EM Energy scale and ECAL SC corrections
Overview:
- on the energy scale: in principle, it would be possible to tune the energy scale with a single number. We just need to put together studies from Zee with Wenu, pi0, photons etc. and decide for an energy reference
- the resolution of the peak seems a priori a much more sensitive value, but we must check this assumption
- we know that f(ET,eta) dominates the resolution in inclusive plots (all ECAL, or all EB, no stringent cuts)
- so we need to tune f(ET,eta) and in particular its eta dependence, to put the resolution down to a reasonable value, which could be of the order of 1 GeV according to our previous studies on the Z peak
- beforehand, we must have estimated the impact of cracks and other systematic parameters on the resolution
- once we'll have a global comprehension of the various effects impacting the energy scale, once the latest calibration constants will be integrated for all sources of EM objects (photons don't currently have the latest laser corrections implemented), and taken into account the systematics on the resolution, we'll be ready to deliver a DP plot of the Zee peak.
Deliverables: (timescale: January 2011)
- Energy scale in ECAL with Z→ee events after latest calibration and laser corrections are applied, estimation of systematic effects
- Extraction of new f(ET, η) correction factors for a minimal Z peak resolution
- DP validated Zee invariant mass figure
Energy scale with Z→ee events
We compare ee invariant mass distributions in latest data and Monte Carlo: we estimate the shift in the peak position, and measure the influence of various systematic effects on the peak position and width, as well as on eventual tails in the distribution.
Status
Systematic studies
Effect of the selection criteria
- ET
- η
- brem, r9, σiηiη
- trigger requirements
- H/E ratio
- matchings (to GsfElectrons, to tracks?)
- isolation
See summary on this page.
Effect of the inter-module and inter-supermodule cracks
Effect of the fitting procedure
- Fit function
- Breit-Wigner and Crystal Ball FFT convolution
- Cruijff function (bifurcated Crystal Ball)
- Double gaussian
- Fit range (Vladlen)
See Vladlen's talk from October 20th
Other possible effects
SC Energy corrections
Method to determine SC corrections from MC
Testing the SC corrections from data
- Apply first-step correction to SCs: f(brem) + C(η) if in barrel
- Fit the Z invariant mass peak
with a crystal ball function, assuming linear background. We must move to FFT convolution of a BW and CB as for the energy scale studies.
- Get the chi2 to measure how well the reconstructed Z peak agrees with the nominal value of the Z° mass
- For xcheck, draw 2D plot of peak position vs eta of the 2 leading ET SCs.
Specific pending questions/issues
- What are the dominant dependencies of f(Et, η)? ET, η, brem
- How off is the energy scale as a function of η?
- Is the statistics sufficient to rederive the atan shape of the correction?
- What strategy for the EE?
- How much luminosity is needed to achieve full parameterization of f(ET, η)?
- Test fEtEta from E/p with Zee
- Update f(brem) with 38X MC
- What would be a f(brem) correction derived from data?
Documentation
Internal note
EGM-10-003
Progress on derive new energy corrections
See this link for optimization studies on Z->ee data
See here for actual status of energy corrections derived from dielectron Monte Carlo
Tools
- Z→ee sample:
see this link for information about data skim with latest laser corrections and alignment constants, produced by Riccardo
- Ntuples: flat ntuples for both Zee and energy correction studies are produced with this code
- Analysis code: both Zee studies and energy corrections are done using this package
(bulk) To-do-list
- Update results with 35 pb-1
- Check performance macro
- Change fit function for f(ET,η) october exercise to achieve smallest and most meaningful errors
- Internal note: scope, update with new results.
Topic revision: r13 - 2012-02-06
- unknown