Made Pylint Happy

dect/directions.py

@@ -1,9 +1,10 @@
 """
-Helper function to generate a structured set of directions in 2 and 3 dimensions.
+Helper function to generate a structured set of directions in 2 and 3
+dimensions.
 """
 
-import torch
 import itertools
+import torch
 
 
 def generate_uniform_directions(num_thetas: int = 64, d: int = 3, device: str = "cpu"):
@@ -33,9 +34,9 @@ def generate_uniform_2d_directions(num_thetas: int = 64, device: str = "cpu"):
     """
     Generate uniformly sampled directions on the unit circle in two dimensions.
 
-    Provides a structured set of directions in two dimensions. First the interval
-    [0,2*pi] is devided into a regular grid and the corresponding angles on the
-    unit circle calculated.
+    Provides a structured set of directions in two dimensions. First the
+    interval [0,2*pi] is devided into a regular grid and the corresponding
+    angles on the unit circle calculated.
 
     Parameters
     ----------
@@ -60,15 +61,15 @@ def generate_multiview_directions(num_thetas: int, bump_steps: int, d: int):
     """
     Generates multiple sets of structured directions in n dimensions.
 
-    We generate sets of directions by embedding the 2d unit circle in
-    d dimensions and sample this unit circle in a structured fashion.
-    This generates d choose 2 structured directions that are organized
-    in channels, compatible with the ECT calculations.
+    We generate sets of directions by embedding the 2d unit circle in d
+    dimensions and sample this unit circle in a structured fashion. This
+    generates d choose 2 structured directions that are organized in channels,
+    compatible with the ECT calculations.
 
-    After computing the ECT, we obtain an d choose 2 channel image where
-    each channel consists of a structured ect along a hyperplane.
-    For the 3-d case we would obtain a 3 channel ect with direction sampled
-    along the xy, xz and yz planes in three dimensions.
+    After computing the ECT, we obtain an d choose 2 channel image where each
+    channel consists of a structured ect along a hyperplane. For the 3-d case we
+    would obtain a 3 channel ect with direction sampled along the xy, xz and yz
+    planes in three dimensions.
     """
     w = torch.vstack(
         [

dect/ect.py

@@ -75,9 +75,9 @@ def compute_ecc(
         The shape of the resulting tensor after summation. It has to be of the
         shape [num_discretization_steps, batch_size, num_thetas]
     scale: torch.FloatTensor
-        A single number that scales the sigmoid function by multiplying the sigmoid
-        with the scale. With high (100>) values, the ect will resemble a discrete
-        ECT and with lower values it will smooth the ECT.
+        A single number that scales the sigmoid function by multiplying the
+        sigmoid with the scale. With high (100>) values, the ect will resemble a
+        discrete ECT and with lower values it will smooth the ECT.
     """
     ecc = torch.nn.functional.sigmoid(scale * torch.sub(lin, nh))
 
@@ -115,7 +115,8 @@ def compute_ect_edges(data: Batch, v: torch.FloatTensor, lin: torch.FloatTensor)
     Parameters
     ----------
     batch : Batch
-        A batch of data containing the node coordinates, the edges and batch index.
+        A batch of data containing the node coordinates, the edges and batch
+        index.
     v: torch.FloatTensor
         The direction vector that contains the directions.
     lin: torch.FloatTensor