Fix Segmentation Fault and ZeroDivisionError in Group Lasso (#292)

Co-authored-by: mathurinm <[email protected]>
mathurinm · Apr 8, 2024 · 1df21cf · 1df21cf
1 parent 1788a4f
commit 1df21cf
Show file tree

Hide file tree

Showing 3 changed files with 8 additions and 6 deletions.
diff --git a/celer/dropin_sklearn.py b/celer/dropin_sklearn.py
@@ -646,7 +646,7 @@ def fit(self, X, y):
                              % (n_samples, y.shape[0]))
 
         X, y, X_offset, y_offset, X_scale = _preprocess_data(
-            X, y, self.fit_intercept, copy=False)
+            X, y, fit_intercept=self.fit_intercept, copy=False)
 
         if not self.warm_start or not hasattr(self, "coef_"):
             self.coef_ = None

diff --git a/celer/group_fast.pyx b/celer/group_fast.pyx
@@ -89,10 +89,11 @@ cpdef floating dnorm_grp(
 
     else:  # scaling only with features in C
         for g_idx in range(ws_size):
+            g = C[g_idx]
+
             if weights[g] == INFINITY:
                 continue
 
-            g = C[g_idx]
             tmp = 0
             for k in range(grp_ptr[g], grp_ptr[g + 1]):
                 j = grp_indices[k]
@@ -418,8 +419,11 @@ cpdef celer_grp(
                                     &inc) / lc_groups[g]
                     norm_wg += w[j] ** 2
                 norm_wg = sqrt(norm_wg)
-                bst_scal = max(0.,
+                if norm_wg != 0.:
+                    bst_scal = max(0.,
                                1. - alpha * weights[g] / lc_groups[g] * n_samples / norm_wg)
+                else:
+                    bst_scal = 0.
 
                 for k in range(grp_ptr[g + 1] - grp_ptr[g]):
                     j = grp_indices[grp_ptr[g] + k]
@@ -448,5 +452,3 @@ cpdef celer_grp(
             'Fitting data with very small alpha causes precision issues.',
             ConvergenceWarning)
     return np.asarray(w), np.asarray(theta), np.asarray(gaps[:t + 1])
-
-
diff --git a/celer/tests/test_mtl.py b/celer/tests/test_mtl.py
@@ -18,7 +18,7 @@
 
 
 @pytest.mark.parametrize("sparse_X, fit_intercept",
-                         itertools.product([0, 1], [0, 1]))
+                         itertools.product([0, 1], [False, True]))
 def test_GroupLasso_Lasso_equivalence(sparse_X, fit_intercept):
     """Check that GroupLasso with groups of size 1 gives Lasso."""
     n_features = 1000