GammaLib

To compute the ARF we have to integrate the point spread function multiplied by the effective area over the selection region.

This can be done by using the GResponse::npred(const GSource& source, const GObservation& obs) method.

Recall that GSource stores the physical attributes of a gamma-ray source, which is the distribution of photon arrival directions in form of a spatial source model, the photon energy, and the photon arrival time. As no energy dispersion is actually used, the energy attribute of GSource can be used as the measured energy. Some thinking is needed how this should be implement in the future once energy dispersion is supported. But for the time being it is okay.

So we could write a GArf method that computes the response given a spatial source model and an observation:

void GArf::set(const GResponse& rsp, const GRoi& roi, const GModelSpatial& spatial);

But the actual code needed if GResponse::npred should be used is actually too specific, hence we better implement this on the GCTAOnOffObservation level for the moment.