found a bug in w-formula, 1/P_s should be outside sums, starting to think we should...