the issue stems mostly from the fact that dist_freq became a dictionary by distance-bin instead of being an ndarray (before v1.0) - then there 2 different interpretation of that dict happen:
keys being dist_bin themselves in __init__, add_pair, add_pairs_fromdataframe_chunk ...
keys being bin_id of a given distance (i.e. bin_id = np.searchsorted(self._dist_bins, some_distance)) in places like from_file and __gfetitem__
This is minimal fix PR - where we switch to dist_freq - as a dict with dist-keys - i.e. changing from_file and __getitem__
Then there is just a simple bug/typo in flatten, where dist is incorrectly defined for the very last open-ended distance interval - it is fixed here as well
addressing stats issue #199
the issue stems mostly from the fact that
dist_freqbecame a dictionary by distance-bin instead of being anndarray(before v1.0) - then there 2 different interpretation of that dict happen:dist_binthemselves in__init__,add_pair,add_pairs_fromdataframe_chunk...bin_idof a givendistance(i.e.bin_id = np.searchsorted(self._dist_bins, some_distance)) in places likefrom_fileand__gfetitem__This is minimal fix PR - where we switch to
dist_freq- as a dict withdist-keys - i.e. changingfrom_fileand__getitem__Then there is just a simple bug/typo in
flatten, wheredistis incorrectly defined for the very last open-ended distance interval - it is fixed here as well