Choose timezone
Your profile timezone:
DIRT2 Meeting
→
Europe/Zurich
Description
https://umn.zoom.us/j/92623331343?pwd=MWNkaXJadHNTbmZkL3RKYXdEK2RwZz09
Thanks Adam for the good work !
Action items
Rik+Ben for sure will meet with Laura and Stephen about using nusystematics.
Rik will reach out to Andy Furmanski for an orientation to one of the Spectral Function codepaths in GENIE and loop Laura and Stephen in.
Stephen and Rik will look at the GENIE LFG and see how and why Emissing and Pmissing don’t have the expected coverage of phase space like other LFG do. Is it an easy mod to get that feature back ?
First part of Laura’s talk focused on initial state options, focused on coverage of the 2D plot Emiss and Pmiss.
NuWro has a servicable LFG and SF. But to switch to NuWro, we lose the other reweighting features in GENIE. And NuWro has problems with other aspects of its model, such as the SIS region. This is probably a bad choice. (Can we run NuWro for QE only and splice it in Frankenmodel style?)
The GENIE Valencia and Genie 1802a RFG have a delta function fixed removal energy and only has variation in P_miss. We would only be able to make a reweight in Pmiss. Stephen thinks GENIE2 and non-QE in Genie3 have the parabolic delta-function RFG shape in this space.
The CRPA+SuSAv2 code in GENIE has some desirable properties and looks like a RFG with a lot of variation rather than a delta function. But it has some shortcomings that probably take it out of contention.
If we had to live with the GENIE, but wanted to access this type of physics, we would have to resort to shifting true and reco energies after reco and recalculating any other quantities that depend on them, which is a bookkeeping issue. There is an attempt along these lines in nusystematics. This problem is mostly QE only, missing reco energy in other channels is larger from other effects like neutrons. So we can go with a QE only solution.
Action items are to look into patching up GENIE to deliver the variations we want. We need to commit to something. Especially the SF, we should reach out to Andy and find out what state the code is in at this moment.
Do we have enough experience in the group with SF, or would inexperience be a barrier ? Laura says T2K folks have a lot of experience with it.
Stephen’s experience with CRPA+SuSA means he knows the QE framework in GENIE3 and thinks it might be a short hack to put in variations in Emiss that look like the expected LFG. Stephen and Rik might take stock of that together.
Second half of the talk.
Implement an ad-hoc energy dependent bias in Emiss (including neutron energy in this case), which directly connects to the oscillation spectra. Running all the models shows how their ground state and FSI assumptions are different. Running in energy bins shows more deviation from whichever one serves as a baseline. For the QE neutrino mode, the two bits of physics are removal energy (first half of the talk) and FSI which creates neutrons. For other channels, the effect expands to all the different neutron mechanisms. Capture this as a model-spread uncertainty in Enu *relative* to a fixed model at a fixed energy. Importantly, this way allows knobs that change the missing energy (like FSI knobs for example) to be independently dialed up and down.
That proposal for a knob seems like a good one in the queue. We are starting to build such a queue, and in a few months will probably have them organized in priority order, based on easy vs hard and high payoff vs low. This looks like a hard high-payoff one.
We should start thinking about how we are going to handhold our customers with these new knobs. Our customers are LBL folks doing fits, plus ND folks who are going to start doing parameter constraint type things. Management thinks this is a more important part of DIRT2 than it was for DIRT1.
There are minutes attached to this event.
Show them.