Merge branch 'GUDHI:master' into matrix_vine_opt

.circleci/config.yml

@@ -1,12 +1,11 @@
 version: 2.0
 jobs:
-
-### With all third parties
+  ### With all third parties
 
   examples:
     docker:
       # cf. https://github.com/GUDHI/gudhi-deploy/blob/main/Dockerfile_for_circleci_image
-      - image: gudhi/ci_for_gudhi:2024.06.02
+      - image: gudhi/ci_for_gudhi:2024.09.01
     steps:
       - checkout
       - run:
@@ -25,7 +24,7 @@ jobs:
 
   tests:
     docker:
-      - image: gudhi/ci_for_gudhi:2024.06.02
+      - image: gudhi/ci_for_gudhi:2024.09.01
     steps:
       - checkout
       - run:
@@ -44,7 +43,7 @@ jobs:
 
   debug_tests:
     docker:
-      - image: gudhi/ci_for_gudhi:2024.06.02
+      - image: gudhi/ci_for_gudhi:2024.09.01
     steps:
       - checkout
       - run:
@@ -63,7 +62,7 @@ jobs:
 
   utils:
     docker:
-      - image: gudhi/ci_for_gudhi:2024.06.02
+      - image: gudhi/ci_for_gudhi:2024.09.01
     steps:
       - checkout
       - run:
@@ -82,7 +81,7 @@ jobs:
 
   python:
     docker:
-      - image: gudhi/ci_for_gudhi:2024.06.02
+      - image: gudhi/ci_for_gudhi:2024.09.01
     resource_class: large # Delaunay complex requires about 5 Gb of RAM to compile
     steps:
       - checkout
@@ -94,7 +93,6 @@ jobs:
       - run:
           name: Build and test python module. Generates and tests the python documentation
           command: |
-            pip3 install pydata-sphinx-theme
             mkdir build
             cd build
             cmake -DWITH_GUDHI_THIRD_PARTY=OFF -DUSER_VERSION_DIR=version ..
@@ -167,8 +165,7 @@ jobs:
             latexmk -pdf -interaction=nonstopmode test_biblio.tex
             latexmk -pdf -interaction=nonstopmode test_gudhi_citation.tex
 
-
-### With all third parties, except CGAL and Eigen
+  ### With all third parties, except CGAL and Eigen
 
   examples_without_cgal_eigen:
     docker:
@@ -248,8 +245,7 @@ jobs:
             python3 setup.py build_ext --inplace
             ctest --output-on-failure
 
-
-### With all third parties, except CGAL
+  ### With all third parties, except CGAL
 
   examples_without_cgal:
     docker:

src/Alpha_complex/test/Incremental_delaunay_test.cpp

@@ -12,6 +12,7 @@
 #define BOOST_TEST_MODULE "incremental_delaunay"
 #include <boost/test/unit_test.hpp>
 #include <boost/range/adaptor/sliced.hpp>
+#include <boost/mpl/list.hpp>
 
 #include <CGAL/Epick_d.h>
 #include <CGAL/Epeck_d.h>

src/Subsampling/include/gudhi/choose_n_farthest_points.h

@@ -168,7 +168,7 @@ using radius_priority_ds =
                           boost::heap::mutable_<true>, boost::heap::constant_time_size<false>>;
 template<class FT>
 struct Landmark_info {
-  std::size_t far; FT radius;
+  std::size_t farthest; FT radius;
   // The points that are closer to this landmark than to other landmarks
   std::vector<std::pair<std::size_t, FT>> voronoi;
   // For a landmark A, the list of landmarks B such that picking a Voronoi
@@ -251,7 +251,7 @@ void choose_n_farthest_points_metric(Distance dist_,
       }
     }
     landmarks[i].radius = r;
-    landmarks[i].far = jmax;
+    landmarks[i].farthest = jmax;
   };
   auto update_radius = [&](std::size_t i)
   {
@@ -280,7 +280,7 @@ void choose_n_farthest_points_metric(Distance dist_,
   for (std::size_t current_number_of_landmarks = 1; current_number_of_landmarks != final_size; current_number_of_landmarks++) {
     std::size_t l_parent = radius_priority.top();
     auto& parent_info = landmarks[l_parent];
-    std::size_t l = parent_info.far;
+    std::size_t l = parent_info.farthest;
     FT radius = parent_info.radius;
     auto& info = landmarks[l];
     *output_it++ = input_pts[l];
@@ -309,8 +309,8 @@ void choose_n_farthest_points_metric(Distance dist_,
       if (it != ngb_info.voronoi.end()) { // modified, always true for ngb==l_parent
         ngb_info.voronoi.erase(it, ngb_info.voronoi.end());
         modified_neighbors.push_back(ngb);
-        // We only need to recompute the radius if far was removed, which we can test here with
-        //   if (dist(l, ngb_info.far) < ngb_info.radius)
+        // We only need to recompute the radius if farthest was removed, which we can test here with
+        //   if (dist(l, ngb_info.farthest) < ngb_info.radius)
         // to avoid a costly test for each w in the loop above, but it does not seem to help.
         update_radius(ngb);
         // if (ngb_info.voronoi.empty()) radius_priority.erase(ngb_info.position_in_queue);

src/cmake/modules/GUDHI_third_party_libraries.cmake

@@ -185,6 +185,7 @@ if (WITH_GUDHI_PYTHON)
     find_python_module("sphinx_paramlinks")
     find_python_module("pydata_sphinx_theme")
     find_python_module("sphinxcontrib.bibtex")
+    find_python_module("sphinx-autodoc-typehints")
     find_python_module("networkx")
     # Specific case for PyKeops on Windows where import fails because it uses fcntl (not available on Windows)
     if (NOT WIN32)

src/python/CMakeLists.txt

@@ -178,6 +178,11 @@ if(SPHINXCONTRIB.BIBTEX_FOUND)
 else()
   disable_python_documentation("sphinxcontrib-bibtex")
 endif()
+if(SPHINX-AUTODOC-TYPEHINTS_FOUND)
+  add_gudhi_debug_info("sphinx-autodoc-typehints version ${SPHINX-AUTODOC-TYPEHINTS_VERSION}")
+else()
+  disable_python_documentation("sphinx-autodoc-typehints")
+endif()
 if(NETWORKX_FOUND)
   add_gudhi_debug_info("NetworkX version ${NETWORKX_VERSION}")
 endif()

src/python/doc/conf.py

@@ -8,18 +8,24 @@
 sys.path.insert(0, os.path.abspath("."))
 
 extensions = [
+    "sphinx.ext.napoleon",
+    # let napoleon be the first extension, cf. https://github.com/tox-dev/sphinx-autodoc-typehints/issues/15
     "matplotlib.sphinxext.plot_directive",
     "sphinx.ext.autodoc",
+    "sphinx_autodoc_typehints",
     "sphinx.ext.doctest",
     "sphinx.ext.mathjax",
     "sphinx.ext.ifconfig",
     "sphinx.ext.viewcode",
-    "sphinx.ext.napoleon",
     "sphinxcontrib.bibtex",
     "sphinx_paramlinks",
 ]
 
 autodoc_class_signature = "separated"
+# cf. https://github.com/tox-dev/sphinx-autodoc-typehints#options
+autodoc_typehints = "none"
+napoleon_use_rtype = True
+always_use_bars_union = True
 
 bibtex_bibfiles = ["../../biblio/bibliography.bib"]
 

src/python/doc/installation.rst

@@ -183,6 +183,7 @@ A complete configuration would be :
     Sphinx-paramlinks version 0.6.0
     pydata_sphinx_theme version 0.15.2
     sphinxcontrib-bibtex version 2.6.2
+    sphinx-autodoc-typehints version 2.4.4
     NetworkX version 3.3
     Eigen3 version 3.4.0
     Boost version 1.84.0
@@ -204,6 +205,7 @@ Documentation
 
 To build the documentation, `sphinx-doc <http://www.sphinx-doc.org>`_,
 `sphinxcontrib-bibtex <https://sphinxcontrib-bibtex.readthedocs.io>`_,
+`sphinx-autodoc-typehints <https://github.com/tox-dev/sphinx-autodoc-typehints>`_,
 `sphinxcontrib-paramlinks <https://github.com/sqlalchemyorg/sphinx-paramlinks>`_ and
 `pydata-sphinx-theme <https://github.com/pydata/pydata-sphinx-theme>`_ :math:`\geq` 0.8.0 are
 required. As the documentation is auto-tested, `CGAL`_, `Eigen`_,

src/python/gudhi/sklearn/rips_persistence.py

@@ -52,23 +52,24 @@ def __init__(
         homology_coeff_field: int = 11,
         n_jobs: Optional[int] = None,
     ):
-        """
-        Constructor for the RipsPersistence class.
+        """Constructor for the RipsPersistence class.
 
         Parameters:
-            homology_dimensions (int or list of int): The returned persistence diagrams dimension(s).
+            homology_dimensions: The returned persistence diagrams dimension(s).
                 Short circuit the use of :class:`~gudhi.representations.preprocessing.DimensionSelector` when only one
                 dimension matters (in other words, when `homology_dimensions` is an int).
-            threshold (float): Rips maximal edge length value. Default is +Inf. Ignored if input_type is 'distance coo_matrix'.
-            input_type (str): Can be 'point cloud' when inputs are point clouds, 'full distance matrix',
+            threshold: Rips maximal edge length value. Default is +Inf. Ignored if input_type is
+                'distance coo_matrix'.
+            input_type: Can be 'point cloud' when inputs are point clouds, 'full distance matrix',
                 'lower distance matrix' when inputs are lower triangular distance matrix (can be full square,
                 but the upper part will not be considered), or 'distance coo_matrix' for a distance matrix in SciPy's
                 sparse format, which should contain each edge at most once (avoid the symmetric) and no diagonal entry.
                 Default is 'point cloud'.
-            num_collapses (int|str): Specify the number of iterations of :func:`~gudhi.flag_filtration.edge_collapse.reduce_graph`
-                (edge collapse) to perform on the graph. Default value is 'auto'.
-            homology_coeff_field (int): The homology coefficient field. Must be a prime number. Default value is 11.
-            n_jobs (Optional[int]): Number of jobs to run in parallel. `None` (default value) means `n_jobs = 1` unless in a
+            num_collapses: Specify the number of iterations of
+                :func:`~gudhi.flag_filtration.edge_collapse.reduce_graph` (edge collapse) to perform on the graph.
+                Default value is 'auto'.
+            homology_coeff_field: The homology coefficient field. Must be a prime number. Default value is 11.
+            n_jobs: Number of jobs to run in parallel. `None` (default value) means `n_jobs = 1` unless in a
                 joblib.parallel_backend context. `-1` means using all processors. cf.
                 https://joblib.readthedocs.io/en/latest/generated/joblib.Parallel.html for more details.
         """
@@ -165,7 +166,7 @@ def __transform(self, inp):
             dgm = _sparse(inp.row, inp.col, inp.data, inp.shape[0], max_dimension=max_dimension, max_edge_length=threshold, homology_coeff_field=self.homology_coeff_field)
         else:
             raise ValueError("Only 'point cloud', 'lower distance matrix', 'full distance matrix' and 'distance coo_matrix' are valid input_type") # move to __init__?
-        
+
         # dgm stops at n-2
         return [dgm[dim] if dim < len(dgm) else np.empty((0,2)) for dim in self.dim_list_]
 
@@ -177,7 +178,7 @@ def transform(self, X, Y=None):
 
         :return: Persistence diagrams in the format:
 
-              - If `homology_dimensions` was set to `n`: `[array( Hn(X[0]) ), array( Hn(X[1]) ), ...]` 
+              - If `homology_dimensions` was set to `n`: `[array( Hn(X[0]) ), array( Hn(X[1]) ), ...]`
               - If `homology_dimensions` was set to `[i, j]`:
                 `[[array( Hi(X[0]) ), array( Hj(X[0]) )], [array( Hi(X[1]) ), array( Hj(X[1]) )], ...]`
         :rtype: list of numpy ndarray of shape (,2) or list of list of numpy ndarray of shape (,2)