Cosmic Ray Pipe Calibration of 1kt

1
Cosmic Ray Pipe Calibration of 1kt
K2K Collaboration Meeting Jan. 2004

• New Data Taken in Summer 2003
• Improve Event Reconstruction
• Test of Particle ID (Jeremys Talk)
• Check of through-m basic plots (see Asias
  Talk)

• J. Argyriades, J. Griskevich, K. Hayashi,
• S. Mine, M. B. Smy, M. Tuchscher, J. Zalipska

2
Trigger Counters for Cosmic Ray Pipe
Michael Smy, UC Irvine
3
Summer 2003 New Data

• Improved (circular) trigger counters
• Took data at z4m to compare with entering muons
  (0 is center of tank and z is vertical, not beam
  direction)
• Improved statistics at z2m, z0m, and,
  especially, z-2m (sssllloooowww!!)

Spring 2004 More Data

• do z2m with high statistics
• If possible, do z1m, z-1m

Michael Smy, UC Irvine
4
(No Transcript)
5
Bonsai Improved Point Fit Algorithm for 1kt
Michael Smy, UC Irvine
6

• Timing
• Charge Pattern
• Particle ID

• Timing
• Charge Pattern

Auto-Fit
MS-Fit
Michael Smy, UC Irvine
7
First Step of Auto-Fit Fit PMT times assuming
all light comes from a single point (point fit)
Michael Smy, UC Irvine
8
Point Fit in Cherenkov Detectors

• Selection of hit PMTs
• Definition of goodness variable for a given point
  x based on relative timing alone, NOT charge
  pattern
• Search detector volume to get close to absolute
  maximum of goodness
• Fine-tune best-fit point for maximum goodness

Michael Smy, UC Irvine
9
Goodness Definition

• pfit defines the goodness and chooses t0
  as center of a sliding timing window
• Bonsai fits t0 with likelihood

Michael Smy, UC Irvine
10
Annealing AlgorithmsFind Absolute Minimum!

• Analogy with solidification
• Nature can find global (energy) minimum, if
  system cools slowly enough
• Usual implementation take sometimes random steps
  uphill (governed by a Boltzmann factor), then
  slowly reduce temperature
• See Numerical Recipes for more details

Michael Smy, UC Irvine
11
Branch OptimizationNavigating SuccessiveAnnealin
g Iterations
Trace all branches for a range of likelihoods!
A Search Tree
Michael Smy, UC Irvine
12
Bonsai Start Position Where to Grow?

• Hit Selection (High Charge Hits)
• Time Ordering
• Use Suitable Combination of Four each
  combination gives one/two vertex
• Use Best (Highest Likelihood) Vertex

Michael Smy, UC Irvine
13
Four-Hit Combinations

• Need to Solve
• or with
• Choose
• and convert to
• Solve

Michael Smy, UC Irvine
14
Bonsai Results
Michael Smy, UC Irvine
15
Further Improvements and Checks

• Fit neutrino data and MC with Bonsai check if
  auto-fit gives better results, if started from
  Bonsai-vertex
• Fit partially contained events, where systematic
  effect (longitudinal vertex pull) is largest
• Try position average for probability density
  function (possibly with more data from Spring!)
  and check for systematic position dependence
• Try to understand auto-fit better to incorporate
  charge pattern into Bonsai
• Maybe allow more early high-charge hits the
  larger the charge asymmetry is in a vertex, so
  that stopping/exit point hits dont pull the
  vertex

Michael Smy, UC Irvine
16
All hits except near exit/stopping point (3m
circle at bottom)
hits near exit/stopping point(1m circle at
bottom)
Michael Smy, UC Irvine
17
Conclusions

• Improved Cosmic Ray Pipe Hardware
• Took a lot of data in 2003
• Want to take more in 2004
• Data important for several calibrations /
  investigations of systematics
• Improved point fit vertex resolution and
  systematic pull

Michael Smy, UC Irvine