DOCS: Missing `shapeU` description

[andrzejnovak]

shapeU nuisance type is not documented in the docs

[amarini]

I think the idea has been that after rateParam have been implemented, they should obtain the same thing in a cleaner way and that was the preferred way we suggested to implement the same effect.

[andrzejnovak]

I'm not sure not sure this is just as easy to implement for my use-case actually. Eg I want to fit for peak position by interpolating between sth like np.random.normal(3, 1, 10000) and np.random.normal(6, 1, 10000). The functionality of shapeU is convenient because I can easily switch from when there's an outside measurement available and supply shape prior or when there isn't and supply shapeU.

[nucleosynthesis]

I think shapeU is indeed for a different application than lnU (for which the rateParam is advised) For a flat prior on a shape uncertainty,.I think shapeU is still our recommended method so we should probably have some documentation on it. However, it's possible to remove the constraint by promoting the parameter to a POI and setting a range - thereby the effect is the same. Perhaps we should think about also describing this too, Best. Nick

…

On Thu, 10 Apr 2025, 18:30 Andrzej Novak, ***@***.***> wrote: I'm not sure not sure this is just as easy to implement for my use-case actually. Eg I want to fit for peak position by interpolating between sth like np.random.normal(3, 1, 10000) and np.random.normal(6, 1, 10000). The functionality of shapeU is convenient because I can easily switch from when there's an outside measurement available and supply shape prior or when there isn't and supply shapeU. — Reply to this email directly, view it on GitHub <#1071 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMEVW6RDYAZ4L5TSP6FAOD2Y2TCXAVCNFSM6AAAAAB24FXIGGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDOOJUGYZTGNJXGU> . You are receiving this because you are subscribed to this thread.Message ID: <cms-analysis/HiggsAnalysis-CombinedLimit/issues/1071/2794633575@ github.com> *andrzejnovak* left a comment (cms-analysis/HiggsAnalysis-CombinedLimit#1071) <#1071 (comment)> I'm not sure not sure this is just as easy to implement for my use-case actually. Eg I want to fit for peak position by interpolating between sth like np.random.normal(3, 1, 10000) and np.random.normal(6, 1, 10000). The functionality of shapeU is convenient because I can easily switch from when there's an outside measurement available and supply shape prior or when there isn't and supply shapeU. — Reply to this email directly, view it on GitHub <#1071 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMEVW6RDYAZ4L5TSP6FAOD2Y2TCXAVCNFSM6AAAAAB24FXIGGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDOOJUGYZTGNJXGU> . You are receiving this because you are subscribed to this thread.Message ID: <cms-analysis/HiggsAnalysis-CombinedLimit/issues/1071/2794633575@ github.com>