PyTorch Wavelet Toolbox Versions Save

Differentiable fast wavelet transforms in PyTorch with GPU support.

v0.1.8

1 month ago

This release adopts the

  • adopt the sphinx_book_theme,
  • adds an introduction explaining how wavelets works,
  • improves our speed tests,
  • and improves the function docstrings, and overall code quality thanks @cthoyt .

What's Changed

Full Changelog: https://github.com/v0lta/PyTorch-Wavelet-Toolbox/compare/v0.1.7...v0.1.8

This release is available at: https://pypi.org/project/ptwt/0.1.8/ install via:

pip install ptwt==0.1.8

v0.1.7

7 months ago

What's New

  • pywt compatible axis and axes argument support along with tests.

Breaking changes:

  • The new axis or axes argument appears before arguments related to boundary wavelet support. Pywt does not support boundary wavelets, so to respect pywt argument order we have added the new arguments before anything related to boundary wavelets. This makes it easy to migrate from pywt to ptwt and back. Unfortunately, this choice breaks calls to boundary wavelet arguments which follow the old order. To migrate, call arguments related to boundary wavelets by name. This affects only matrix-fwt and wavelet-packet code that uses boundary wavelet support.

What's Changed

New Contributors

Full Changelog: https://github.com/v0lta/PyTorch-Wavelet-Toolbox/compare/v0.1.6...v0.1.7

This release is available at: https://pypi.org/project/ptwt/0.1.7/ install via:

pip install ptwt==0.1.7

v0.1.6

9 months ago

What's new:

  • support for symmetric padding for all convolution-based transformations (the default in pywt).
  • channel dimensions are now automatically processed along with batch dimensions for 1d and 2d code for easier integration into machine learning models.

What's Changed

Full Changelog: https://github.com/v0lta/PyTorch-Wavelet-Toolbox/compare/v0.1.5...v0.1.6

v0.1.5

1 year ago

What's Changed

  • added additional test cases for our wavedec3 and 1D-matrix transformation code.
  • 3d matrix transformation support. The new matrix-wavedecs3 returns a list of dictionaries like https://pywavelets.readthedocs.io/en/latest/ref/nd-dwt-and-idwt.html#multilevel-decomposition-wavedecn .
  • The separable_conv_transform module now provides support for separable padded transforms.
  • improved handling of the internal "channel"-dimension for the padded 2d-fwt case. Output tensors from wavedec2 now have the shape [batch, height, width] instead of [batch, 1, height, width].
  • Added shape checks for the transformed coefficients, for more informative error messages.
  • Added tests for just in time compilation. See test/test_jit.py for more details.

Breaking changes

  • since https://arxiv.org/pdf/2210.14874.pdf found them to work well, the faster separable transforms are now the default for all boundary wavelet computations. If you require wavedec2 and MatrixWavedec2 to produce identical coefficients in the Haar case, set separable to False.
  • to simplify the documentation _get_pad is now private.
  • To ensure our API is compatible with pywt the padded analysis transforms argument order has changed. The order is now always data, wavelet, mode, level.

Full Changelog: https://github.com/v0lta/PyTorch-Wavelet-Toolbox/compare/v0.1.4...v0.1.5

v0.1.4.post1

1 year ago

What's new:

This release is available at: https://pypi.org/project/ptwt/0.1.4/ install via:

pip install ptwt==0.1.4

v0.1.3

1 year ago

What's new:

  • support for inverse packet computations.
  • extended the documentation for the wavelet packet code.
  • renamed the packets max_level argument to maxlevel for pywt compatibility (breaking change).

This release is available at: https://pypi.org/project/ptwt/0.1.3/ install via:

pip install ptwt==0.1.3

v0.1.2

2 years ago

What's new:

  • three-dimensional padded analysis and synthesis transform operations are now supported.
  • the code has been refactored for improved readability.

This release is available at: https://pypi.org/project/ptwt/0.1.2/ install via:

pip install ptwt==0.1.2

v0.1.1

2 years ago

What's new:

  • Features:

    • wavedec and wavedec2 now additionally support constant and periodic padding.
    • Further expanded the test coverage.
    • Added Python 3.10.2 support.
    • Moved the CI-Pipeline from tox to nox ( https://nox.thea.codes/en/stable/index.html ).

  • Bugfixes:

    • Fixed a bug in the 2d-boundary-wavelet code. Using non-separated two-dimensional boundary wavelets on non-square images now works as expected. We thank @raoulheese for reporting this one!

  • Breaking changes:

    • Removed the redundant level argument from the fwt_pad2 function.

This release is available at: https://pypi.org/project/ptwt/0.1.1/ install via:

pip install ptwt==0.1.1

v0.1.0

2 years ago

For major new release we have:

  • changed the names of the matrix wavelet objects from 2d to 2 for naming consistency within ptwt and with pywt (breaking change).
  • added support for 1d continuous wavelet transforms in PyTorch.
  • Extended test coverage.

This release is available at: https://pypi.org/project/ptwt/0.1.0/ install via:

pip install ptwt==0.1.0