Fix window bounds check in event_triggered_traces function

bnpm/timeSeries.py

@@ -425,7 +425,7 @@ def event_triggered_traces(
     idx_triggers_clean = torch.as_tensor(idx_triggers[~torch.isnan(idx_triggers)], dtype=torch.long)  ## remove nans from idx_triggers and convert to torch.long
 
     windows = torch.stack([xAxis + i for i in torch.as_tensor(idx_triggers_clean, dtype=torch.long)], dim=0)  ## make windows. shape = (n_triggers, len_window)
-    win_toInclude = (torch.any(windows<0, dim=1)==0) * (torch.any(windows>arr.shape[dim], dim=1)==0)  ## boolean array of windows that are within the bounds of the length of 'dim'
+    win_toInclude = (torch.any(windows<0, dim=1)==0) * (torch.any(windows>=arr.shape[dim], dim=1)==0)  ## boolean array of windows that are within the bounds of the length of 'dim'
     n_win_excluded = torch.sum(win_toInclude==False)  ## number of windows excluded due to window bounds. Only used for printing currently
     windows = windows[win_toInclude]  ## windows after pruning out windows that are out of bounds
     n_windows = windows.shape[0]  ## number of windows. Only used for printing currently
@@ -436,7 +436,7 @@ def event_triggered_traces(
     shape = list(arr.shape)  ## original shape
     dims_perm = [dim] + list(range(dim)) + list(range(dim+1, len(shape)))  ## new dims for indexing. put dim at dim 0
     arr_perm = arr.permute(*dims_perm)  ## permute to put 'dim' to dim 0
-    arr_idx = arr_perm.index_select(0, windows.reshape(-1))  ## index out windows along dim 0
+    arr_idx = arr_perm.index_select(dim=0, index=windows.reshape(-1))  ## index out windows along dim 0
     rs = list(arr_perm.shape[1:]) + [n_windows, win_bounds[1]-win_bounds[0]]  ## new shape for unflattening. 'dim' will be moved to dim -1, then reshaped to n_windows x len_window
     arr_idx_rs = arr_idx.permute(*(list(range(1, arr_idx.ndim)) + [0])).reshape(*rs)  ## permute to put current 'dim' (currently dim 0) to end, then reshape