Set amplitude scalings equal zero for zero template units (#4455)

Author: chrishalcrow

src/spikeinterface/postprocessing/amplitude_scalings.py

@@ -201,6 +201,9 @@ def __init__(
             max_margin_collisions = delta_collision_samples + margin_waveforms
             self._margin = max_margin_collisions
 
+        # for some edge cases a template can be zero, leading to problems later
+        template_is_zero = [np.all(template == 0) for template in all_templates]
+
         self._all_templates = all_templates
         self._sparsity_mask = sparsity_mask
         self._nbefore = nbefore
@@ -209,6 +212,7 @@ def __init__(
         self._cut_out_after = cut_out_after
         self._handle_collisions = handle_collisions
         self._delta_collision_samples = delta_collision_samples
+        self._template_is_zero = template_is_zero
 
         self._kwargs.update(
             all_templates=all_templates,
@@ -220,6 +224,7 @@ def __init__(
             return_in_uV=return_in_uV,
             handle_collisions=handle_collisions,
             delta_collision_samples=delta_collision_samples,
+            template_is_zero=template_is_zero,
         )
 
     def get_dtype(self):
@@ -239,6 +244,7 @@ def compute(self, traces, peaks):
         cut_out_after = self._cut_out_after
         handle_collisions = self._handle_collisions
         delta_collision_samples = self._delta_collision_samples
+        template_is_zero = self._template_is_zero
 
         # local_spikes_within_margin = peaks
         # i0 = np.searchsorted(local_spikes_within_margin["sample_index"], left_margin)
@@ -265,7 +271,14 @@ def compute(self, traces, peaks):
             if spike_index in collisions.keys():
                 # we deal with overlapping spikes later
                 continue
+
             unit_index = spike["unit_index"]
+
+            if template_is_zero[unit_index]:
+                # if template is zero, linregress will fail so we intervene
+                scalings[spike_index] = 0
+                continue
+
             sample_centered = spike["sample_index"]
             (sparse_indices,) = np.nonzero(sparsity_mask[unit_index])
             template = all_templates[unit_index][:, sparse_indices]