Because of changes in how np.longdouble is handled on Windows (it has always only had double precision, but now they're being more strict about that somehow), it errors on an operation between a Python int and a np.longdouble. This adds an explicit conversion from int to np.uint64 first.
Because of changes in how
np.longdoubleis handled on Windows (it has always only had double precision, but now they're being more strict about that somehow), it errors on an operation between a Pythonintand anp.longdouble. This adds an explicit conversion frominttonp.uint64first.