Changes:
- Fix unnecessary generator
- Iterate dictionary directly instead of calling .keys()
- Remove global statement at the module level
- Use list() instead of a list comprehension
- Use with statement to open the file
- Merge isinstance calls
* WIP: real valued fft functions
Note: The transpose rule is not correct yet (hence the failing tests).
* Fix transpose rules for rfft and irfft
* Typo fix
* fix test failures in x64 mode
* Add 1d/2d real fft functions, plus docs
* first commit with placeholders for tests
* added tests for the following:
1 - inverse
2 - dtypes
3 - size
4 - axis
added error tests for the following:
1 - multiple axes provided instead of single axis
2 - axis out of bounds
* removed placeholders
added functions to .rst file
* Support transforms along arbitrary axes with jax.numpy.fft
Fixes GH-1878
The logic that attempted to check for transformations along non-innermost axes
was broken.
Rather than fixing it, this PR adds support for these transformations by
transposing and untransposing arrays. This adds some overhead over the LAX
implementation, but it suspect it is minimal in most cases and it should be
worthwhile for the sake of completeness.
* Fixes per review
* Move internal type-related functions into a new (internal) jax.types module.
Avoid calling onp type functions in lieu of the wrappers in jax.types. Currently these do the same thing, but future changes will make the behavior of the jax type functions diverge from the classic NumPy versions in some cases.
Move xla_bridge.canonicalize_dtype into jax.types, since it fits there more naturally.
* Rename jax.types to jax.dtypes.
* s/types/dtypes/ in tests.
This change creates a new fft primitive in lax, and uses it to implement numpy's np.fft.fftn function.
Not-yet-implemented functionality:
- vmap
- 's' argument of fftn
- other numpy np.fft functions
Resolves#505.
