Both for the 2D scan, but also for the 1D ones, we have to figure out a suitable figure-of-merit and methodology, possibly parametrized by input.
- It should combine assignment fidelity, QND, and QND-$\pi$.
- It may favor stable maxima, e.g. by smoothing with a convolutional mask (moving average), to improve noise-resilience.
- It may expose some of the figure-of-merit parameters as inputs of the protocol.
Both for the 2D scan, but also for the 1D ones, we have to figure out a suitable figure-of-merit and methodology, possibly parametrized by input.