Choose timezone
Your profile timezone:
NIUWG/DIRT2
→
Europe/Zurich
Not in chronological order, the notetaker is a bit of a time traveler.
Action item: Abi will look at choices that make bins land on non-human-readable boundaries, and make minor adjustments.
Action item: Rik and the UMD summer undergraduate research crew take this and show what Delta systematics do in this space. (Tom Holvey should try? Or is he thesis writing?)
Action item: what part of this belongs at the Collaboration meeting in two weeks ?
(Action item: several times we mention a meeting with Callum Cox who might adopt this sooner rather than later. There is a meeting week from today with a few experts on the SIS and DIS knobs in GENIE.)
Abi’s samples are from NuWro with stats of about 1/10 of one year of operations. (It was 10M but one year of NDLAr will be 100M)
Some axis labels are not quite right, but we were quickly able to re-orient ourselves to them. Abi has some 1D plots she might dump into a file if she can do it efficiently enough. For experts at the meeting, this set of 2D plots was digestible, but for others in DUNE or for exploratory work on microtopics we might need the 1D plots to efficiently make sense of things.
We thought this had all the features we hoped it would. Talked through the oscillation max region. Talked through the evolution of QE and Delta samples vs pT. Observed the finer binning at oscillation max and QE peak.
Abi’s demonstration of the oscillation max region was comforting, this binning will constrain it explicitly. Stephen notes that in the limit of tiny bins, Enu_reco and the MINERvA 3D are just a transformation from each other, so it doesn’t matter. Rik is advocating that we go with the one that allows the ND to best express the Interaction physics (and detector systematics) that need to be constrained, rather than the oscillation signal.
Callum W. and Richie point out NDLAr has severe acceptance limitations. Callum shares a plot via slack that illustrates the acceptance carves a hole that has impacts as low as Q2 = 1. Richie says NDLAr measures x transverse much better than y transverse. Even before real Reco samples, the NIUWG working with generator/NUISANCE output probably want to code up an approximate acceptance mask.
We will want to build in the ability to switch between Available Energy and Enu_reco, just in case.
Stephen notes we will want the ability to merge a handful of bins. Even our new systematics knobs may fail to cover some bins, leading to unworkable chi2s and fit misbehavior. Rik suggests one place this might be true (despite our best efforts) are the first two or three Eavail bins. Better that we do this for known physics shortcomings rather than ignorance.
In general, we can anticipate needing a rebinning layer from the rectangular binning Abi is making and some effective binning that re-aggregates Abi’s bins for low stats and/or minimal physics impact. Laura mentions a root facility to do this, Rik mentions that Dan Ruterbories implemented something equivalent on MINERvA.
Do we want to process in 2x finer bins anticipating an immediate coarsening of the bins ? Can go one way but not the other. Not sure, we’re not practiced with what we have from Abi already.
Twice, Anne and Rik express that the DIS region is not as feature-rich as the QE-2p2h-Delta region. That means fewer binnings are needed. Just enough to give some sensitivity to DIS shape and detector systematic effects. Are side conversation also touches on how high in W is this sample really ? W < 3 as Rik remembers ? Higher ? Disconnected, but this secretly refers back to Callum’s comment about acceptance limitations. It will fall to us (Anne and Rik are at the meeting) to figure out what DIS + SIS constraint might be missed because of these limitations.
Not able to get someone from Mach3 to attend this time. This is 1000 bins or so. Is that ok ? We think so. We think other aspects of the systematics are harder for Mach3 than the bin counts. Not sure.
There are minutes attached to this event.
Show them.