Data Analysis and Detector Characterization

1
Data Analysis and Detector Characterization

• Paul Lebrun
• Auger/Fermilab/CD

2
Outline

• Intro from our physics program to detector
  studies.
• Infrastructure for Auger data analysis at
  Fermilab.
• Data Analysis
• Composition hadronic physics
• Anisotropies
• Detector studies.
• SD reconstruction systematics
• SD PMT and PMT electronics studies.

3
Introduction.

• The list of data analysis topics maps onto the
  science of Auger, with a particular focus on the
  intersection between the Cosmic Frontier and the
  Energy Frontier, namely, the issues of mass
  composition, leading us to hadronic physics at or
  well above the LHC energy.
• Our Holy Grail Composition hadronic
  interactions in the one EeV -gt 100 EeV range.
  Anisotropy an additional handle for determining
  the composition.
• While we are statistics limited above the GZK,
  systematics also matter ? back to detector
  characterization.

4
Infrastructure for Auger data analysis at
Fermilab.

• Hardware (modest!)
• One 4-core data server 10 Tb of data on
  networked disk.
• Fermi-Grid customers (small impact at Fermilab)
• Workstations for 5 physicists
• Data serving
• Non-event databases about 10 active MySQL
  databases, collecting data from experts servers
  and making it available as need be.
• U.S. event data repository.

5
Auger Offline Analysis, Software Methods

• All Offline analysis package installed and
  running. Thanks to the effort of the Offline
  group (Northeastern Univ. (U.S.), UNAM
  (Mexico)), the installation is now easy and does
  not take much effort.
• Reconstructing the entire SD data set take few
  days, allowing complete and systematic evaluation
  of detector performance and effective resolution.
  
• Full access of hadronic interaction Monte-Carlo
  expertise, as Eun-Joo Ahn is the current lead
  developer for Sibyll. These packages are running
  on the Grid (Open Fermi)
• Multiple groups works on the same topic, either
  using the same base software, or independent
  reconstruction packages.

6
Hadronic Studies Composition

• Rigorous analysis of the slanted depth (Xmax)
  data
• Using existing models, study mixed composition.
• Assuming proton, (or proton/Iron mix), extract
  self-consistent p-air cross-section vs energy.
• New (and unique) analysis technique Optimum use
  of Monte Carlo samples via the use of weights.
  That is, a small change in cross-section in a
  limited energy range is handled consistently via
  a change of these weights, not by recreating a
  completely new Monte Carlo data set.

7
Hadronic Physics, Results

• These analysis do not support a proton to Fe
  transition, from one EeV to 15 EeV range. Other
  nuclei are involved, or hadronic model needs to
  be revisited.
• If proton only, the p-air cross section rises
  with energy faster than most models predicts. Or,
  again, the hadronic model needs to be modified,
  at energies 10 times higher than the LHC can
  produce.

8
Composition with X-sectionextrapolated from low
energy data.
9
Anisotropies Composition Examples

• Point like source at 1 EeV, on the Galactic
  plane. Given known Galactic magnetic fields then
  this ray must be neutral. Based on shower
  profile and lateral distributions one can
  determine that such rays are not photons gt must
  be a neutron.
• Activity at Fermilab (2 years ago), check of the
  Galactic center prescription).
• Consider an extra-galactic source whose direction
  is away from the Galactic plane thereby
  influenced only by the local (few kpc) magnetic
  field. This leads to the possibility to
  determine the mass composition at the highest
  energy using the local region of the Galaxy as a
  mass spectrometer.

10
Anisotropies Mass CompositionThe Lemoine
Waxman thesis

• Stringent limit on composition at 55 EeV, should
  the Centaurus A be confirmed.
• Discussed previously by P. Kasper.
• Credits Fred Kuehn and Ivonne Albuquerque
  spearheaded this analysis effort and became a
  lead co-author

11
AGN Anisotropy Signal
12
Trans-GZK anisotropy at Fermilab

• Check (i.e., recalculation from raw data) of this
  anisotropy. Includes the SD reconstruction.
• In depth analysis of the confidence level
  associated to the AGN signal. Investigate
  stability vs time.
• Includes a closer look at possible detector
  systematics In-depth study of the angular and
  energy resolution.

13
2-Point Correlation vs time.

• A 2Pt correlation analysis result was presented
  at the Rencontre de Blois, 2009 meeting,
  showing a compelling signal.
• Current (2004 -gt 2010 data) probability for
  isotropy is 1 gt only upper limit. (Recent
  publication..)
• We checked the stability of the signal, and it
  follows (time-wise) the same evolution as the AGN
  signal.

14
Global Likelihood optimization

• Motivation So far our published analysis of the
  AGN correlation
• Is based on event counting above a fixed energy.
• Assumes that the angular and energy
  uncertainties have a negligible impact of the
  significance of the result.
• and does not use models of the Galactic magnetic
  fields.

15
In our more precise likelihood analysis, the AGN
and 2pt analysis are no longer based on a simple
binomial test For each event, one has to compute
a probability for correlation instead of sorting
the sample into correlated vs not correlated
sub-samples. Uncertainties from SD
reconstruction propagated into the physics
analysis
16
Status of these analysis

• At Fermilab In progress. Log likelihood
  function coded, based on simple extra-galactic
  propagator. Parallelism implemented based on MPI.
  Needed Galactic field Runge-Kutta propagator.
  
• At NYU (G. Farrar et al), .... Study of
  Galactic maps from radio astronomy data, and used
  in the AGN analysis.

17
Outline

• Intro From our physics program to detector
  studies.
• Infrastructure for Auger data analysis at
  Fermilab.
• Data analysis
• Composition hadronic physics
• Anisotropies
• Detector studies.
• SD reconstruction systematics
• SD PMT and PMT electronics studies at Lab 2.

18
From Physics Analysis to Detector
Characterization

• The three measured quantities that enters the
  anisotropy analysis are
• Time of arrival of EAS gt know from GPS data
  with much more accuracy than needed.
• Direction of arrival, obtained via timing.
• Energy of the shower anisotropy are expected to
  depend on the energy of the ray.
• Two of them are obtained indirectly. One needs
  calibration reconstruction software.

19
In order to determine the systematic
uncertainties at each step of this process, the
SD reconstruction has been done many times, to
refine the knowledge of the time of arrival
variance, integrated pulses (station signals) and
so forth. Cross-checks Quality of the EAS fits
(angles and energy) In the context of small
changes in the detector response.
20
Surface Detector Long Term Performance

• Our SD monitoring data (single muons) indicate a
  small (few /year) reduction of the pulse length,
  due to a commensurate change in the light
  collection efficiency.
• We can run for a few more years without losing
  sensitivity, as the trigger parameters can be
  re-adjusted.
• Time integrated signal calibrated in-situ. To
  first order, the calibration procedure is robust
  against changes in light collection efficiency.

21
Angular Resolution Results

• Variance of the EAS arrival time directly
  measured with special, close proximity (10 m.)
  proximity tanks.
• Angular resolution is better than 0.9 degrees,
  based on low energy data and does improve at
  higher energy.
• Adequate for AGN signal. But contribute to the
  final uncertainties on the parameters of the
  correlation.

22
SD Energy Method for determination

• SD Energy resolution/calibration also done based
  on data.
• Time integrated pulse converted to Vertical
  Equivalent Muons (VEM) -gt station signal.
  Based on in-situ calibration, from single muon
  data.
• From VEM to SD estimate of the energy, via the
  lateral distribution fit, with correction for
  zenith angle.
• Calibrated with hybrid data.
• These last two step have been checked, as they
  are done offline, based on raw data.

23
SD Energy Scale Resolution studies.

• Conservative estimate of the station signal
  variance, estimated for the entire array, using
  information from the monitoring data .
• Study of the EAS lateral distribution fits (LDF)
  in Offline reconstruction.
• Study of the combined lateral curvature fits.
• In presence of the known SD PMT saturation
  features
• PMT base
• FADC
• Studied at Fermilab.

24
Findings Impact on Physics (Spectrum,
Anisotropy)

• At the highest energies, the angular resolution
  ranges from 0.5 to 0.9 degrees which is adequate
  for anisotropy studies, for most conventional
  UHECR scenarios.
• The energy resolution (13 to 15 above 10 EeV)
  and the stability of the energy scale (20 at 10
  EeV) is a contributing factor to the flux and
  signal to noise ratio uncertainties at or above
  57 EeV. Note that we have no direct calibration
  (from the FD) at these energies.

25
SD Energy Scale and Resolution Stability

• However, the energy spectrum is stable. The
  event count above 55 EeV vs time is proportional
  to the exposure (estimated from monitoring data
  and known with high accuracy).
• This constrains the change(s) in the energy scale
  and resolution. Within hybrid statistical
  accuracy (more data will help!.)
• These uncertainties are being implemented in the
  AGN likelihood analysis. A more plausible
  estimate of the AGN correlation significance will
  be extracted.
• Status in progress...

26
Conclusions

• We are focusing on particle physics at the
  highest energy X-section, Hadronic models and
  composition.
• Trans-GZK anisotropy (AGN 2Pt), prospects
• Auger has the real potential to open the field of
  UHECR astronomy. Important for particle physics
  at the highest energies.
• More work is needed to assess all systematic
  uncertainties that enter the final estimates of
  the parameters of these anisotropy signals.