manodeep / Corrfunc

⚡️⚡️⚡️Blazing fast correlation functions on the CPU.
https://corrfunc.readthedocs.io
MIT License
166 stars 51 forks source link

Test wtheta totncells #108

Closed manodeep closed 7 years ago

manodeep commented 7 years ago

Fixed two niggling issues with wtheta

  1. Binning in ra after binning in dec was a complete kludge. Now, the calculation follows from the haversine formula
  2. Extra cells that were allocated within gridlink_mocks but could never have any particles. Now, a first pass determines the total number of ra cells and keeps track of the ra offset for each dec bin.
manodeep commented 7 years ago

Apparently the fudge factor of 1.01 was not good enough. Now, I am no longer convinced that the fudge-factor was for numerical precision issues - need to re-check my math.

manodeep commented 7 years ago

The logic for the dz equation needs to be fixed. However, even in current state, this PR seems to fix #107

manodeep commented 7 years ago

I need a dip in the Ganges now. Upper, cleaner parts obviously :)