File python-autoray.changes of Package python-autoray
-------------------------------------------------------------------
Sun Mar 22 17:27:13 UTC 2026 - Dirk Müller <dmueller@suse.com>
- update to 0.8.10:
* `tensorflow`: support for `cholesky` with `upper` arg,
batched svd, and `scipy.linalg.solve_triangular`
* **Full Changelog**:
https://github.com/jcmgray/autoray/compare/v0.8.9...v0.8.10
- update to 0.8.9:
* add `swapaxes` function for tensorflow
- update to 0.8.8:
* **Enhancements:**
* `autoray.lazy`: add support for `moveaxis` and `swapaxes`
* **Full Changelog**:
https://github.com/jcmgray/autoray/compare/v0.8.7...v0.8.8
- update to 0.8.7:
* **Enhancements**
* Torch improvements:
* Added torch implementation for `trace` with axes specified
plus tests.
* Allowed torch `default_rng` seed to be None.
* Added axis kwarg support for torch `count_nonzero`.
* Added decorator support for registration APIs:
* `register_function` can now be used as a decorator.
* `register_custom_wrapper` can now be used as a decorator.
* Refactored internals to use the new decorator style for
function registration.
- update to 0.8.6:
* add stop_gradient
* Add torch support for scipy.linalg.solve_triangular.
* Add tensorflow support for random.default_rng.
* Add allclose support for lazy arrays.
* Add Python 3.14 support.
* Fix and sort internal backend registration functions.
- update to 0.8.4:
* **Enhancements**
* add `autoray.is_scalar` function
* `lazy.einsum`: for numpy backend set `optimize=True` by
default
* `lazy.linalg.norm`: add kwargs support
* `torch.random.default_rng`: add choice support
* move ci to pixi
* **Full Changelog**:
https://github.com/jcmgray/autoray/compare/v0.8.3...v0.8.4
- update to 0.8.3:
* **Enhancements**
* register `array` and `asarray` as creation routines, so that
they pick up device by default from `like`
* `LazyArray.show`: show more information about call signature
by default
* lazy creation routines, support single int as 1D shape
specifier
* remove custom `torch.count_nonzero` which `torch` now
natively implements
* bump min python version to 3.10
* `torch.take` implementation: support vmapping over both
scalar and sequence indices
* `get_namespace`: support `scipy` as a custom submodule.
* custom `autograd.numpy.take` to support grad
(https://github.com/HIPS/autograd/issues/743).
* **Full Changelog**:
https://github.com/jcmgray/autoray/compare/v0.8.2...v0.8.3
- update to 0.8.2:
* **Ehancements**
* Speedup creation of `get_namespace`.
* Reduce namespace attribute retrieval overhead to essentially
zero
* **Full Changelog**:
https://github.com/jcmgray/autoray/compare/v0.8.1...v0.8.2
- update to 0.8.1:
* **Bug fixes**
* enable jax batched qr including for flat, grad inputs
- update to 0.8.0:
* **Breaking changes**
* `LazyArray.__iter__`: now iterates over slices of array
rather than the computational graph nodes
* **Enhancements**
* add `xp = autoray.get_namespace(like)` as an alternative api
* `LazyArray`: add support for python array api via
`.__array_namespace__()`
* `LazyArray`: support broadcasted linear algebra
* alias `("torch" "equal")` to `torch.eq`
* `lazy`: support caching of more kwargs
* `lazy`: add `take_along_axis`
* `lazy`: add `equal`
* add `"random.default_rng"` implementation for `jax` and
`torch` to support random pure functions
* **Bug fixes**
* python compiler supports calls with kwargs
* `jax`: re wrap fat QR to support gradient
* `pytensor`: qr fix mode to 'economic'
- update to 0.7.2:
* support axis kwargs for torch diagonal
* no longer wrap jax qr for 'fat' matrices
- update to 0.7.1:
* optimize lazy computational graph traversal
* complexity tracing: support pytrees and add a couple of
missing functions
* make `get_dtype_name` default impl more robust
* add `paddle` support
* `torch`: add flip, copy, `keepdims` translation for
reductions.
* `tensorflow`, fix cast location
-------------------------------------------------------------------
Tue Nov 19 12:32:33 UTC 2024 - Dirk Müller <dmueller@suse.com>
- update to 0.7.0:
* prefer using `tensorflow.experimental.numpy` api over custom
translations
* add many more unit tests
* move to `pyproject.toml` file
* move to uv for CI setup
* set python min version to 3.9
-------------------------------------------------------------------
Fri Aug 30 12:46:56 UTC 2024 - Ben Greiner <code@bnavigator.de>
- Unpin numpy 2
- Clean Build Requirements
-------------------------------------------------------------------
Tue Jul 9 05:22:46 UTC 2024 - Steve Kowalik <steven.kowalik@suse.com>
- Restrict numpy to < 2.
-------------------------------------------------------------------
Thu Jun 6 08:45:43 UTC 2024 - Dirk Müller <dmueller@suse.com>
- update to 0.6.12:
* add `torch.indices` and `tensorflow.indices`
- update to 0.6.11:
* Fix creation functions with builtin `like` kwarg
* Allow additional keyword arguments in `tensorflow_diag`
- update to 0.6.10:
* The following functions now inherit default `dtype` and
possibly `device` (when using torch) when the `like` kwarg is
an explicit array:
* `"empty"`
* `"eye"`
* `"full"`
* `"identity"`
* `"ones"`
* `"zeros"`
* fix NumpyMimic special attribute access
* fix `"diag"` for tensorflow.
-------------------------------------------------------------------
Sat Apr 6 18:52:03 UTC 2024 - Dirk Müller <dmueller@suse.com>
- update to 0.6.9:
* `autojit`: fix jax when kwargs are used
* `autojit`: simplify torch and python compiler
* torch: alias min/max to amin/amax
-------------------------------------------------------------------
Tue Jan 30 12:16:25 UTC 2024 - Dirk Müller <dmueller@suse.com>
- update to 0.6.8:
* Alias jax.scipy
-------------------------------------------------------------------
Fri Nov 10 13:34:32 UTC 2023 - Dirk Müller <dmueller@suse.com>
- update to 0.6.7:
* `lazy.einsum`: allow `cotengra` or `opt_einsum` for advanced
parsing, fall back to basic when neither present
* add `torch.expand_dims`.
* **Full Changelog**:
https://github.com/jcmgray/autoray/compare/v0.6.6...v0.6.7
-------------------------------------------------------------------
Mon Sep 25 02:48:25 UTC 2023 - jun wang <junguo.wang@suse.com>
- update to 0.6.6
* autoray.lazy: allow more functions to work on pytrees
* Fix importlib import error and update tensorflow CI
* add multi-dispatch for a few more relevant functions
* change plot_history_functions default style to scatter
* ensure composed function names are kept
* add autoray.size
* added docs
* autoray.lazy.LazyArray support all fancy indexing
* add experimental complexity cost tracing
* fix: matmul shape for lazy by @yangguohao in #16
-------------------------------------------------------------------
Sun Apr 16 17:00:31 UTC 2023 - Dirk Müller <dmueller@suse.com>
- update to 0.6.3:
* `autoray.lazy.compute`: allow computing multiple outputs
simultaneously
* `autoray.lazy.Function` allow pickling and viewing of
uncompiled source
* make sure `shape` and `ndim` work for builtins similarly to
`numpy.{shape,ndim}`
* Add: `autoray.lazy.where` function for `LazyArray`
* Add: `autoray.lazy.take` function for `LazyArray`
* Add`LazyArray.plot_history_stats` pie charts
* Add `autoray.shape` and `autoray.ndim` as preferred shape
functions
* Add basic support for `aesara`
* `LazyArray`: fix negative axis reductions
* fix fancy indexing of `LazyArray` objects
-------------------------------------------------------------------
Thu Jan 26 09:29:06 UTC 2023 - Ben Greiner <code@bnavigator.de>
- Update to 0.6.0
* autoray.lazy: Much better tools for inspecting the
computational graph:
- LazyArray.show()
- LazyArray.plot_circuit()
- LazyArray.plot_history_functions() and variants
- LazyArray.history_fn_frequencies
* Global submodule aliases:
- enables e.g. dispatching do("scipy.linalg.lu", x:
numpy.ndarray) and cupy.scipy to cupyx.scipy
* Fix for complex and builtins and newer numpy (#11).
-------------------------------------------------------------------
Sat Jan 14 11:48:05 UTC 2023 - Ben Greiner <code@bnavigator.de>
- Update to 0.5.3
* add lazy.diag
* update infrastructure, including moving from versioneer to
setuptools_scm
- Release 0.5.1
* allow Composed.register to be used as a decorator
- Release 0.3.2
* tweaks to compiler.py
- No relevant release or tag notes for other releases since 0.2.5
- PEP517 build
-------------------------------------------------------------------
Mon Jan 24 16:30:37 UTC 2022 - Benjamin Greiner <code@bnavigator.de>
- SciPy and dask are optional
- Don't test dask with python310: not supported yet
-------------------------------------------------------------------
Fri Feb 19 06:18:20 UTC 2021 - andy great <andythe_great@pm.me>
- Update to version 0.2.5.
* No changelog given.
- Skip python36 because numpy no longer support it.
-------------------------------------------------------------------
Wed Aug 19 10:39:15 UTC 2020 - Tomáš Chvátal <tchvatal@suse.com>
- Fix source url to fetch from github
-------------------------------------------------------------------
Wed Aug 19 08:59:05 UTC 2020 - Tomáš Chvátal <tchvatal@suse.com>
- Remove not stated dependencies
- Do not run coverage testing
-------------------------------------------------------------------
Tue Aug 18 09:48:36 UTC 2020 - andy great <andythe_great@pm.me>
- Initial package release.
