Detach

test/test_grad/test_pos.py

@@ -169,13 +169,11 @@ def test_autograd(dtype: torch.dtype, name: str) -> None:
     }
 
     # variable to be differentiated
-    positions.requires_grad_(True)
+    pos = positions.clone().requires_grad_(True)
 
     # automatic gradient
-    energy = torch.sum(dftd3(numbers, positions, param))
-    (grad,) = torch.autograd.grad(energy, positions)
-
-    positions.detach_()
+    energy = torch.sum(dftd3(numbers, pos, param))
+    (grad,) = torch.autograd.grad(energy, pos)
 
     assert pytest.approx(ref.cpu(), abs=tol) == grad.cpu()
 
@@ -243,11 +241,14 @@ def test_functorch(dtype: torch.dtype, name: str) -> None:
         "a2": torch.tensor(5.00000000, **dd),
     }
 
-    def dftd3_func(pos: Tensor) -> Tensor:
-        return dftd3(numbers, pos, param).sum()
+    # variable to be differentiated
+    pos = positions.clone().requires_grad_(True)
+
+    def dftd3_func(p: Tensor) -> Tensor:
+        return dftd3(numbers, p, param).sum()
 
-    grad = jacrev(dftd3_func)(positions)
+    grad = jacrev(dftd3_func)(pos)
     assert isinstance(grad, Tensor)
 
     assert grad.shape == ref.shape
-    assert pytest.approx(ref.cpu(), abs=tol) == grad.cpu()
+    assert pytest.approx(ref.cpu(), abs=tol) == grad.detach().cpu()