remove requires_grad from random rotations

Summary: Because rotations and (rotation) quaternions live on curved manifolds, it doesn't make sense to optimize them directly. Having a prominent option to require gradient on random ones may cause people to try, and isn't particularly useful.

Reviewed By: theschnitz

Differential Revision: D29160734

fbshipit-source-id: fc9e320672349fe334747c5b214655882a460a62

pytorch3d/transforms/rotation_conversions.py

@@ -282,9 +282,7 @@ def matrix_to_euler_angles(matrix, convention: str):
     return torch.stack(o, -1)
 
 
-def random_quaternions(
-    n: int, dtype: Optional[torch.dtype] = None, device=None, requires_grad=False
-):
+def random_quaternions(n: int, dtype: Optional[torch.dtype] = None, device=None):
     """
     Generate random quaternions representing rotations,
     i.e. versors with nonnegative real part.
@@ -294,21 +292,17 @@ def random_quaternions(
         dtype: Type to return.
         device: Desired device of returned tensor. Default:
             uses the current device for the default tensor type.
-        requires_grad: Whether the resulting tensor should have the gradient
-            flag set.
 
     Returns:
         Quaternions as tensor of shape (N, 4).
     """
-    o = torch.randn((n, 4), dtype=dtype, device=device, requires_grad=requires_grad)
+    o = torch.randn((n, 4), dtype=dtype, device=device)
     s = (o * o).sum(1)
     o = o / _copysign(torch.sqrt(s), o[:, 0])[:, None]
     return o
 
 
-def random_rotations(
-    n: int, dtype: Optional[torch.dtype] = None, device=None, requires_grad=False
-):
+def random_rotations(n: int, dtype: Optional[torch.dtype] = None, device=None):
     """
     Generate random rotations as 3x3 rotation matrices.
 
@@ -317,21 +311,15 @@ def random_rotations(
         dtype: Type to return.
         device: Device of returned tensor. Default: if None,
             uses the current device for the default tensor type.
-        requires_grad: Whether the resulting tensor should have the gradient
-            flag set.
 
     Returns:
         Rotation matrices as tensor of shape (n, 3, 3).
     """
-    quaternions = random_quaternions(
-        n, dtype=dtype, device=device, requires_grad=requires_grad
-    )
+    quaternions = random_quaternions(n, dtype=dtype, device=device)
     return quaternion_to_matrix(quaternions)
 
 
-def random_rotation(
-    dtype: Optional[torch.dtype] = None, device=None, requires_grad=False
-):
+def random_rotation(dtype: Optional[torch.dtype] = None, device=None):
     """
     Generate a single random 3x3 rotation matrix.
 
@@ -339,13 +327,11 @@ def random_rotation(
         dtype: Type to return
         device: Device of returned tensor. Default: if None,
             uses the current device for the default tensor type
-        requires_grad: Whether the resulting tensor should have the gradient
-            flag set
 
     Returns:
         Rotation matrix as tensor of shape (3, 3).
     """
-    return random_rotations(1, dtype, device, requires_grad)[0]
+    return random_rotations(1, dtype, device)[0]
 
 
 def standardize_quaternion(quaternions):

tests/test_rotation_conversions.py

@@ -76,7 +76,8 @@ class TestRotationConversion(TestCaseMixin, unittest.TestCase):
 
     def test_quat_grad_exists(self):
         """Quaternion calculations are differentiable."""
-        rotation = random_rotation(requires_grad=True)
+        rotation = random_rotation()
+        rotation.requires_grad = True
         modified = quaternion_to_matrix(matrix_to_quaternion(rotation))
         [g] = torch.autograd.grad(modified.sum(), rotation)
         self.assertTrue(torch.isfinite(g).all())
@@ -131,7 +132,8 @@ class TestRotationConversion(TestCaseMixin, unittest.TestCase):
 
     def test_euler_grad_exists(self):
         """Euler angle calculations are differentiable."""
-        rotation = random_rotation(dtype=torch.float64, requires_grad=True)
+        rotation = random_rotation(dtype=torch.float64)
+        rotation.requires_grad = True
         for convention in self._all_euler_angle_conventions():
             euler_angles = matrix_to_euler_angles(rotation, convention)
             mdata = euler_angles_to_matrix(euler_angles, convention)
@@ -218,7 +220,8 @@ class TestRotationConversion(TestCaseMixin, unittest.TestCase):
 
     def test_quaternion_application(self):
         """Applying a quaternion is the same as applying the matrix."""
-        quaternions = random_quaternions(3, torch.float64, requires_grad=True)
+        quaternions = random_quaternions(3, torch.float64)
+        quaternions.requires_grad = True
         matrices = quaternion_to_matrix(quaternions)
         points = torch.randn(3, 3, dtype=torch.float64, requires_grad=True)
         transform1 = quaternion_apply(quaternions, points)