I'm working on some fairly computational intensive calculations that deal with numpy matrices and ndarrays, and from some digging around, there are about a dozen ways not to implement memoization, generally full of collisions, and issues with ndarrays being mutable objects.
Has anyone come across a fairly general memoisation decorator that can handle numpy objects?
How about this package:
http://packages.python.org/joblib/memory.html
2021 update https://joblib.readthedocs.io/en/latest/generated/joblib.Memory.html