Generalized move_func_strides to arbitrary dimensions #81

Merged
merged 1 commit into from Feb 18, 2014

Conversation

Projects
None yet
2 participants
@shoyer
Contributor

shoyer commented Feb 18, 2014

For context: strides are generally the fastest way to calculate moving window
functions with pure numpy (if sufficient memory is available). Since I like
using bottleneck for fast rolling window functions, I thought it would make
sense to be able to use bottleneck's implementation of striding moving windows
in higher dimensions.

Because there was no unit test coverage for this function, I also adapted
move_test.py to check against the 'strides' method for slow calculations as
well as the 'loop' method.

Generalized move_func_strides to arbitrary dimensions
For context: strides are generally the fastest way to calculate moving window
functions with pure numpy (if sufficient memory is available). Since I like
using bottleneck for fast rolling window functions, I thought it would make
sense to be able to use bottleneck's implementation of striding moving windows
in higher dimensions.

Because there was no unit test coverage for this function, I also adapted
move_test.py to check against the 'strides' method for slow calculations as
well as the 'loop' method.
@kwgoodman

This comment has been minimized.

Show comment Hide comment
@kwgoodman

kwgoodman Feb 18, 2014

Owner

Very nice!

Owner

kwgoodman commented Feb 18, 2014

Very nice!

kwgoodman added a commit that referenced this pull request Feb 18, 2014

Merge pull request #81 from shoyer/better_move_func_strides
Generalized move_func_strides to arbitrary dimensions

@kwgoodman kwgoodman merged commit a608b86 into kwgoodman:master Feb 18, 2014

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment