minor fix typing and np deprecation warning

e2cnn/group/group.py

@@ -151,7 +151,7 @@ def subgroup(self, id) -> Tuple[e2cnn.group.Group, Callable, Callable]:
         pass
 
     @property
-    def irreps(self) -> Dict[int, e2cnn.group.IrreducibleRepresentation]:
+    def irreps(self) -> Dict[str, e2cnn.group.IrreducibleRepresentation]:
         r"""
         Dictionary containing all irreducible representations (:class:`~e2cnn.group.IrreducibleRepresentation`)
         instantiated for this group.
@@ -163,7 +163,7 @@ def irreps(self) -> Dict[int, e2cnn.group.IrreducibleRepresentation]:
         return self._irreps
 
     @property
-    def representations(self) -> Dict[int, e2cnn.group.Representation]:
+    def representations(self) -> Dict[str, e2cnn.group.Representation]:
         r"""
         Dictionary containing all representations (:class:`~e2cnn.group.Representation`)
         instantiated for this group.
@@ -176,7 +176,7 @@ def representations(self) -> Dict[int, e2cnn.group.Representation]:
 
     @property
     @abstractmethod
-    def trivial_representation(self) -> e2cnn.group.Representation:
+    def trivial_representation(self) -> e2cnn.group.IrreducibleRepresentation:
         r"""
         Builds the trivial representation of the group.
         The trivial representation is a 1-dimensional representation which maps any element to 1,

e2cnn/group/representation.py

@@ -844,7 +844,7 @@ def build_induced_representation(group: e2cnn.group.Group,
     character = {}
 
     for g in group.elements:
-        repr_g = np.zeros((size, size), dtype=np.float)
+        repr_g = np.zeros((size, size), dtype=float)
         for r in representatives:
             gr = group.combine(g, r)
 
@@ -909,7 +909,7 @@ def build_commuting_matrix(rho: e2cnn.group.IrreducibleRepresentation, t):
 
     # rectangular matrix contained in one of the blocks (associated with the coset containing the trivial element)
     # of the change of basis matrix
-    v = np.zeros((size, repr.size), dtype=np.float)
+    v = np.zeros((size, repr.size), dtype=float)
 
     # position in the matrix
     p = 0
@@ -1096,8 +1096,8 @@ def null(A: Union[np.matrix, sparse.linalg.LinearOperator],
 
     # print(u.shape, s.shape, vh.shape)
     # print(min(s))
-    null_space = sp.compress((s <= eps), vh, axis=0)
-    return sp.transpose(null_space)
+    null_space = np.compress((s <= eps), vh, axis=0)
+    return np.transpose(null_space)
 
 
 def find_orthogonal_matrix(basis: np.ndarray, shape):