Suave

=================

**Breaking Changes**
---------------------
- Package will be renamed to ``corrfunc`` from ``Corrfunc``

New features
------------
- conda installable package
- GPU version

==================
This is a bug-fix release and contains general code quality improvements.


Enhancements
------------
- A new helper routine to find the combination of (RA, DEC) refinements that produces fastest runtime in ``DDtheta_mocks`` [216]
- Further testing via GitHub Actions [220]
- Added Ubuntu-Xenial on Travis [222]

Bug fixes
----------
- Fixing docs build failure on Travis [215]
- Fixing compile failure on missing 'CC' in environment [226]

==================
This is a bug-fix release and contains general code quality improvements.


Enhancements
------------



Bug fixes
----------
- Installation does not require python(3)-config anymore [209, 211]
- Better handling of terminal colours for unknown terminals [209]
- Prevent incorrect calculations with periodic boundaries for large ratios of (zmax, Rmax) to Lbox [210]

===================
This is a release for bug-fixes and general code quality improvements. Travis
now also tests for ``python3.7``.


Enhancements
------------
- Improved code quality and adherence to PEP8 [189]
- Documentation no longer shows duplicate entries [205]



Bug fixes
----------
- Incorrect calculations for non-native endian data [191]
- Large Rmax to Lbox ratio now supported for periodic boundaries [192]
- Workaround for GNU Assembler bug causing incorrect calculations [196]
- Only report runtime isa support if we also have compiler support [200]
- Example code to illustrate how to code custom weights with AVX512F [205]

================

Enhancements
------------
- Reduce memory footprint of the cell pairs [186]

==================

**Breaking Changes**
--------------------

New features
------------
- AVX512F kernels for all pair-counters, faster code from new optimizations using the minimum separation between pairs of cells, option to use the input particle arrays directly and not make a copy of the particle positions, internal code changes to (hopefully) achieve better OpenMP scaling [167, 170, 173]

Bug fixes
---------
- Fix segmentation fault in vpf_mocks [168]
- Fix automatic uniform weights array when only one set of weights (or a scalar) is passed [180]
- Fix memory leak due to Python reference leak when using weights [181]