Merge pull request #183 from PyLops/dev

Release v0.9.0
PyLops · Aug 16, 2024 · 1e9e91b · 1e9e91b
2 parents 8a4fc96 + 8efa282
commit 1e9e91b
Show file tree

Hide file tree

Showing 20 changed files with 809 additions and 78 deletions.
diff --git a/.readthedocs.yaml b/.readthedocs.yaml
@@ -10,8 +10,6 @@ build:
   os: ubuntu-20.04
   tools:
     python: "3.9"
-  apt_packages:
-    - libopenblas-dev
 
 # Build documentation in the docs/ directory with Sphinx
 sphinx:

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,3 +1,16 @@
+Changelog
+=========
+
+# 0.9.0
+
+* Added :py:class:`pyproximal.optimization.palm.iPALM` solver
+* Added :py:func:`pyproximal.optimization.palm._backtracking` method to be used when `gammaf=None` and/or `gammag=None`
+* Added :py:func:`pyproximal.utils.gradtest.gradtest_proximal` and :py:func:`pyproximal.utils.gradtest.gradtest_bilinear` methods
+* Added `tol` to :py:class:`pyproximal.optimization.primal.ProximalPoint` and
+  :py:class:`pyproximal.optimization.primal.ProximalGradient` solvers
+* Modified :py:class:`pyproximal.ProxOperator.precomposition` to allow `b` being also a vector
+
+
 # 0.8.0
 
 * Added ``pyproximal.projection.L01BallProj`` and ``pyproximal.proximal.L01Ball`` operators

diff --git a/README.md b/README.md
@@ -1,13 +1,12 @@
 ![PyProximal](https://github.com/PyLops/pyproximal/blob/dev/docs/source/_static/pyproximal_b.png)
 
 [![PyPI version](https://badge.fury.io/py/pyproximal.svg)](https://badge.fury.io/py/pyproximal)
-[![Build Status](https://travis-ci.com/PyLops/pyproximal.svg?branch=main)](https://travis-ci.com/PyLops/pyproximal)
-[![AzureDevOps Status](https://dev.azure.com/matteoravasi/PyLops/_apis/build/status%2FPyLops.pyproximal?branchName=refs%2Fpull%2F129%2Fmerge)](https://dev.azure.com/matteoravasi/PyLops/_build/latest?definitionId=10&branchName=refs%2Fpull%2F129%2Fmerge)
+[![AzureDevOps Status](https://dev.azure.com/matteoravasi/PyLops/_apis/build/status%2FPyLops.pyproximal?branchName=refs%2Fpull%2F180%2Fmerge)](https://dev.azure.com/matteoravasi/PyLops/_build/latest?definitionId=10&branchName=refs%2Fpull%2F180%2Fmerge)
 ![GithubAction Status](https://github.com/PyLops/pyproximal/workflows/PyProx/badge.svg)
 [![Documentation Status](https://readthedocs.org/projects/pyproximal/badge/?version=latest)](https://pyproximal.readthedocs.io/en/latest/?badge=latest)
 [![OS-support](https://img.shields.io/badge/OS-linux,osx-850A8B.svg)](https://github.com/PyLops/pyproximal)
 [![Slack Status](https://img.shields.io/badge/chat-slack-green.svg)](https://pylops.slack.com)
-
+[![DOI](https://joss.theoj.org/papers/10.21105/joss.06326/status.svg)](https://doi.org/10.21105/joss.06326)
 
 
 :vertical_traffic_light: :vertical_traffic_light: This library is under early development.
@@ -184,9 +183,17 @@ make docupdate
 Note that if a new example or tutorial is created (and if any change is made to a previously available example or tutorial)
 you are required to rebuild the entire documentation before your changes will be visible.
 
+## Citing
+When using PyProximal in scientific publications, please cite the following paper:
+
+- Ravasi M, Örnhag M. V., Luiken N., Leblanc O. and Uruñuela E., 2024, *<b>PyProximal - scalable convex optimization in Python</b>*,
+  Journal of Open Source Software, 9(95), 6326. doi: 10.21105/joss.06326 [(link)](https://joss.theoj.org/papers/10.21105/joss.06326)
+
+
 
 ## Contributors
 * Matteo Ravasi, mrava87
 * Nick Luiken, NickLuiken
 * Eneko Uruñuela, eurunuela
 * Marcus Valtonen Örnhag, marcusvaltonen
+* Olivier Leblanc, olivierleblanc
diff --git a/docs/source/api/index.rst b/docs/source/api/index.rst
@@ -130,6 +130,18 @@ Other operators
     BilinearOperator
     LowRankFactorizedMatrix
 
+Utility functions
+-----------------
+
+.. currentmodule:: pyproximal.utils.gradtest
+
+.. autosummary::
+   :toctree: generated/
+
+    gradtest_proximal
+    gradtest_bilinear
+
+
 
 Solvers
 -------
@@ -158,6 +170,7 @@ Primal
    :toctree: generated/
 
     PALM
+    iPALM
 
 .. currentmodule:: pyproximal.optimization.pnp
 

diff --git a/docs/source/changelog.rst b/docs/source/changelog.rst
@@ -3,6 +3,18 @@
 Changelog
 =========
 
+Version 0.9.0
+--------------
+*Released on: 16/08/2024*
+
+* Added :py:class:`pyproximal.optimization.palm.iPALM` solver
+* Added :py:func:`pyproximal.optimization.palm._backtracking` method to be used when `gammaf=None` and/or `gammag=None`
+* Added :py:func:`pyproximal.utils.gradtest.gradtest_proximal` and :py:func:`pyproximal.utils.gradtest.gradtest_bilinear` methods
+* Added `tol` to :py:class:`pyproximal.optimization.primal.ProximalPoint` and
+  :py:class:`pyproximal.optimization.primal.ProximalGradient` solvers
+* Modified :py:class:`pyproximal.ProxOperator.precomposition` to allow `b` being also a vector
+
+
 Version 0.8.0
 --------------
 *Released on: 11/03/2024*

diff --git a/docs/source/credits.rst b/docs/source/credits.rst
@@ -6,4 +6,5 @@ Contributors
 *  `Matteo Ravasi <https://github.com/mrava87>`_, mrava87
 *  `Nick Luiken <https://github.com/NickLuiken>`_, NickLuiken
 *  `Eneko Uruñuela <https://github.com/eurunuela>`_, eurunuela
-*  `Marcus Valtonen Örnhag <https://github.com/marcusvaltonen>`_, marcusvaltonen
+*  `Marcus Valtonen Örnhag <https://github.com/marcusvaltonen>`_, marcusvaltonen
+*  `Olivier Leblanc <https://github.com/olivierleblanc>`_, olivierleblanc
diff --git a/environment-dev-arm.yml b/environment-dev-arm.yml
@@ -0,0 +1,27 @@
+name: pyproximal
+channels:
+  - defaults
+  - conda-forge
+  - numba
+dependencies:
+  - python>=3.8.12
+  - numpy>=1.15.0, <2.0.0
+  - scipy>=1.8.0
+  - pylops>=2.0.0
+  - scikit-image
+  - matplotlib
+  - ipython
+  - pytest
+  - Sphinx
+  - numpydoc
+  - numba
+  - icc_rt
+  - pip:
+    - bm3d
+    - pytest-runner
+    - setuptools_scm
+    - pydata-sphinx-theme
+    - sphinx-gallery
+    - nbsphinx
+    - image
+    - sphinxemoji
diff --git a/environment-dev.yml b/environment-dev.yml
@@ -5,7 +5,7 @@ channels:
   - numba
 dependencies:
   - python>=3.8.12
-  - numpy>=1.15.0
+  - numpy>=1.15.0, <2.0.0
   - scipy>=1.8.0
   - pylops>=2.0.0
   - scikit-image
@@ -17,7 +17,8 @@ dependencies:
   - numba
   - icc_rt
   - pip:
-    - bm3d
+    - bm4d<4.2.4 # temporary fix as GLIBC_2.32 not found by readthedocs
+    - bm3d<4.0.2 # temporary fix as GLIBC_2.32 not found by readthedocs
     - pytest-runner
     - setuptools_scm
     - pydata-sphinx-theme

diff --git a/environment.yml b/environment.yml
@@ -3,6 +3,6 @@ channels:
   - defaults
 dependencies:
   - python>=3.8.12
-  - numpy>=1.15.0
+  - numpy>=1.15.0, <2.0.0
   - scipy>=1.8.0
   - pylops>=2.0.0
diff --git a/pyproject.toml b/pyproject.toml
@@ -30,7 +30,7 @@ classifiers = [
     "Topic :: Scientific/Engineering :: Mathematics",
 ]
 dependencies = [
-    "numpy >= 1.15.0",
+    "numpy >= 1.15.0, <2.0.0",
     "scipy >= 1.8.0",
     "pylops >= 2.0.0",
 ]

diff --git a/pyproximal/ProxOperator.py b/pyproximal/ProxOperator.py
@@ -73,17 +73,16 @@ def _proxdual_moreau(self, x, tau, **kwargs):
     def prox(self, x, tau, **kwargs):
         """Proximal operator applied to a vector
 
-        The  proximal operator can always be computed given its dual
+        The proximal operator can always be computed given its dual
         proximal operator using the Moreau decomposition as defined in
-        :func:`pyprox.moreau`. For this reason we can easily create a common
+        :func:`pyproximal.moreau`. For this reason we can easily create a common
         method for all proximal operators that can be evaluated provided the
         dual proximal is implemented.
 
         However, direct implementations are generally available. This can
         be done by simply implementing ``prox`` for a specific proximal
         operator, which will overwrite the general method.
 
-
         Parameters
         ----------
         x : :obj:`np.ndarray`
@@ -100,9 +99,9 @@ def proxdual(self, x, tau, **kwargs):
 
         The dual of a proximal operator can always be computed given its
         proximal operator using the Moreau decomposition as defined in
-        :func:`pyprox.moreau`. For this reason we can easily create a common
+        :func:`pyproximal.moreau`. For this reason we can easily create a common
         method for all dual proximal operators that can be evaluated provided
-        he proximal is implemented.
+        the proximal is implemented.
 
         However, since the dual of a proximal operator of a function is
         equivalent to the proximal operator of the conjugate function, smarter
@@ -163,7 +162,7 @@ def affine_addition(self, v):
         if isinstance(v, np.ndarray):
             return _SumOperator(self, v)
         else:
-            return NotImplemented
+            raise NotImplementedError('v must be of type numpy.ndarray')
 
     def postcomposition(self, sigma):
         r"""Postcomposition
@@ -191,7 +190,7 @@ def postcomposition(self, sigma):
         if isinstance(sigma, float):
             return _PostcompositionOperator(self, sigma)
         else:
-            return NotImplemented
+            raise NotImplementedError('sigma must be of type float')
 
     def precomposition(self, a, b):
         r"""Precomposition
@@ -203,8 +202,8 @@ def precomposition(self, a, b):
         ----------
         a : :obj:`float`
             Multiplicative scalar
-        b : :obj:`float`
-            Additive Scalar
+        b : :obj:`float` or obj:`np.ndarray`
+            Additive scalar (or vector)
 
         Notes
         -----
@@ -217,10 +216,12 @@ def precomposition(self, a, b):
             prox_{a^2 \tau f} (a \mathbf{x} + b) - b)
 
         """
-        if isinstance(a, float) and isinstance(b, float):
+        if isinstance(a, float) and isinstance(b, (float, np.ndarray)):
             return _PrecompositionOperator(self, a, b)
         else:
-            return NotImplemented
+            raise NotImplementedError('a must be of type float and b '
+                                      'must be of type float or '
+                                      'numpy.ndarray')
 
     def chain(self, g):
         r"""Chain
@@ -347,7 +348,7 @@ def __init__(self, f, a, b):
         #    raise ValueError('First input must be a ProxOperator')
         if not isinstance(a, float):
             raise ValueError('Second input must be a float')
-        if not isinstance(b, float):
+        if not isinstance(b, (float, np.ndarray)):
             raise ValueError('Second input must be a float')
         self.f, self.a, self.b = f, a, b
         super().__init__(None, True if f.grad else False)

diff --git a/pyproximal/optimization/__init__.py b/pyproximal/optimization/__init__.py
@@ -34,6 +34,7 @@
 
     SR3                             Sparse Relaxed Regularized algorithm
     PALM                            Proximal Alternating Linearized Minimization
+    iPALM                           Inertial Proximal Alternating Linearized Minimization
 
 Finally this subpackage contains also a solver for image segmentation based
 on a special use of the Primal-Dual algorithm: