debug sinkhorn divergence gradients

Author: rflamary

ot/bregman.py

@@ -3173,8 +3173,7 @@ def empirical_sinkhorn2(X_s, X_t, reg, a=None, b=None, metric='sqeuclidean',
             return loss
 
     else:
-        M = dist(nx.to_numpy(X_s), nx.to_numpy(X_t), metric=metric)
-        M = nx.from_numpy(M, type_as=a)
+        M = dist(X_s, X_t, metric=metric)
 
         if log:
             sinkhorn_loss, log = sinkhorn2(a, b, M, reg, numItermax=numIterMax,
@@ -3287,6 +3286,10 @@ def empirical_sinkhorn_divergence(X_s, X_t, reg, a=None, b=None, metric='sqeucli
         International Conference on Artficial Intelligence and Statistics,
         (AISTATS) 21, 2018
     '''
+    X_s, X_t = list_to_array(X_s, X_t)
+
+    nx = get_backend(X_s, X_t)
+
     if log:
         sinkhorn_loss_ab, log_ab = empirical_sinkhorn2(X_s, X_t, reg, a, b, metric=metric,
                                                        numIterMax=numIterMax,
@@ -3313,7 +3316,7 @@ def empirical_sinkhorn_divergence(X_s, X_t, reg, a=None, b=None, metric='sqeucli
         log['log_sinkhorn_a'] = log_a
         log['log_sinkhorn_b'] = log_b
 
-        return max(0, sinkhorn_div), log
+        return nx.maximum(0, sinkhorn_div), log
 
     else:
         sinkhorn_loss_ab = empirical_sinkhorn2(X_s, X_t, reg, a, b, metric=metric,
@@ -3332,7 +3335,7 @@ def empirical_sinkhorn_divergence(X_s, X_t, reg, a=None, b=None, metric='sqeucli
                                               warn=warn, **kwargs)
 
         sinkhorn_div = sinkhorn_loss_ab - 0.5 * (sinkhorn_loss_a + sinkhorn_loss_b)
-        return max(0, sinkhorn_div)
+        return nx.maximum(0, sinkhorn_div)
 
 
 def screenkhorn(a, b, M, reg, ns_budget=None, nt_budget=None, uniform=False,

test/test_bregman.py

@@ -879,6 +879,34 @@ def test_empirical_sinkhorn_divergence(nx):
     ot.bregman.empirical_sinkhorn_divergence(X_sb, X_tb, 1, a=ab, b=bb, log=True)
 
 
+def test_empirical_sinkhorn_divergence_gradient():
+    # Test sinkhorn divergence
+    n = 10
+    a = np.linspace(1, n, n)
+    a /= a.sum()
+    b = ot.unif(n)
+    X_s = np.reshape(np.arange(n, dtype=np.float64), (n, 1))
+    X_t = np.reshape(np.arange(0, n * 2, 2, dtype=np.float64), (n, 1))
+
+    nx = ot.backend.TorchBackend()
+
+    ab, bb, X_sb, X_tb = nx.from_numpy(a, b, X_s, X_t)
+
+    ab.requires_grad = True
+    bb.requires_grad = True
+    X_sb.requires_grad = True
+    X_tb.requires_grad = True
+
+    emp_sinkhorn_div = ot.bregman.empirical_sinkhorn_divergence(X_sb, X_tb, 1, a=ab, b=bb)
+
+    emp_sinkhorn_div.backward()
+
+    assert ab.grad is not None
+    assert bb.grad is not None
+    assert X_sb.grad is not None
+    assert X_tb.grad is not None
+
+
 def test_stabilized_vs_sinkhorn_multidim(nx):
     # test if stable version matches sinkhorn
     # for multidimensional inputs