Perform best-fit at upper limits?

[kdund]

For some sorts of checks it might be useful to know the value of all profiled parameters for the upper and lower confidence intervals-- would there be a way to have the alea toy runner to re-compute fits at the confidence intervals?

[hammannr]

Indeed this is quite useful to have, i.e. for getting the signal strength at the limits. So you're suggesting to include this as an option in the runner config? I think it should not be an issue conceptually to implement. It could go into the simulate_and_fit method in the runner and simply be included as a separate field in the fit_result, I think.

[kdund]

yep! The one thing I see it as annoying is that it would have to be called after the confidence interval calculation, so you'd need two places to call fits

[hammannr]

I think so too, yes. In principle, we could also include it in the confidence_interval method but it would make the return dependent on whether or not this is requested. In any case I think we won't get around an additional function call though without completely reqriting the confidence_interval method.