A very general note about the implementation: We could also consider to split all functions by their NP dependency. Because then in the end we would only have to do one q2 integration and the NP dependencies would be explicit (and all the benchmark points could be calculated in few seconds, because they are just quadratic linear combinations of the pre-calculated integrals).
I think that would also contribute to a better understanding of what impact the different wilson coefficient have.
A very general note about the implementation: We could also consider to split all functions by their NP dependency. Because then in the end we would only have to do one q2 integration and the NP dependencies would be explicit (and all the benchmark points could be calculated in few seconds, because they are just quadratic linear combinations of the pre-calculated integrals).
I think that would also contribute to a better understanding of what impact the different wilson coefficient have.