Corrfunc Versions Save

New

• New pair counters for DD(s, mu) in theory and mocks (see PR #132, #166 )

Improvements

• Failed doctests now fail the build on TRAVIS
• The testing harness can now run tests all combinations of code options for any given kernel
• API now converts arrays into native-endian format (see Issue #140)
• Better handling of python errors in Makefile

Notes

Last Corrfunc release that will support python==2.6

New

• New pair counters for DD(s, mu) in theory and mocks (see PR #132)

Improvements

• Failed doctests now fail the build on TRAVIS
• The testing harness can now run tests all combinations of code options for any given kernel
• API now converts arrays into native-endian format (see Issue #140)

Minor updates, mostly related to user experience:

• New doc pages about weights
• New doc page with a detailed table for all the available clustering statistics and the interfaces
• Directory structure now is more self-explanatory

New Features:

• Cleaner API: Ability to handle both doubles and floats at runtime
• Runtime feature control - the output of all the codes can now be changed at runtime. Previously, even to simply output ravg required a complete recompilation of the package.
• Weighted correlation functions (with arbitrary weighting functions) will be handled by all correlation routines. (Implemented by @lgarrison )
• Automatic detection of CPU: no more illegal instructions and runtime errors
• Future-compatible: The libraries should (hopefully) be binary compatible with future versions.
• Documentation: Read all of the API at corrfunc.rtfd.io
• doctests now run automatically providing another layer of testing

Bugfixes:

• Crash in DDtheta_mocks when using the bruteforce mode (#107)
• Incorrect results in DDtheta_mocks for angular separations (#111)
• Misleading documentation about np.linspace and the number of bins (#96)

New in version 1.1.0:

• Added SSE support and runtime dispatch to all of the theory correlation function routines.

Bugfix:

• Missing factor of 4/3 in \xi normalization. ALL previous \xi calculations were wrong.

Corrfunc v1.0.0 release.

Updates from v0.2.3:

1. Smoother installation process
2. Detection of CPU capabilities at compile-time
3. Various performance fixes

Bugfixes:

1. Double-counting bug in `wp/xi' have been fixed (See #50)
2. Bug in `wp/xi' where empty cells caused a crash is fixed with commit 4c52530e92011638adefeb432c4fcf478c6cd2f6

If you cloned the repo, then do the following to upgrade:

git pull
make -j4 distclean
make 
make install
python setup.py install (--user)

If you used pip, then the following should do the job:

pip install Corrfunc --upgrade

Added a lot of checks to ensure smoother compiling on MACs. Updated version on PyPI as well (the one difference being that the PyPI version suppresses all progressbar output)

Release to mark version uploaded to PyPI.

** Changes **

1. Package size is now smaller by ~50 MB (removed ascii galaxy catalog).
2. USE_AVX is now enabled by default.

Latest version that is mostly ready for publishing. Can not upload to PYPI though since the tar.gz file is too large (~100 MB, limit is ~40 MB).