When running tw_squared_error I ran into a bug that causes the code to fail out if the first object in the tuple for interval_where_one is an xr.DataArray and the second is a float, e.g. np.inf.
I think I pinpointed the issue:
|
a, b = interval_where_one |
|
|
|
if isinstance(a, (float, int)): |
|
a = xr.DataArray(a) |
|
b = xr.DataArray(b) |
Where b in this case will not be converted into a xr.DataArray, resulting in any methods, e.g. .min() failing.
I'll get a PR set up ASAP for y'all!