MRG: Add support for smoothing_steps=0

mne/viz/_brain/_timeviewer.py

@@ -74,7 +74,7 @@ def __init__(self, brain):
         smoothing_slider = self.plotter.add_slider_widget(
             set_smoothing,
             value=default_smoothing_value,
-            rng=[1, 15], title="smoothing",
+            rng=[0, 15], title="smoothing",
             pointa=(0.82, 0.90),
             pointb=(0.98, 0.90)
         )

mne/viz/_brain/tests/test_brain.py

@@ -85,7 +85,7 @@ def test_brain_add_data(renderer):
 
     brain_data.add_data(hemi_data, fmin=fmin, hemi=hemi, fmax=fmax,
                         colormap='hot', vertices=hemi_vertices,
-                        colorbar=False, time=None)
+                        smoothing_steps=0, colorbar=False, time=None)
     brain_data.add_data(hemi_data, fmin=fmin, hemi=hemi, fmax=fmax,
                         colormap='hot', vertices=hemi_vertices,
                         initial_time=0., colorbar=True, time=None)

mne/viz/_brain/utils.py

@@ -36,6 +36,23 @@ def mesh_edges(faces):
     return edges
 
 
+def _nearest(vertices, adj_mat):
+    from scipy.sparse.csgraph import dijkstra
+    # Vertices can be out of order, so sort them to start ...
+    order = np.argsort(vertices)
+    vertices = vertices[order]
+    _, _, sources = dijkstra(adj_mat, False, indices=vertices, min_only=True,
+                             return_predecessors=True)
+    col = np.searchsorted(vertices, sources)
+    # ... then get things back to the correct configuration.
+    col = order[col]
+    row = np.arange(len(col))
+    data = np.ones(len(col))
+    mat = sparse.coo_matrix((data, (row, col)))
+    assert mat.shape == (adj_mat.shape[0], len(vertices)), mat.shape
+    return mat
+
+
 @verbose
 def smoothing_matrix(vertices, adj_mat, smoothing_steps=20, verbose=None):
     """Create a smoothing matrix.
@@ -67,6 +84,9 @@ def smoothing_matrix(vertices, adj_mat, smoothing_steps=20, verbose=None):
 
     logger.info("Updating smoothing matrix, be patient..")
 
+    if smoothing_steps == 0:
+        return _nearest(vertices, adj_mat)
+
     e = adj_mat.copy()
     e.data[e.data == 2] = 1
     n_vertices = e.shape[0]