Reduce usage of Fortran kD-tree

Description

Ideally, we wouldn't use the fortran kD-tree at all in yt because it adds an additional dependency, and fortran doesn't play super nice with CPython/numpy arrays, leading to memory duplication. Right now there are only two places that it's used, in parallel hop halo finder, and in the two point functions stuff. In the latter, it may be practical to replace that with the Cython tree, but for the former, we still haven't found a satisfactorily fast C/C++ kD-tree to replace the fortran one.

In summary, in the shorter term I should try switching out the kD-tree in the two point functions.

Environment

All

Status

Assignee

Stephen Skory

Reporter

Stephen Skory

Labels

None

Components

Fix versions

Affects versions

3.0

Priority

Minor
Configure