Blind Analyses - PowerPoint PPT Presentation

1 / 4
About This Presentation
Title:

Blind Analyses

Description:

Other analyses, near detector physics, cosmic neutrinos, etc may require ... I see no reason for messing up our events, for example by distorting energies as ... – PowerPoint PPT presentation

Number of Views:39
Avg rating:3.0/5.0
Slides: 5
Provided by: litch2
Category:

less

Transcript and Presenter's Notes

Title: Blind Analyses


1
Blind Analyses
  • Principles
  • Simple
  • No unnecessary blindness
  • I consider only neutrino oscillation analyses.
    Other analyses, near detector physics, cosmic
    neutrinos, etc may require differences
  • Four data sets
  • Near detector
  • Cosmic ray set, selected by a beam timing
    requirement
  • Long event set, selected in preliminary
    reconstruction to be longer than a constant to be
    determined
  • Short event set, the rest of the events
  • Only the short event set is blinded, initially

2
  • The near detector data does not need blinding, it
    is used to calibrate the beam and the detector
    function. This analysis however should be done
    independently of the far detector data.
  • The cosmic ray muons are used to give the energy
    calibration of the detectors. They do not need
    to be blinded but again the calibration should be
    independent of the data analysis.
  • The long events are essentially unambiguous ? cc
    events. I contend that they should not be
    blinded
  • We need a first look at real data (i.e. the
    oscillation dip) to determine our beam strategy.
    We will look silly if Super-K is wrong and ?m2 is
    high (6-7x10-3, the Soudan 2 and Kamioka
    prefered values) and we have been running for
    years in the low energy beam
  • Having seen the dip the important questions are
    its position and shape. These are predominantly
    determined by the energy calibration, which is
    blind to the data and secondly by the
    contribution of the short events which are
    blinded.

3
  • Since the long events are almost entirely ? cc
    there are no significant selection problems which
    need blinding
  • Once the dip is observed one can choose an
    energy region away from the dip and use this to
    tune the detector MC and beam extrapolation,
    independently of the physics, maybe using
    different beam energies.
  • Should there be a surprise and the initial data
    not be consistent with oscillations, all bets are
    off and we will have to reconsider the analyses
    but the short events are still blinded.
  • The short events are where the discovery physics
    lies
  • Electron cc events for ?13
  • NC events for sterile neutrinos (plus extra
    constraints on the oscillation parameters, T
    test)
  • Low energy ? cc events for observing the rise
    below the dip

4
  • In each case the physics depends, at least to
    first order, on the absolute rate of these events
    compared to the expectation from the beam and the
    high energy ? cc events
  • This can be blinded according to Garys
    suggestion of withholding an unknown fraction of
    these events, while tuning the analysis on the
    remainder
  • Do we need more complicated distortions of
    events?
  • I see no reason for messing up our events, for
    example by distorting energies as has been
    suggested
  • The energy calibration is already blind to the
    data
  • Assuming oscillations are the answer I see no
    reason for (unconsciously) preferring one ?m2 or
    sin22? over another.
  • Atmospheric neutrinos are a special problem. I
    dont see any clear way of blinding an analysis.
    One could use a similar long/short separation but
    the background problems are much more severe and
    predominantly occur in the large majority of
    short events. No previous atmospheric analysis
    has been blind. Fortunately with a beam time cut
    the atmospheric analysis can essentially be
    carried out independently of the beam analysis
Write a Comment
User Comments (0)
About PowerShow.com