pre commit

Author: Samuel Boïté

ot/backend.py

@@ -1952,13 +1952,13 @@ def forward(ctx, val, grads, *inputs):
             def backward(ctx, grad_output):
                 # the gradients are grad
                 return (None, None) + tuple(g * grad_output for g in ctx.grads)
-        
+
         # define a differentiable SPD matrix sqrt
         # with closed-form VJP
         class MatrixSqrtFunction(Function):
             @staticmethod
             def forward(ctx, a):
-                a_sym = .5 * (a + a.transpose(-2, -1))
+                a_sym = 0.5 * (a + a.transpose(-2, -1))
                 L, V = torch.linalg.eigh(a_sym)
                 s = L.clamp_min(0).sqrt()
                 y = (V * s.unsqueeze(-2)) @ V.transpose(-2, -1)
@@ -1969,7 +1969,7 @@ def forward(ctx, a):
             @once_differentiable
             def backward(ctx, g):
                 s, V = ctx.saved_tensors
-                g_sym = .5 * (g + g.transpose(-2, -1))
+                g_sym = 0.5 * (g + g.transpose(-2, -1))
                 ghat = V.transpose(-2, -1) @ g_sym @ V
                 d = s.unsqueeze(-1) + s.unsqueeze(-2)
                 xhat = ghat / d

test/test_backend.py

@@ -823,16 +823,16 @@ def fun(a, b, d):
 
 
 def test_sqrtm_backward_torch():
-  if not torch:
-    pytest.skip("Torch not available")
-  nx = ot.backend.TorchBackend()
-  torch.manual_seed(42)
-  d = 5
-  A = torch.randn(d, d, dtype=torch.float64, device="cpu")
-  A = A @ A.T
-  A.requires_grad_(True)
-  func = lambda x: nx.sqrtm(x).sum()
-  assert torch.autograd.gradcheck(func, (A,), atol=1e-4, rtol=1e-4)
+    if not torch:
+        pytest.skip("Torch not available")
+    nx = ot.backend.TorchBackend()
+    torch.manual_seed(42)
+    d = 5
+    A = torch.randn(d, d, dtype=torch.float64, device="cpu")
+    A = A @ A.T
+    A.requires_grad_(True)
+    func = lambda x: nx.sqrtm(x).sum()
+    assert torch.autograd.gradcheck(func, (A,), atol=1e-4, rtol=1e-4)
 
 
 def test_get_backend_none():