Fix ruff errors

modules/data/utils/utils.py

@@ -401,12 +401,8 @@ def load_pointcloud_dataset(cfg):
             file_path=osp.join(data_dir, "stanford_bunny.npy"),
             accept_license=False,
         )
-
-        num_points = cfg["num_points"] if "num_points" in cfg else len(pos)
         pos = torch.tensor(pos)
 
-        pos = pos[np.random.choice(pos.shape[0], num_points, replace=False)]
-
     return CustomDataset(
         [
             torch_geometric.data.Data(
@@ -420,8 +416,10 @@ def load_pointcloud_dataset(cfg):
 def annulus_2d(D, N, R1=0.8, R2=1, A=0):
     n = 0
     P = np.array([[0.0] * D] * N)
+
+    rng = np.random.default_rng()
     while n < N:
-        p = np.random.uniform(-R2, R2, D)
+        p = rng.uniform(-R2, R2, D)
         if np.linalg.norm(p) > R2 or np.linalg.norm(p) < R1:
             continue
         if (p[0] > 0) and (np.abs(p[1]) < A / 2):

modules/transforms/liftings/pointcloud2graph/cover_lifting.py

@@ -6,34 +6,35 @@
 import statsmodels.stats.multitest as mt
 import torch
 import torch_geometric
-from gudhi import cover_complex
 from statsmodels.distributions.empirical_distribution import ECDF
 from torch_geometric.utils.convert import from_networkx
 
 from modules.transforms.liftings.pointcloud2graph.base import PointCloud2GraphLifting
 
+rng = np.random.default_rng()
 
-def persistent_homology(points: torch.Tensor, subcomplex_inds: list[int] = None):
+
+def persistent_homology(points: torch.Tensor, subcomplex_inds=None):
     st = gudhi.AlphaComplex(points=points).create_simplex_tree()
 
     if subcomplex_inds is not None:
-        subcomplex = []
-        for simplex in st.get_simplices():
-            if all(x in subcomplex_inds for x in simplex[0]):
-                subcomplex.append(simplex[0])
+        subcomplex = [
+            simplex
+            for simplex, _ in st.get_simplices()
+            if all(x in subcomplex_inds for x in simplex)
+        ]
 
         new_vertex = st.num_vertices()
         st.insert([new_vertex], 0)
         for simplex in subcomplex:
-            st.insert(simplex + [new_vertex], st.filtration(simplex))
+            st.insert([*simplex, new_vertex], st.filtration(simplex))
 
     persistence = st.persistence()
-    diagram = np.array(
+
+    return np.array(
         [(birth, death) for (dim, (birth, death)) in persistence if dim == 1]
     )
 
-    return diagram
-
 
 def transform(diagram):
     b, d = diagram[:, 0], diagram[:, 1]
@@ -43,7 +44,7 @@ def transform(diagram):
 
 def get_empirical_distribution(dim: int):
     """Generates empirical distribution of pi values for random pointcloud in R^{dim}"""
-    random_pc = np.random.uniform(size=(10000, dim))
+    random_pc = rng.uniform(size=(10000, dim))
     dgm_rand = persistent_homology(random_pc)
     return ECDF(transform(dgm_rand))
 
@@ -55,9 +56,7 @@ def test_weak_universality(emp_cdf: ECDF, diagram, alpha: float = 0.05):
 
 
 def sample_points(points: torch.Tensor, n=300):
-    return points[
-        np.random.choice(points.shape[0], min(n, points.shape[0]), replace=False)
-    ]
+    return points[rng.choice(points.shape[0], min(n, points.shape[0]), replace=False)]
 
 
 class CoverLifting(PointCloud2GraphLifting):
@@ -72,7 +71,6 @@ class CoverLifting(PointCloud2GraphLifting):
     def __init__(
         self,
         ambient_dim: int = 2,
-        cover_complex: gudhi.cover_complex.CoverComplex = None,
         **kwargs,
     ):
         super().__init__(**kwargs)

test/transforms/liftings/pointcloud2graph/test_cover_lifting.py

@@ -1,7 +1,6 @@
 """Test the message passing module."""
 
 import networkx as nx
-import torch
 from torch_geometric.utils.convert import to_networkx
 
 from modules.data.utils.utils import load_annulus
@@ -21,12 +20,7 @@ def setup_method(self):
     def test_lift_topology(self):
         """Test the lift_topology method."""
         # Test the lift_topology method
-        lifted_dataset = self.lifting(self.data)
+        lifted_data = self.lifting(self.data)
 
-        # g = nx.Graph()
-        # us, vs = lifted_dataset["edge_index"]
-
-        # for u, v in zip(us, vs):
-        #     g.add_edge(u, v)
-
-        # nx.cycles.find_cycle(g)
+        g = to_networkx(lifted_data, to_undirected=True)
+        nx.find_cycle(g)

tutorials/pointcloud2graph/cover_lifting.ipynb

@@ -45,7 +45,6 @@
     "from modules.utils.utils import (\n",
     "    describe_data,\n",
     "    load_dataset_config,\n",
-    "    load_model_config,\n",
     "    load_transform_config,\n",
     ")"
    ]
@@ -257,8 +256,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "import torch\n",
-    "from torch_geometric.nn import global_mean_pool\n",
     "from torch_geometric.nn.models import GraphSAGE"
    ]
   },