Merge pull request #538 from VincentRouvreau/empty_diagram_management_for_representations

Empty diagram management for representations

7 files changed, 216 insertions, 41 deletions

diff --git a/src/python/gudhi/cubical_complex.pyx b/src/python/gudhi/cubical_complex.pyx
index 97c69a2d..8e244bb8 100644
--- a/src/python/gudhi/cubical_complex.pyx
+++ b/src/python/gudhi/cubical_complex.pyx
@@ -281,4 +281,8 @@ cdef class CubicalComplex:
             launched first.
         """
         assert self.pcohptr != NULL, "compute_persistence() must be called before persistence_intervals_in_dimension()"
-        return np.array(self.pcohptr.intervals_in_dimension(dimension))
+        piid = np.array(self.pcohptr.intervals_in_dimension(dimension))
+        # Workaround https://github.com/GUDHI/gudhi-devel/issues/507
+        if len(piid) == 0:
+            return np.empty(shape = [0, 2])
+        return piid
diff --git a/src/python/gudhi/periodic_cubical_complex.pyx b/src/python/gudhi/periodic_cubical_complex.pyx
index ef1d3080..6c21e902 100644
--- a/src/python/gudhi/periodic_cubical_complex.pyx
+++ b/src/python/gudhi/periodic_cubical_complex.pyx
@@ -280,4 +280,8 @@ cdef class PeriodicCubicalComplex:
             launched first.
         """
         assert self.pcohptr != NULL, "compute_persistence() must be called before persistence_intervals_in_dimension()"
-        return np.array(self.pcohptr.intervals_in_dimension(dimension))
+        piid = np.array(self.pcohptr.intervals_in_dimension(dimension))
+        # Workaround https://github.com/GUDHI/gudhi-devel/issues/507
+        if len(piid) == 0:
+            return np.empty(shape = [0, 2])
+        return piid
diff --git a/src/python/gudhi/representations/vector_methods.py b/src/python/gudhi/representations/vector_methods.py
index 84bc99a2..e883b5dd 100644
--- a/src/python/gudhi/representations/vector_methods.py
+++ b/src/python/gudhi/representations/vector_methods.py
@@ -6,6 +6,7 @@
 #
 # Modification(s):
 #   - 2020/06 Martin: ATOL integration
+#   - 2021/11 Vincent Rouvreau: factorize _automatic_sample_range
 
 import numpy as np
 from sklearn.base          import BaseEstimator, TransformerMixin
@@ -45,10 +46,14 @@ class PersistenceImage(BaseEstimator, TransformerMixin):
             y (n x 1 array): persistence diagram labels (unused).
         """
         if np.isnan(np.array(self.im_range)).any():
-            new_X = BirthPersistenceTransform().fit_transform(X)
-            pre = DiagramScaler(use=True, scalers=[([0], MinMaxScaler()), ([1], MinMaxScaler())]).fit(new_X,y)
-            [mx,my],[Mx,My] = [pre.scalers[0][1].data_min_[0], pre.scalers[1][1].data_min_[0]], [pre.scalers[0][1].data_max_[0], pre.scalers[1][1].data_max_[0]]
-            self.im_range = np.where(np.isnan(np.array(self.im_range)), np.array([mx, Mx, my, My]), np.array(self.im_range))
+            try:
+                new_X = BirthPersistenceTransform().fit_transform(X)
+                pre = DiagramScaler(use=True, scalers=[([0], MinMaxScaler()), ([1], MinMaxScaler())]).fit(new_X,y)
+                [mx,my],[Mx,My] = [pre.scalers[0][1].data_min_[0], pre.scalers[1][1].data_min_[0]], [pre.scalers[0][1].data_max_[0], pre.scalers[1][1].data_max_[0]]
+                self.im_range = np.where(np.isnan(np.array(self.im_range)), np.array([mx, Mx, my, My]), np.array(self.im_range))
+            except ValueError:
+                # Empty persistence diagram case - https://github.com/GUDHI/gudhi-devel/issues/507
+                pass
         return self
 
     def transform(self, X):
@@ -94,6 +99,28 @@ class PersistenceImage(BaseEstimator, TransformerMixin):
         """
         return self.fit_transform([diag])[0,:]
 
+def _automatic_sample_range(sample_range, X, y):
+        """
+        Compute and returns sample range from the persistence diagrams if one of the sample_range values is numpy.nan.
+
+        Parameters:
+            sample_range (a numpy array of 2 float): minimum and maximum of all piecewise-linear function domains, of
+                the form [x_min, x_max].
+            X (list of n x 2 numpy arrays): input persistence diagrams.
+            y (n x 1 array): persistence diagram labels (unused).
+        """
+        nan_in_range = np.isnan(sample_range)
+        if nan_in_range.any():
+            try:
+                pre = DiagramScaler(use=True, scalers=[([0], MinMaxScaler()), ([1], MinMaxScaler())]).fit(X,y)
+                [mx,my] = [pre.scalers[0][1].data_min_[0], pre.scalers[1][1].data_min_[0]]
+                [Mx,My] = [pre.scalers[0][1].data_max_[0], pre.scalers[1][1].data_max_[0]]
+                return np.where(nan_in_range, np.array([mx, My]), sample_range)
+            except ValueError:
+                # Empty persistence diagram case - https://github.com/GUDHI/gudhi-devel/issues/507
+                pass
+        return sample_range
+
 class Landscape(BaseEstimator, TransformerMixin):
     """
     This is a class for computing persistence landscapes from a list of persistence diagrams. A persistence landscape is a collection of 1D piecewise-linear functions computed from the rank function associated to the persistence diagram. These piecewise-linear functions are then sampled evenly on a given range and the corresponding vectors of samples are concatenated and returned. See http://jmlr.org/papers/v16/bubenik15a.html for more details.
@@ -119,10 +146,7 @@ class Landscape(BaseEstimator, TransformerMixin):
             X (list of n x 2 numpy arrays): input persistence diagrams.
             y (n x 1 array): persistence diagram labels (unused).
         """
-        if self.nan_in_range.any():
-            pre = DiagramScaler(use=True, scalers=[([0], MinMaxScaler()), ([1], MinMaxScaler())]).fit(X,y)
-            [mx,my],[Mx,My] = [pre.scalers[0][1].data_min_[0], pre.scalers[1][1].data_min_[0]], [pre.scalers[0][1].data_max_[0], pre.scalers[1][1].data_max_[0]]
-            self.sample_range = np.where(self.nan_in_range, np.array([mx, My]), np.array(self.sample_range))
+        self.sample_range = _automatic_sample_range(np.array(self.sample_range), X, y)
         return self
 
     def transform(self, X):
@@ -218,10 +242,7 @@ class Silhouette(BaseEstimator, TransformerMixin):
             X (list of n x 2 numpy arrays): input persistence diagrams.
             y (n x 1 array): persistence diagram labels (unused).
         """
-        if np.isnan(np.array(self.sample_range)).any():
-            pre = DiagramScaler(use=True, scalers=[([0], MinMaxScaler()), ([1], MinMaxScaler())]).fit(X,y)
-            [mx,my],[Mx,My] = [pre.scalers[0][1].data_min_[0], pre.scalers[1][1].data_min_[0]], [pre.scalers[0][1].data_max_[0], pre.scalers[1][1].data_max_[0]]
-            self.sample_range = np.where(np.isnan(np.array(self.sample_range)), np.array([mx, My]), np.array(self.sample_range))
+        self.sample_range = _automatic_sample_range(np.array(self.sample_range), X, y)
         return self
 
     def transform(self, X):
@@ -307,10 +328,7 @@ class BettiCurve(BaseEstimator, TransformerMixin):
             X (list of n x 2 numpy arrays): input persistence diagrams.
             y (n x 1 array): persistence diagram labels (unused).
         """
-        if np.isnan(np.array(self.sample_range)).any():
-            pre = DiagramScaler(use=True, scalers=[([0], MinMaxScaler()), ([1], MinMaxScaler())]).fit(X,y)
-            [mx,my],[Mx,My] = [pre.scalers[0][1].data_min_[0], pre.scalers[1][1].data_min_[0]], [pre.scalers[0][1].data_max_[0], pre.scalers[1][1].data_max_[0]]
-            self.sample_range = np.where(np.isnan(np.array(self.sample_range)), np.array([mx, My]), np.array(self.sample_range))
+        self.sample_range = _automatic_sample_range(np.array(self.sample_range), X, y)
         return self
 
     def transform(self, X):
@@ -374,10 +392,7 @@ class Entropy(BaseEstimator, TransformerMixin):
             X (list of n x 2 numpy arrays): input persistence diagrams.
             y (n x 1 array): persistence diagram labels (unused).
         """
-        if np.isnan(np.array(self.sample_range)).any():
-            pre = DiagramScaler(use=True, scalers=[([0], MinMaxScaler()), ([1], MinMaxScaler())]).fit(X,y)
-            [mx,my],[Mx,My] = [pre.scalers[0][1].data_min_[0], pre.scalers[1][1].data_min_[0]], [pre.scalers[0][1].data_max_[0], pre.scalers[1][1].data_max_[0]]
-            self.sample_range = np.where(np.isnan(np.array(self.sample_range)), np.array([mx, My]), np.array(self.sample_range))
+        self.sample_range = _automatic_sample_range(np.array(self.sample_range), X, y)
         return self
 
     def transform(self, X):
@@ -396,9 +411,13 @@ class Entropy(BaseEstimator, TransformerMixin):
         new_X = BirthPersistenceTransform().fit_transform(X)        
 
         for i in range(num_diag):
-
             orig_diagram, diagram, num_pts_in_diag = X[i], new_X[i], X[i].shape[0]
-            new_diagram = DiagramScaler(use=True, scalers=[([1], MaxAbsScaler())]).fit_transform([diagram])[0]
+            try:
+                new_diagram = DiagramScaler(use=True, scalers=[([1], MaxAbsScaler())]).fit_transform([diagram])[0]
+            except ValueError:
+                # Empty persistence diagram case - https://github.com/GUDHI/gudhi-devel/issues/507
+                assert len(diagram) == 0
+                new_diagram = np.empty(shape = [0, 2])
 
             if self.mode == "scalar":
                 ent = - np.sum( np.multiply(new_diagram[:,1], np.log(new_diagram[:,1])) )
@@ -412,12 +431,11 @@ class Entropy(BaseEstimator, TransformerMixin):
                     max_idx = np.clip(np.ceil((py - self.sample_range[0]) / step_x).astype(int), 0, self.resolution)
                     for k in range(min_idx, max_idx):
                         ent[k] += (-1) * new_diagram[j,1] * np.log(new_diagram[j,1])
-                    if self.normalized:
-                        ent = ent / np.linalg.norm(ent, ord=1)
-                    Xfit.append(np.reshape(ent,[1,-1]))
-
-        Xfit = np.concatenate(Xfit, 0)
+                if self.normalized:
+                    ent = ent / np.linalg.norm(ent, ord=1)
+                Xfit.append(np.reshape(ent,[1,-1]))
 
+        Xfit = np.concatenate(Xfit, axis=0)
         return Xfit
 
     def __call__(self, diag):
@@ -478,7 +496,13 @@ class TopologicalVector(BaseEstimator, TransformerMixin):
             diagram, num_pts_in_diag = X[i], X[i].shape[0]
             pers = 0.5 * (diagram[:,1]-diagram[:,0])
             min_pers = np.minimum(pers,np.transpose(pers))
-            distances = DistanceMetric.get_metric("chebyshev").pairwise(diagram)
+            # Works fine with sklearn 1.0, but an ValueError exception is thrown on past versions
+            try:
+                distances = DistanceMetric.get_metric("chebyshev").pairwise(diagram)
+            except ValueError:
+                # Empty persistence diagram case - https://github.com/GUDHI/gudhi-devel/issues/507
+                assert len(diagram) == 0
+                distances = np.empty(shape = [0, 0])
             vect = np.flip(np.sort(np.triu(np.minimum(distances, min_pers)), axis=None), 0)
             dim = min(len(vect), thresh)
             Xfit[i, :dim] = vect[:dim]
diff --git a/src/python/gudhi/simplex_tree.pyx b/src/python/gudhi/simplex_tree.pyx
index 9c51cb46..c3720936 100644
--- a/src/python/gudhi/simplex_tree.pyx
+++ b/src/python/gudhi/simplex_tree.pyx
@@ -9,8 +9,7 @@
 
 from cython.operator import dereference, preincrement
 from libc.stdint cimport intptr_t
-import numpy
-from numpy import array as np_array
+import numpy as np
 cimport gudhi.simplex_tree
 
 __author__ = "Vincent Rouvreau"
@@ -542,7 +541,11 @@ cdef class SimplexTree:
             function to be launched first.
         """
         assert self.pcohptr != NULL, "compute_persistence() must be called before persistence_intervals_in_dimension()"
-        return np_array(self.pcohptr.intervals_in_dimension(dimension))
+        piid = np.array(self.pcohptr.intervals_in_dimension(dimension))
+        # Workaround https://github.com/GUDHI/gudhi-devel/issues/507
+        if len(piid) == 0:
+            return np.empty(shape = [0, 2])
+        return piid
 
     def persistence_pairs(self):
         """This function returns a list of persistence birth and death simplices pairs.
@@ -583,8 +586,8 @@ cdef class SimplexTree:
         """
         assert self.pcohptr != NULL, "lower_star_persistence_generators() requires that persistence() be called first."
         gen = self.pcohptr.lower_star_generators()
-        normal = [np_array(d).reshape(-1,2) for d in gen.first]
-        infinite = [np_array(d) for d in gen.second]
+        normal = [np.array(d).reshape(-1,2) for d in gen.first]
+        infinite = [np.array(d) for d in gen.second]
         return (normal, infinite)
 
     def flag_persistence_generators(self):
@@ -602,19 +605,19 @@ cdef class SimplexTree:
         assert self.pcohptr != NULL, "flag_persistence_generators() requires that persistence() be called first."
         gen = self.pcohptr.flag_generators()
         if len(gen.first) == 0:
-            normal0 = numpy.empty((0,3))
+            normal0 = np.empty((0,3))
             normals = []
         else:
             l = iter(gen.first)
-            normal0 = np_array(next(l)).reshape(-1,3)
-            normals = [np_array(d).reshape(-1,4) for d in l]
+            normal0 = np.array(next(l)).reshape(-1,3)
+            normals = [np.array(d).reshape(-1,4) for d in l]
         if len(gen.second) == 0:
-            infinite0 = numpy.empty(0)
+            infinite0 = np.empty(0)
             infinites = []
         else:
             l = iter(gen.second)
-            infinite0 = np_array(next(l))
-            infinites = [np_array(d).reshape(-1,2) for d in l]
+            infinite0 = np.array(next(l))
+            infinites = [np.array(d).reshape(-1,2) for d in l]
         return (normal0, normals, infinite0, infinites)
 
     def collapse_edges(self, nb_iterations = 1):
diff --git a/src/python/test/test_cubical_complex.py b/src/python/test/test_cubical_complex.py
index d0e4e9e8..29d559b3 100755
--- a/src/python/test/test_cubical_complex.py
+++ b/src/python/test/test_cubical_complex.py
@@ -174,3 +174,28 @@ def test_periodic_cofaces_of_persistence_pairs_when_pd_has_no_paired_birth_and_d
     assert np.array_equal(pairs[1][0], np.array([0]))
     assert np.array_equal(pairs[1][1], np.array([0, 1]))
     assert np.array_equal(pairs[1][2], np.array([1]))
+
+def test_cubical_persistence_intervals_in_dimension():
+    cub = CubicalComplex(
+        dimensions=[3, 3],
+        top_dimensional_cells=[1, 2, 3, 4, 5, 6, 7, 8, 9],
+    )
+    cub.compute_persistence()
+    H0 = cub.persistence_intervals_in_dimension(0)
+    assert np.array_equal(H0, np.array([[ 1., float("inf")]]))
+    assert cub.persistence_intervals_in_dimension(1).shape == (0, 2)
+
+def test_periodic_cubical_persistence_intervals_in_dimension():
+    cub = PeriodicCubicalComplex(
+        dimensions=[3, 3],
+        top_dimensional_cells=[1, 2, 3, 4, 5, 6, 7, 8, 9],
+        periodic_dimensions = [True, True]
+    )
+    cub.compute_persistence()
+    H0 = cub.persistence_intervals_in_dimension(0)
+    assert np.array_equal(H0, np.array([[ 1., float("inf")]]))
+    H1 = cub.persistence_intervals_in_dimension(1)
+    assert np.array_equal(H1, np.array([[ 3., float("inf")], [ 7., float("inf")]]))
+    H2 = cub.persistence_intervals_in_dimension(2)
+    assert np.array_equal(H2, np.array([[ 9., float("inf")]]))
+    assert cub.persistence_intervals_in_dimension(3).shape == (0, 2)
diff --git a/src/python/test/test_representations.py b/src/python/test/test_representations.py
index cda1a15b..93461f1e 100755
--- a/src/python/test/test_representations.py
+++ b/src/python/test/test_representations.py
@@ -3,9 +3,23 @@ import sys
 import matplotlib.pyplot as plt
 import numpy as np
 import pytest
+import random
 
 from sklearn.cluster import KMeans
 
+# Vectorization
+from gudhi.representations import (Landscape, Silhouette, BettiCurve, ComplexPolynomial,\
+  TopologicalVector, PersistenceImage, Entropy)
+
+# Preprocessing
+from gudhi.representations import (BirthPersistenceTransform, Clamping, DiagramScaler, Padding, ProminentPoints, \
+  DiagramSelector)
+
+# Kernel
+from gudhi.representations import (PersistenceWeightedGaussianKernel, \
+  PersistenceScaleSpaceKernel, SlicedWassersteinDistance,\
+  SlicedWassersteinKernel, PersistenceFisherKernel, WassersteinDistance)
+
 
 def test_representations_examples():
     # Disable graphics for testing purposes
@@ -98,3 +112,60 @@ def test_infinity():
     assert c[1] == 0
     assert c[7] == 3
     assert c[9] == 2
+
+
+def test_preprocessing_empty_diagrams():
+    empty_diag = np.empty(shape = [0, 2])
+    assert not np.any(BirthPersistenceTransform()(empty_diag))
+    assert not np.any(Clamping().fit_transform(empty_diag))
+    assert not np.any(DiagramScaler()(empty_diag))
+    assert not np.any(Padding()(empty_diag))
+    assert not np.any(ProminentPoints()(empty_diag))
+    assert not np.any(DiagramSelector()(empty_diag))
+
+def pow(n):
+  return lambda x: np.power(x[1]-x[0],n)
+
+def test_vectorization_empty_diagrams():
+    empty_diag = np.empty(shape = [0, 2])
+    random_resolution = random.randint(50,100)*10 # between 500 and 1000
+    print("resolution = ", random_resolution)
+    lsc = Landscape(resolution=random_resolution)(empty_diag)
+    assert not np.any(lsc)
+    assert lsc.shape[0]%random_resolution == 0
+    slt = Silhouette(resolution=random_resolution, weight=pow(2))(empty_diag)
+    assert not np.any(slt)
+    assert slt.shape[0] == random_resolution
+    btc = BettiCurve(resolution=random_resolution)(empty_diag)
+    assert not np.any(btc)
+    assert btc.shape[0] == random_resolution
+    cpp = ComplexPolynomial(threshold=random_resolution, polynomial_type="T")(empty_diag)
+    assert not np.any(cpp)
+    assert cpp.shape[0] == random_resolution
+    tpv = TopologicalVector(threshold=random_resolution)(empty_diag)
+    assert tpv.shape[0] == random_resolution
+    assert not np.any(tpv)
+    prmg = PersistenceImage(resolution=[random_resolution,random_resolution])(empty_diag)
+    assert not np.any(prmg)
+    assert prmg.shape[0] == random_resolution * random_resolution
+    sce = Entropy(mode="scalar", resolution=random_resolution)(empty_diag)
+    assert not np.any(sce)
+    assert sce.shape[0] == 1
+    scv = Entropy(mode="vector", normalized=False, resolution=random_resolution)(empty_diag)
+    assert not np.any(scv)
+    assert scv.shape[0] == random_resolution
+
+def test_kernel_empty_diagrams():
+    empty_diag = np.empty(shape = [0, 2])
+    assert SlicedWassersteinDistance(num_directions=100)(empty_diag, empty_diag) == 0.
+    assert SlicedWassersteinKernel(num_directions=100, bandwidth=1.)(empty_diag, empty_diag) == 1.
+    assert WassersteinDistance(mode="hera", delta=0.0001)(empty_diag, empty_diag) == 0.
+    assert WassersteinDistance(mode="pot")(empty_diag, empty_diag) == 0.
+    assert BottleneckDistance(epsilon=.001)(empty_diag, empty_diag) == 0.
+    assert BottleneckDistance()(empty_diag, empty_diag) == 0.
+#    PersistenceWeightedGaussianKernel(bandwidth=1., kernel_approx=None, weight=arctan(1.,1.))(empty_diag, empty_diag)
+#    PersistenceWeightedGaussianKernel(kernel_approx=RBFSampler(gamma=1./2, n_components=100000).fit(np.ones([1,2])), weight=arctan(1.,1.))(empty_diag, empty_diag)
+#    PersistenceScaleSpaceKernel(bandwidth=1.)(empty_diag, empty_diag)
+#    PersistenceScaleSpaceKernel(kernel_approx=RBFSampler(gamma=1./2, n_components=100000).fit(np.ones([1,2])))(empty_diag, empty_diag)
+#    PersistenceFisherKernel(bandwidth_fisher=1., bandwidth=1.)(empty_diag, empty_diag)
+#    PersistenceFisherKernel(bandwidth_fisher=1., bandwidth=1., kernel_approx=RBFSampler(gamma=1./2, n_components=100000).fit(np.ones([1,2])))(empty_diag, empty_diag)
diff --git a/src/python/test/test_simplex_tree.py b/src/python/test/test_simplex_tree.py
index a3eacaa9..31c46213 100755
--- a/src/python/test/test_simplex_tree.py
+++ b/src/python/test/test_simplex_tree.py
@@ -9,6 +9,7 @@
 """
 
 from gudhi import SimplexTree, __GUDHI_USE_EIGEN3
+import numpy as np
 import pytest
 
 __author__ = "Vincent Rouvreau"
@@ -404,3 +405,46 @@ def test_boundaries_iterator():
 
     with pytest.raises(RuntimeError):
         list(st.get_boundaries([6])) # (6) does not exist
+
+def test_persistence_intervals_in_dimension():
+    # Here is our triangulation of a 2-torus - taken from https://dioscuri-tda.org/Paris_TDA_Tutorial_2021.html
+    #   0-----3-----4-----0
+    #   | \   | \   | \   | \   |
+    #   |   \ |   \ |    \|   \ | 
+    #   1-----8-----7-----1
+    #   | \   | \   | \   | \   |
+    #   |   \ |   \ |   \ |   \ |
+    #   2-----5-----6-----2
+    #   | \   | \   | \   | \   |
+    #   |   \ |   \ |   \ |   \ |
+    #   0-----3-----4-----0
+    st = SimplexTree()
+    st.insert([0,1,8])
+    st.insert([0,3,8])
+    st.insert([3,7,8])
+    st.insert([3,4,7])
+    st.insert([1,4,7])
+    st.insert([0,1,4])
+    st.insert([1,2,5])
+    st.insert([1,5,8])
+    st.insert([5,6,8])
+    st.insert([6,7,8])
+    st.insert([2,6,7])
+    st.insert([1,2,7])
+    st.insert([0,2,3])
+    st.insert([2,3,5])
+    st.insert([3,4,5])
+    st.insert([4,5,6])
+    st.insert([0,4,6])
+    st.insert([0,2,6])
+    st.compute_persistence(persistence_dim_max=True)
+    
+    H0 = st.persistence_intervals_in_dimension(0)
+    assert np.array_equal(H0, np.array([[ 0., float("inf")]]))
+    H1 = st.persistence_intervals_in_dimension(1)
+    assert np.array_equal(H1, np.array([[ 0., float("inf")], [ 0., float("inf")]]))
+    H2 = st.persistence_intervals_in_dimension(2)
+    assert np.array_equal(H2, np.array([[ 0., float("inf")]]))
+    # Test empty case
+    assert st.persistence_intervals_in_dimension(3).shape == (0, 2)
+    
\ No newline at end of file