* Make a global `qnm.modes_cache` object so the user doesn't have to * Run a startup function to ensure common jitted functions are compiled at load time (they are cached thereafter)
0.2.7
- A numba improvement for a big speedup - `KerrSpinSeq.__call__` can now store computations in the sequence - Improve input validation, docs, and add one `__repr__`
0.2.5
Mainly a docs improvement, focusing on documenting spherical-spheroidal decomposition