rpca anomaly projection changed

qolmat/imputations/rpca/rpca_noisy.py

@@ -141,6 +141,7 @@ def decompose_on_basis(
             A_Omega = rpca_utils.soft_thresholding(D - L @ Q, lam)
             A_Omega_C = D - L @ Q
             A = np.where(Omega, A_Omega, A_Omega_C)
+            # A = np.where(Omega, A_Omega, 0)
 
             Ac = np.linalg.norm(A - A_prev, np.inf)
             Lc = np.linalg.norm(L - L_prev, np.inf)
@@ -229,11 +230,7 @@ def decompose_rpca(
                     f"than the number of rows in the matrix but {period} >= {n_rows}!"
                 )
 
-        print("before")
-        print(D)
         D = utils.linear_interpolation(D)
-        print("after")
-        print(D)
 
         M, A, L, Q = self.decompose_rpca_algorithm(
             D,
@@ -421,8 +418,9 @@ def decompose_rpca_algorithm(
                 )
 
             A_Omega = rpca_utils.soft_thresholding(D - X, lam)
-            A_Omega_C = D - X
-            A = np.where(Omega, A_Omega, A_Omega_C)
+            # A_Omega_C = D - X
+            # A = np.where(Omega, A_Omega, A_Omega_C)
+            A = np.where(Omega, A_Omega, 0)
 
             Q = scp.linalg.solve(
                 a=tau * Ir + mu * (L.T @ L),

qolmat/imputations/rpca/rpca_utils.py

@@ -30,8 +30,6 @@ def approx_rank(
     """
     if threshold == 1:
         return min(M.shape)
-    print("approx_rank")
-    print(M)
     _, values_singular, _ = np.linalg.svd(M, full_matrices=True)
 
     cum_sum = np.cumsum(values_singular) / np.sum(values_singular)

tests/imputations/rpca/test_rpca_noisy.py

@@ -121,8 +121,6 @@ def test_rpca_noisy_zero_tau_zero_lambda(X: NDArray, X_interpolated: NDArray):
     """Test RPCA noisy results if tau and lambda equal zero."""
     rpca = RPCANoisy(tau=0, lam=0, norm="L2")
     Omega = ~np.isnan(X)
-    print(X)
-    print(Omega)
     X_result, A_result, _, _ = rpca.decompose_rpca(X, Omega)
     np.testing.assert_allclose(X_result, X_interpolated, atol=1e-4)
     np.testing.assert_allclose(A_result, np.full_like(X, 0), atol=1e-4)