ errors for you etc.
There are three basic directories:
Files relevant to this project are stored under
/project/projectdirs/lsst/LSSWGon NERSC systems (e.g. cori)
The following subdirectories are available: