We have a fraction accumulator now, but it does not work correctly for weighted samples. However, we definitely need this in high-energy physics because we often estimate the efficiency of some detector from weighted simulation samples.
There are several ways how this could be implemented
1) New class weighted_fraction<T>
2) Specialization fraction<weighted_sum<T>>
3) Template specialization fraction<T, bool weight_support = False>
Number 3) is my least favored. I like 2), if that's feasible.
We have a fraction accumulator now, but it does not work correctly for weighted samples. However, we definitely need this in high-energy physics because we often estimate the efficiency of some detector from weighted simulation samples.
There are several ways how this could be implemented 1) New class
weighted_fraction<T>
2) Specializationfraction<weighted_sum<T>>
3) Template specializationfraction<T, bool weight_support = False>
Number 3) is my least favored. I like 2), if that's feasible.
Ping @henryiii @gohil-jay @jpivarski