fix bug

CostaLab · Aug 20, 2021 · 9de86fc · 9de86fc
1 parent b46a846
commit 9de86fc
Show file tree

Hide file tree

Showing 2 changed files with 9 additions and 14 deletions.
diff --git a/scopen/MF.py b/scopen/MF.py
@@ -198,21 +198,18 @@ def _compute_regularization(alpha, l1_ratio, regularization):
     return l1_reg_W, l1_reg_H, l2_reg_W, l2_reg_H
 
 
-def _loss(X, W, H, square_root=False):
+def _loss(X, W, H):
     """Compute the Frobenius *squared* norm of X - dot(W, H).
     Parameters
     ----------
     X : float or array-like, shape (n_samples, n_features)
         Numpy masked arrays or arrays containing NaN are accepted.
     W : float or dense array-like, shape (n_samples, n_components)
     H : float or dense array-like, shape (n_components, n_features)
-    square_root : boolean, default False
-        If True, return np.sqrt(2 * res)
-        For beta == 2, it corresponds to the Frobenius norm.
     Returns
     -------
         res : float
-            Beta divergence of X and np.dot(X, H)
+            Frobenius norm of X and np.dot(X, H)
     """
     # The method can be called with scalars
     if not sp.issparse(X):
@@ -232,10 +229,7 @@ def _loss(X, W, H, square_root=False):
 
     assert not np.isnan(res)
     assert res >= 0
-    if square_root:
-        return np.sqrt(res * 2)
-    else:
-        return res
+    return np.sqrt(res * 2)
 
 
 def _initialize_nmf(X, n_components, init=None, eps=1e-6,
@@ -522,7 +516,7 @@ def _fit_coordinate_descent(X, W, H, tol=1e-4, max_iter=200, l1_reg_W=0,
                   f"violation: {_violation: .8f}")
 
         elif verbose == 2:
-            err = _loss(X, W, Ht.T, square_root=True)
+            err = _loss(X, W, Ht.T)
             print(f"{datetime.now().strftime('%m/%d/%Y %H:%M:%S')}, iteration: {n_iter: }, "
                   f"violation: {_violation: .8f}, error: {err: .8f}")
 
@@ -688,7 +682,6 @@ def non_negative_factorization(X, W=None, H=None, n_components=None,
     W, H, n_iter = _fit_coordinate_descent(X, W, H, tol, max_iter,
                                            l1_reg_W, l1_reg_H,
                                            l2_reg_W, l2_reg_H,
-                                           update_H=True,
                                            verbose=verbose,
                                            shuffle=shuffle,
                                            random_state=random_state)
@@ -876,7 +869,7 @@ def fit_transform(self, X, y=None, W=None, H=None):
             random_state=self.random_state, verbose=self.verbose,
             shuffle=self.shuffle)
 
-        self.reconstruction_err_ = _loss(X, W, H, square_root=True)
+        self.reconstruction_err_ = _loss(X, W, H)
 
         self.n_components_ = H.shape[0]
         self.components_ = H

diff --git a/scopen/Main.py b/scopen/Main.py
@@ -129,7 +129,7 @@ def estimate_rank(data, args):
         for n_components in n_components_list:
             arguments = (data, n_components, args.alpha,
                          args.max_iter, args.verbose,
-                         args.random_state. args.init)
+                         args.random_state, args.init)
 
             res = run_nmf(arguments)
             w_hat_dict[n_components] = res[0]
@@ -139,7 +139,9 @@ def estimate_rank(data, args):
     elif args.nc > 1:
         arguments_list = list()
         for n_components in n_components_list:
-            arguments = (data, n_components, args.alpha, args.max_iter, args.verbose, args.random_state, args.init)
+            arguments = (data, n_components, args.alpha,
+                         args.max_iter, args.verbose,
+                         args.random_state, args.init)
             arguments_list.append(arguments)
 
         with Pool(processes=args.nc) as pool: