Merge pull request #366 from j-towns/close-matmul

Express matmul vjps in terms of matmul

autograd/numpy/numpy_vjps.py

@@ -317,19 +317,48 @@ def grad_inner(argnum, ans, A, B):
         return lambda G: tensordot_adjoint_1(A, G, axes, A_ndim, B_ndim)
 defvjp(anp.inner, partial(grad_inner, 0), partial(grad_inner, 1))
 
-def grad_matmul(argnum, ans, A, B):
-    A_ndim, B_ndim = anp.ndim(A), anp.ndim(B)
-    if A_ndim == 0 or B_ndim == 0:
-        raise ValueError("Scalar operands are not allowed, use '*' instead")
-    elif A_ndim == 1 or B_ndim == 1 or (A_ndim == 2 and B_ndim == 2):
-        axes = ([A_ndim - 1], [max(0, B_ndim - 2)])
-        if argnum == 0:
-            return lambda G: match_complex(A, tensordot_adjoint_0(B, G, axes, A_ndim, B_ndim))
-        elif argnum == 1:
-            return lambda G: match_complex(B, tensordot_adjoint_1(A, G, axes, A_ndim, B_ndim))
-    else:
-        return grad_einsum(argnum + 1, ans, ("...ij,...jk->...ik", A, B), None)
-defvjp(anp.matmul, partial(grad_matmul, 0), partial(grad_matmul, 1))
+def matmul_adjoint_0(B, G, A_meta, B_ndim):
+    if anp.ndim(G) == 0:  # A_ndim == B_ndim == 1
+        return unbroadcast(G * B, A_meta)
+    _, A_ndim, _, _ = A_meta
+    if A_ndim == 1:
+        G = anp.expand_dims(G, anp.ndim(G) - 1)
+    if B_ndim == 1:  # The result we need is an outer product
+        B = anp.expand_dims(B, 0)
+        G = anp.expand_dims(G, anp.ndim(G))
+    else:  # We need to swap the last two axes of B
+        B = anp.swapaxes(B, B_ndim - 2, B_ndim - 1)
+    result = anp.matmul(G, B)
+    return unbroadcast(result, A_meta)
+
+def matmul_adjoint_1(A, G, A_ndim, B_meta):
+    if anp.ndim(G) == 0:  # A_ndim == B_ndim == 1
+        return unbroadcast(G * A, B_meta)
+    _, B_ndim, _, _ = B_meta
+    B_is_vec = (B_ndim == 1)
+    if B_is_vec:
+        G = anp.expand_dims(G, anp.ndim(G))
+    if A_ndim == 1:  # The result we need is an outer product
+        A = anp.expand_dims(A, 1)
+        G = anp.expand_dims(G, anp.ndim(G) - 1)
+    else:  # We need to swap the last two axes of A
+        A = anp.swapaxes(A, A_ndim - 2, A_ndim - 1)
+    result = anp.matmul(A, G)
+    if B_is_vec:
+        result = anp.squeeze(result, anp.ndim(G) - 1)
+    return unbroadcast(result, B_meta)
+
+def matmul_vjp_0(ans, A, B):
+    A_meta = anp.metadata(A)
+    B_ndim = anp.ndim(B)
+    return lambda g: matmul_adjoint_0(B, g, A_meta, B_ndim)
+
+def matmul_vjp_1(ans, A, B):
+    A_ndim = anp.ndim(A)
+    B_meta = anp.metadata(B)
+    return lambda g: matmul_adjoint_1(A, g, A_ndim, B_meta)
+
+defvjp(anp.matmul, matmul_vjp_0, matmul_vjp_1)
 
 @primitive
 def dot_adjoint_0(B, G, A_ndim, B_ndim):