FIX: More

mne/forward/forward.py

@@ -317,7 +317,7 @@ def _block_diag(A, n):
     jj = jj * np.ones(ma, dtype=np.int64)[:, None]
     jj = jj.T.ravel()  # column indices foreach sparse bd
 
-    bd = sparse.coo_matrix((A.T.ravel(), np.c_[ii, jj].T)).tocsc()
+    bd = sparse.coo_matrix((A.T.ravel(), np.c_[ii, jj].T), copy=True).tocsc()
 
     return bd
 

mne/label.py

@@ -2575,7 +2575,7 @@ def _labels_to_stc_surf(labels, values, tmin, tstep, subject):
         data[hemi] = np.concatenate(data[hemi], axis=0).astype(float)
         cols = np.arange(len(vertices[hemi]))
         vertices[hemi], rows = np.unique(vertices[hemi], return_inverse=True)
-        mat = sparse.coo_matrix((np.ones(len(rows)), (rows, cols))).tocsr()
+        mat = sparse.coo_matrix((np.ones(len(rows)), (rows, cols)), copy=True).tocsr()
         mat = mat * sparse.diags(1.0 / np.asarray(mat.sum(axis=-1))[:, 0])
         data[hemi] = mat.dot(data[hemi])
     vertices = [vertices[hemi] for hemi in hemis]

mne/morph.py

@@ -341,7 +341,7 @@ def _compute_sparse_morph(vertices_from, subject_from, subject_to, subjects_dir=
     rows = np.arange(len(cols))
     data = np.ones(len(cols))
     morph_mat = sparse.coo_matrix(
-        (data, (rows, cols)), shape=(len(cols), len(cols))
+        (data, (rows, cols)), shape=(len(cols), len(cols)), copy=True
     ).tocsr()
     return vertices, morph_mat
 

mne/source_estimate.py

@@ -3361,7 +3361,9 @@ def _get_adjacency_from_edges(edges, n_times, verbose=None):
     data = np.ones(
         edges.data.size * n_times + 2 * n_vertices * (n_times - 1), dtype=np.int64
     )
-    adjacency = sparse.coo_matrix((data, (row, col)), shape=(n_times * n_vertices,) * 2)
+    adjacency = sparse.coo_matrix(
+        (data, (row, col)), shape=(n_times * n_vertices,) * 2, copy=True
+    )
     return adjacency
 
 

mne/stats/_adjacency.py

@@ -115,5 +115,7 @@ def combine_adjacency(*structure):
     # Handle the diagonal separately at the end to avoid duplicate entries
     edges[:, n_off:] = vertices.ravel()
     weights[n_off:] = 1.0
-    graph = sparse.coo_matrix((weights, edges), (vertices.size, vertices.size))
+    graph = sparse.coo_matrix(
+        (weights, edges), (vertices.size, vertices.size), copy=True
+    )
     return graph

mne/surface.py

@@ -406,7 +406,7 @@ def _triangle_neighbors(tris, npts):
     rows = tris.ravel()
     cols = np.repeat(np.arange(len(tris)), 3)
     data = np.ones(len(cols))
-    csr = coo_matrix((data, (rows, cols)), shape=(npts, len(tris))).tocsr()
+    csr = coo_matrix((data, (rows, cols)), shape=(npts, len(tris)), copy=True).tocsr()
     neighbor_tri = [
         csr.indices[start:stop] for start, stop in zip(csr.indptr[:-1], csr.indptr[1:])
     ]
@@ -1740,7 +1740,7 @@ def _mesh_edges(tris=None):
     a, b, c = tris.T
     x = np.concatenate((a, b, c))
     y = np.concatenate((b, c, a))
-    edges = coo_matrix((ones_ntris, (x, y)), shape=(npoints, npoints))
+    edges = coo_matrix((ones_ntris, (x, y)), shape=(npoints, npoints), copy=True)
     edges = edges.tocsr()
     edges = edges + edges.T
     return edges
@@ -1768,7 +1768,9 @@ def mesh_dist(tris, vert):
 
     # Euclidean distances between neighboring vertices
     dist = np.linalg.norm(vert[edges.row, :] - vert[edges.col, :], axis=1)
-    dist_matrix = csr_matrix((dist, (edges.row, edges.col)), shape=edges.shape)
+    dist_matrix = csr_matrix(
+        (dist, (edges.row, edges.col)), shape=edges.shape, copy=True
+    )
     return dist_matrix