Hi Chris,
Thanks for the suggestions!
bins is already a numpy array so I can’t do the sum you suggested. I mean, inside the definition of counts_obs it just says that the normal distribution is len(bins) dimensional, doesn’t it?. But I did take your advice, apparently I can sum over a generator. I also changed the prior distributions and the scale, but the Chains still fail. The relative uncertainty in the energies is apparently just too low for the algorithm.