Merge branch 'master' into revisit_overwrite_x

Author: vtavana

.github/workflows/conda-package.yml

@@ -55,7 +55,7 @@ jobs:
 
       - name: Build conda package
         run: |
-          CHANNELS="-c conda-forge -c https://software.repos.intel.com/python/conda --override-channels"
+          CHANNELS="-c https://software.repos.intel.com/python/conda -c conda-forge --override-channels"
           VERSIONS="--python ${{ matrix.python }}"
           TEST="--no-test"
 
@@ -101,8 +101,17 @@ jobs:
           mkdir -p $GITHUB_WORKSPACE/channel/linux-64
           mv ${PACKAGE_NAME}-*.conda $GITHUB_WORKSPACE/channel/linux-64
           conda index $GITHUB_WORKSPACE/channel
-          # Test channel
-          conda search $PACKAGE_NAME -c $GITHUB_WORKSPACE/channel --override-channels
+
+      - name: Test conda channel
+        run: |
+          conda search $PACKAGE_NAME -c $GITHUB_WORKSPACE/channel --override-channels --info --json > $GITHUB_WORKSPACE/ver.json
+          cat $GITHUB_WORKSPACE/ver.json
+
+      - name: Get package version
+        run: |
+          export PACKAGE_VERSION=$(python -c "${{ env.VER_SCRIPT1 }} ${{ env.VER_SCRIPT2 }}")
+          echo PACKAGE_VERSION=${PACKAGE_VERSION}
+          echo "PACKAGE_VERSION=$PACKAGE_VERSION" >> $GITHUB_ENV
 
       - name: Collect dependencies
         run: |
@@ -131,8 +140,7 @@ jobs:
       - name: Install mkl_fft
         run: |
           CHANNELS="-c $GITHUB_WORKSPACE/channel ${{ env.CHANNELS }}"
-          conda create -n ${{ env.TEST_ENV_NAME }} python=${{ matrix.python }} "scipy>=1.10" $CHANNELS
-          conda install -n ${{ env.TEST_ENV_NAME }} $PACKAGE_NAME pytest $CHANNELS
+          conda create -n ${{ env.TEST_ENV_NAME }} $PACKAGE_NAME=${{ env.PACKAGE_VERSION }} python=${{ matrix.python }} pytest "scipy>=1.10" $CHANNELS
           # Test installed packages
           conda list -n ${{ env.TEST_ENV_NAME }}
 

.github/workflows/openssf-scorecard.yml

@@ -39,7 +39,7 @@ jobs:
           persist-credentials: false
 
       - name: "Run analysis"
-        uses: ossf/scorecard-action@f49aabe0b5af0936a0987cfb85d86b75731b0186 # v2.4.1
+        uses: ossf/scorecard-action@05b42c624433fc40578a4040d5cf5e36ddca8cde # v2.4.2
         with:
           results_file: results.sarif
           results_format: sarif

CHANGELOG.md

@@ -4,7 +4,7 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## [dev] (MM/DD/YY)
+## [2.0.0] (05/DD/2025)
 
 ### Added
 * Added Hermitian FFT functions to SciPy interface `mkl_fft.interfaces.scipy_fft`: `hfft`, `ihfft`, `hfftn`, `ihfftn`, `hfft2`, and `ihfft2` [gh-161](https://github.com/IntelPython/mkl_fft/pull/161)

README.md

@@ -4,8 +4,17 @@
 [![Conda package with conda-forge channel only](https://github.com/IntelPython/mkl_fft/actions/workflows/conda-package-cf.yml/badge.svg)](https://github.com/IntelPython/mkl_fft/actions/workflows/conda-package-cf.yml)
 [![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/IntelPython/mkl_fft/badge)](https://securityscorecards.dev/viewer/?uri=github.com/IntelPython/mkl_fft)
 
+# Introduction
 `mkl_fft` started as a part of Intel® Distribution for Python* optimizations to NumPy, and is now being released
-as a stand-alone package. It can be installed into conda environment from Intel's channel using:
+as a stand-alone package. It offers a thin layered interface for the Intel® oneAPI Math Kernel Library (OneMKL) FFT functionality that allows efficient access to native FFT optimizations from a range of NumPy and SciPy functions. As a result, its performance is close to the performance of native C/Intel® OneMKL. The optimizations are provided for real and complex data types in both single and double precisions for in-place and out-of-place modes of operation. For analyzing the performance use [FFT benchmarks](https://github.com/intelpython/fft_benchmark).
+
+Thanks to Intel® OneMKL’s flexibility in its supports for arbitrarily strided input and output arrays both one-dimensional and multi-dimensional Fast Fourier Transforms along distinct axes can be performed directly, without the need to copy the input into a contiguous array first. Furthermore, input strides can be arbitrary, including negative or zero, as long as strides remain an integer multiple of array’s item size, otherwise a copy will be made.
+
+More details can be found in ["Accelerating Scientific Python with Intel Optimizations"](https://proceedings.scipy.org/articles/shinma-7f4c6e7-00f) from Proceedings of the 16th Python in Science Conference (SciPy 2017).
+
+---
+# Installation
+`mkl_fft` can be installed into conda environment from Intel's channel using:
 
 ```
    conda install -c https://software.repos.intel.com/python/conda mkl_fft
@@ -34,22 +43,12 @@ If command above installs NumPy package from the PyPI, please use following comm
 Where `<numpy_version>` should be the latest version from https://software.repos.intel.com/python/conda/
 
 ---
+# How to use?
+## `mkl_fft.interfaces` module
+The recommended way to use `mkl_fft` package is through `mkl_fft.interfaces` module. These interfaces act as drop-in replacements for equivalent functions in NumPy and SciPy. Learn more about these interfaces [here](https://github.com/IntelPython/mkl_fft/blob/master/mkl_fft/interfaces/README.md).
 
-Since MKL FFT supports performing discrete Fourier transforms over non-contiguously laid out arrays, OneMKL can be directly
-used on any well-behaved floating point array with no internal overlaps for both in-place and not in-place transforms of
-arrays in single and double floating point precision.
-
-This eliminates the need to copy input array contiguously into an intermediate buffer.
-
-`mkl_fft` directly supports N-dimensional Fourier transforms.
-
-More details can be found in [SciPy 2017 conference proceedings](https://github.com/scipy-conference/scipy_proceedings/tree/2017/papers/oleksandr_pavlyk).
-
----
-
-The `mkl_fft` package offers interfaces that act as drop-in replacements for equivalent functions in NumPy and SciPy. Learn more about these interfaces [here](https://github.com/IntelPython/mkl_fft/blob/master/mkl_fft/interfaces/README.md).
-
-While using these interfaces is the easiest way to leverage `mk_fft`, one can also use `mkl_fft` directly with the following FFT functions:
+## `mkl_fft` package
+While using the interfaces module is the recommended way to leverage `mk_fft`, one can also use `mkl_fft` directly with the following FFT functions:
 
 ### complex-to-complex (c2c) transforms:
 
@@ -84,6 +83,7 @@ numpy.allclose(mkl_res, np_res)
 ```
 
 ---
+# Building from source
 
 To build `mkl_fft` from sources on Linux with Intel® OneMKL:
   - create a virtual environment: `python3 -m venv fft_env`

conda-recipe-cf/meta.yaml

@@ -1,4 +1,4 @@
-{% set version = "1.3.14" %}
+{% set version = "2.0.0" %}
 {% set buildnumber = 0 %}
 
 package:

conda-recipe/meta.yaml

@@ -1,4 +1,4 @@
-{% set version = "1.3.14" %}
+{% set version = "2.0.0" %}
 {% set buildnumber = 0 %}
 
 package:
@@ -20,6 +20,7 @@ requirements:
       - python
       - setuptools >=77
       - mkl-devel
+      - mkl-devel 2024.2.*  # [py==39 or py==310]
       - cython
       - numpy-base
     run:

mkl_fft/_pydfti.pyx

@@ -203,7 +203,7 @@ cdef cnp.ndarray _pad_array(
     b_shape[axis] = n
 
     # allocating temporary buffer
-    x_arr_is_fortran = cnp.PyArray_CHKFLAGS(x_arr, cnp.NPY_F_CONTIGUOUS)
+    x_arr_is_fortran = cnp.PyArray_CHKFLAGS(x_arr, cnp.NPY_ARRAY_F_CONTIGUOUS)
     b_arr = <cnp.ndarray> cnp.PyArray_EMPTY(
         b_ndim, b_shape, <cnp.NPY_TYPES> b_type, x_arr_is_fortran
     )  # 0 for C-contiguous
@@ -239,9 +239,12 @@ cdef cnp.ndarray _process_arguments(
 
     # convert x to ndarray, ensure that strides are multiples of itemsize
     x_arr = PyArray_CheckFromAny(
-          x, NULL, 0, 0,
-          cnp.NPY_ELEMENTSTRIDES | cnp.NPY_ENSUREARRAY | cnp.NPY_NOTSWAPPED,
-          NULL)
+        x, NULL, 0, 0,
+        cnp.NPY_ARRAY_ELEMENTSTRIDES |
+        cnp.NPY_ARRAY_ENSUREARRAY |
+        cnp.NPY_ARRAY_NOTSWAPPED,
+        NULL
+    )
 
     if (<void *> x_arr) is NULL:
         raise ValueError("An input argument x is not an array-like object")
@@ -309,7 +312,7 @@ cdef cnp.ndarray _allocate_result(
         f_shape[axis_] = n_
 
     # allocating output buffer
-    x_arr_is_fortran = cnp.PyArray_CHKFLAGS(x_arr, cnp.NPY_F_CONTIGUOUS)
+    x_arr_is_fortran = cnp.PyArray_CHKFLAGS(x_arr, cnp.NPY_ARRAY_F_CONTIGUOUS)
     f_arr = <cnp.ndarray> cnp.PyArray_EMPTY(
         f_ndim, f_shape, <cnp.NPY_TYPES> f_type, x_arr_is_fortran
     )  # 0 for C-contiguous
@@ -402,7 +405,9 @@ def _c2c_fft1d_impl(x, n=None, axis=-1, direction=+1, double fsc=1.0, out=None):
         # so we cast to complex double and operate in place
         try:
             x_arr = <cnp.ndarray> cnp.PyArray_FROM_OTF(
-                x_arr, cnp.NPY_CDOUBLE, cnp.NPY_BEHAVED | cnp.NPY_ENSURECOPY)
+                x_arr, cnp.NPY_CDOUBLE,
+                cnp.NPY_ARRAY_BEHAVED | cnp.NPY_ARRAY_ENSURECOPY
+            )
         except:
             raise ValueError(
                 "First argument must be a complex "
@@ -582,9 +587,9 @@ def _r2c_fft1d_impl(
     else:
         # we must cast the input to doubles and allocate the output,
         try:
-            requirement = cnp.NPY_BEHAVED | cnp.NPY_ENSURECOPY
+            requirement = cnp.NPY_ARRAY_BEHAVED | cnp.NPY_ARRAY_ENSURECOPY
             if x_type is cnp.NPY_LONGDOUBLE:
-                requirement = requirement | cnp.NPY_FORCECAST
+                requirement = requirement | cnp.NPY_ARRAY_FORCECAST
             x_arr = <cnp.ndarray> cnp.PyArray_FROM_OTF(
                 x_arr, cnp.NPY_DOUBLE, requirement)
             x_type = cnp.PyArray_TYPE(x_arr)
@@ -684,11 +689,11 @@ def _c2r_fft1d_impl(
         # so we cast to complex double and operate in place
         if x_type is cnp.NPY_FLOAT:
             x_arr = <cnp.ndarray> cnp.PyArray_FROM_OTF(
-                x_arr, cnp.NPY_CFLOAT, cnp.NPY_BEHAVED
+                x_arr, cnp.NPY_CFLOAT, cnp.NPY_ARRAY_BEHAVED
             )
         else:
             x_arr = <cnp.ndarray> cnp.PyArray_FROM_OTF(
-                x_arr, cnp.NPY_CDOUBLE, cnp.NPY_BEHAVED
+                x_arr, cnp.NPY_CDOUBLE, cnp.NPY_ARRAY_BEHAVED
             )
         x_type = cnp.PyArray_TYPE(x_arr)
         in_place = 1
@@ -763,9 +768,12 @@ def _direct_fftnd(
 
     # convert x to ndarray, ensure that strides are multiples of itemsize
     x_arr = PyArray_CheckFromAny(
-          x, NULL, 0, 0,
-          cnp.NPY_ELEMENTSTRIDES | cnp.NPY_ENSUREARRAY | cnp.NPY_NOTSWAPPED,
-          NULL)
+        x, NULL, 0, 0,
+        cnp.NPY_ARRAY_ELEMENTSTRIDES |
+        cnp.NPY_ARRAY_ENSUREARRAY |
+        cnp.NPY_ARRAY_NOTSWAPPED,
+        NULL
+    )
 
     if <void *> x_arr is NULL:
         raise ValueError("An input argument x is not an array-like object")
@@ -783,7 +791,9 @@ def _direct_fftnd(
         pass
     else:
         x_arr = <cnp.ndarray> cnp.PyArray_FROM_OTF(
-            x_arr, cnp.NPY_CDOUBLE, cnp.NPY_BEHAVED | cnp.NPY_ENSURECOPY)
+            x_arr, cnp.NPY_CDOUBLE,
+            cnp.NPY_ARRAY_BEHAVED | cnp.NPY_ARRAY_ENSURECOPY
+        )
         x_type = cnp.PyArray_TYPE(x_arr)
         assert x_type == cnp.NPY_CDOUBLE
         in_place = 1

mkl_fft/_version.py

@@ -1 +1 @@
-__version__ = "1.3.14"
+__version__ = "2.0.0"

mkl_fft/interfaces/numpy_fft.py

@@ -24,9 +24,6 @@
 # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-# Added for completing the namespaces
-from numpy.fft import fftfreq, fftshift, ifftshift, rfftfreq
-
 # pylint: disable=no-name-in-module
 from ._numpy_fft import (
     fft,
@@ -60,8 +57,11 @@
     "irfftn",
     "hfft",
     "ihfft",
-    "fftshift",
-    "ifftshift",
-    "fftfreq",
-    "rfftfreq",
 ]
+
+# It is important to put the following import here to avoid circular imports
+# when patching numpy with mkl_fft
+# Added for completing the namespaces
+from numpy.fft import fftfreq, fftshift, ifftshift, rfftfreq
+
+__all__ += ["fftshift", "ifftshift", "fftfreq", "rfftfreq"]