Missing q2 and MCORRERR functors

We are missing the q2 and MCORERR functors for SL. For q2 there needs to be one for each solution. I think there are the errors as well to be implemented.

I shall implement them.

added ThOr label

assigned to @masmith

changed the description

Hello @masmith, feel encouraged to update the task https://gitlab.cern.ch/lhcb-dpa/project/-/issues/61 as you see fit as many of us are using this task to follow the developments of new functors, ensuring also there is no duplication of work. Advance thanks.

If I remember correctly there is a MASSHYPOTHESES functor which can be used to get q2. We indeed need mcorr err.

ah thanks, I'll take a look at that

@amathad Where is the functor code located? found it

Hi @masmith , Lets not use F.MASSHYPOTHESIS there are some issues with that.

Actually there used to be mass squared functor at some point but it doesn’t seem to exist anymore. This is needed instead of F.MASS functor below in case of negative q2 values and we do not want to take sqrt. If we implement this new functor F.MASS_SQ, we are in business with q2. 



For fully reconstructed to calculate sqrt(q2) one can do: 


CHILD_1_FOURMOM = F.CHILD(1, F.FOURMOMENTUM)
CHILD_2_FOURMOM= F.CHILD(2, F.FOURMOMENTUM)
QSQ = F.MASS_SQ @ (CHILD_1_FOURMOM + CHILD_2_FOURMOM)

For semi-muonic decays with neutrino one can do:



MOTHER_FOURMOM= F.FOURMOMENTUM
QSQ = F.MASS_SQ @ (MOTHER_FOURMOM - CHILD_1_FOURMOM - CHILD_2_FOURMOM)

The F.MASS is declared on python side here and the actual implementation of it is here. Should be straight forward to implement F.MASS_SQ. You might be asked to add a test in which case you need to add it here. Feel free to ping me on Mattermost if you run into problems.

I am working on the MCORRER first.

Is it possible to declare a new four-vector from the components. I have:

      auto const dMcdp = -1 * invSqrtMassPT * A / B * flight + 1 / pt * (mom - A / B * flight);
      auto const dMcdE = invSqrtMassPT * E ;

where flight, and so dMcdp is a three vector. I would like to combine dMcdp and dMcdE into a single 4-vector to multiply it with the four-momentum covariance matrix without having to write out each element. Is that possible?

Wonder if you could do something like this i.e build SMatrix.

Thanks, that looks promising.

mentioned in merge request Rec!3111 (merged)