Error propagation with root


I’d like to know if is there any ROOT function or class to propagate errors of derived quantities.
I’m talking about this formula:

Schermata 2020-08-11 alle 11.21.15

where F(x,y,z) is a function of x,y,z obtained through direct measurements.


Hi Riccardo,

In principle that’s best covered by modeling and symbolic calculus, and that in turn brings us to RooFit. @StephanH is that something that’s possible with RooFit?

Cheers, Axel.

No, RooFit doesn’t do this as far as I know. It uses numeric derivatives to do fits.

Unless @moneta knows another method, you can write down the formula 10.2 using the RichardsonDerivator. Make one derivator for each partial derivative, and then sum them in squares as usual.