MENU

Please re-load this page.

This page describes the EHE neutrino search with the 9-string IceCube detector on 2006 sample prior to accessing EHE signal part of sample.
Our target energy range is any neutrinos above 10^8 GeV. Cut are optimized for a GZK model. Our target sensitivity is to find GZK neutrinos within several years and competitive to the other EHE neutrino searches around world.
Relevant performance study has been performed prior to this analysis for the 80 string IceCube detector and described in the proceedings. Please find physics behind this EHE analysis strategy in this proceedings as the basic strategy is unchanged for 9 string analysis.
The point we need to address in the EHE neutrino search is that we do not have good understanding about the major contribution from atmospheric muon background in UHE/EHE regime, although still we do know for sure that the atmospheric muon distribution has much softer and steeper distribution than that of signal (gzk) neutrino (at both surface and deep in ice) and are able to reject background well if we could cut out lower energy events than signal energy domain. Thus, an empirical model construction using high enough energy range but lower than signal range is required.

EHE Sample

Data used for this analysis is level-1 filtering EHE sample which has filtering condition of number of channels larger than 53 or 80. The documentation of the Level-0 and Level-1 processing can be found here. Offline cut of Nchannel larger than 80 is applied to all data used in this analysis. Data is corrected from June 2, 2006 to Nov 20, 2006. Bad files listed here are excluded (plot). Before cleaning 1654 exist and after file cleaning 1260 files are left. List of files used in this analysis also found here. Also Carsten Rott has made a page on EHE sample verification. Removed are mostly the files which has length less than 300 sec and do not much contribute to livetime and so called "weasel" period. Livetime for current sample is 125.0 days.
The most difficult thing encountered in 2006 real data study is that the number of bins were reduced from what is designed/planned. In 2006, we have 128 bins of channel-0 (gain 16) ATWD which corresponds to 400ns but only 32 bins for channel 1 (gain 2) and 2 (gain 0.25) which corresponds to 100 ns (A combined waveforms look like this) and 50 bins read out for FADC of 1 micro second.

Real Event Distributions

• EHE 9 string data - initial look at basic variable
• EHE 9 string data - other basic variable for different high NPE sets of events (first-guess cos theta, azimuth, center of brightness, IceTop Nch, IceTop Nch vs. firstguess cos theta, in-ice N channel, run ID vs. event ID)
• EHE 9 string data - IceTop activity
• Baseline estimation strategy with SC sample - SC event baseline study for EHE analysis
• EHE 9 string data - with EHE optimized baseline option
• EHE 9 string data - with Channel 0 only (kind of hit cleaning) and data clean-up

MC sample

Detailed Monte Carlo simulation is performed using IceTray modules (relevant modules are included in the EHE meta-project) based on the simulation meta-project v01-09-06a. Droop is simulated for FADC and bin number is reduced for both ATWD and FADC. Muon and taus are injected at 860m from the center of IceCube 23 string. 400K of muon which follows E-1 spectra from 100TeV to 100EeV, 400k of muon with E-2 and 200K of taus with E-1 spectra in the same energy range are generated.

EHE MC Distributions

• Initial look at 9 string MC sample
• MC truth Energy and NPE relation. NPE is obtained from FADC or ATWD whichever gives larger value for the channel. The event-sum NPE distribution was unexceptionally low compared to previous studies because of the number of bin reduction
• MC truth cos theta and linefit cos theta and its resolution at given linefit velocity direction
• Studies on general properties of EHE sample
• Energy loss comparison between MMC and JULIeT
  In the plot of dE/dx as function of in-ice muon energy, blue mark with "Keiichi's new function" indicates Juliet's averaged energy loss behavior and black mark with "Total" is that of MMC. Agreement is pretty good in the energy region of interest and at higher energy some deviation comes in because of difference in photo-nuclear interaction model used.

Signal and Background Simulation

In this study, JULIeT is used as the event simulator near the detector as well as propagator. The weighting schema for the JULIeT can be found in this internal report (pdf). Generated E-1 and E-2 samples are weighed by signal sample of a propagated GZK neutrino distributions in ice and the same with an empirical atmospheric muon model which we described in the following.

We have developed a relation between primary cosmic-ray energy and muon bundle sum energy (pdf), and this is convoluted with measured cosmic-ray fluxes. This indicates that possible muon bundles in ice is represented by single muons of which energy corresponds to the sum of muons in a bundle. A toy simulation shows this is good approximation at this EHE energy. Further explanation of the empirical atmospheric muon (bundle) model can be found here.
An example how flux changes with constant Ethres with varying alpha and similary constant alpha varying Ethres.

We have also checked corsika sample.

In addition to the model, we extracted fluctuation size from corsika sample (proton primary which gives a larger fluctuation value than from heavy primaries) in the relation between in-ice muon bundle energy and primary CR energy. Here one can find fluctuation sizes and corresponding fluctuation fig used in the model.

Plotted here are chi2(/dof) from simulated event NPE distributions weighed by model function using dE/dx ~ -beta(E)*E approximation and real data NPE distribution on the alpha and Ethres plane, in which above fluctuation effects are included. Chi2/ndof from fit without fluctuation effect old alpha and Ethres plane.

One can see a chi2 valley on the plane, in the plane, agreements between data and model are nearly equally good. There is not much difference in chi2 in the sets of the numbers of alpha and Ethres in the valley.
Initially we choose two extreme cases to estimate uncertainty in the model (shown below as old comparison without fluctuation).

Finally this model with two sets of parameters (note that if GZK cut off we are almost background free! and our main charm/bottom contribution is from the CR energy equal or above 10^20 GeV if any) is compared with the 9 string IceCube EHE real sample NPE distributions in the different linefit cos-theta range. Black points are real data and red histograms are obtained atmospheric muon (bundle) model (blue and light blue lines are from GZK model for reference) distributions in both plots.

Comparisons between data and simulation with fluctuation effect

alpha = 2.04, Ethres = 3730 GeV

• NPE
  
• cos theta
  
• NPE(focus)
  

Old Comparisons between data and simulation without fluctuation effect

for two sets of alpha and E_thres

NPE
Set #1 - alpha = 1.9, Ethres = 300 GeV
upper left full cos theta_firstguess, upper right cos theta_firstguess > 0.6,
lower left 0.4 < cos theta_firstguess < 0.6, lower right cos theta_firstguess < 0.4
Set #2 - alpha = 2.0, Ethres = 1.5 TeV
upper left full cos theta_firstguess, upper right cos theta_firstguess > 0.6,
lower left 0.4 < cos theta_firstguess < 0.6, lower right cos theta_firstguess < 0.4
cos theta line-fit Set #1 - alpha = 1.9, Ethres = 300 GeV upper left all logNPE>4.0, upper right 4.0 lower left 4.1 < logNPE < 4.2, logNPE > 4.3
Set #2 - alpha = 2.0, Ethres = 1.5 TeV upper left all logNPE>4.0, upper right 4.0 lower left 4.1 < logNPE < 4.2, logNPE > 4.3


Background simulations: Comparsion with CORSIKA

CORSIKA Study Page

Cut optimization

Here, we obtain GZK and atmospheric muon background distributions on NPE and linefit cos-theta space.
Cut optimization is done by changing signal regions in this space.
In this space, 13 cut lines are drawn (0 to 13). In the each space, the right part of lines are considered to be signal region and the left hand part is rejected.

Passing rate (integrated event number passed the cut) from these 13 cuts are shown for signal and background. We consider cut #6 is well optimized for our purpose, considering a possible NPE - Energy correlation systematics of a factor of ~2 as discussed below.
Alghough in original proposal we requested even harder cut but it must be too tight. We can see that the event distribution after cut #6, #8 corresponds to EHE CRs by plotting event rates as a function of CR energy with cut #6 and event rates as a function of CR energy with cut #8.
Please note that our preffered cut #6 here corresponding to cut level 2 in the ICRC proceedings.

Effective area and sensitivity

With cut #8, muon effective area can be found here (in [km2]) as a function of mc-truth cos theta and in-ice muon energy. Also Muon effective area with cut #6 can be found.
With cut #6, NEUTRINO effective area can be found here (in [km2]) as a function of surface neutrino energy. Also with the same cut #6, contribution from electron neutrino can be found here (in [km2]) as a function of surface neutrino energy. Also for example with cut #6, event rate as a function of GZK neutrino from muon channel event rate as a function of GZK neutrino from muon channel shows that our main signal in-ice muon energy around 10^8 and above which corresponds to approximately surface energy of 10^9 GeV and above. This would be different from previous UHE analysis with AMANDA.
Sensitivity plot with this cut #4, #6, #8 and #10.


On the other considerations

We relies on the simulation especially on the part of NPE - energy correlation. It is important to understand and estimate uncertainty in the number of detected and simulated NPEs. Current simulation, standard candle and golden-dom study indicates that there is about a factor of 2 difference. This study will be continued as this is not only relevant for EHE analysis but also any other analysis.

• NPE distribution comparison with SC simulation (photonics) 1 - ROMEO and Photonics extension with Golden DOMs
• Real data SC detailed NPE distributions - extracted NPE distributions from SC sample

Review iterations

Ordered by date

• initial question from Gary
  
  • reply from Aya
• Analysis call minutes from Gary
  • reply and review request from Aya
• comments from Marek
• reply from Aya
• reply from Marek
• reply from Aya
• reply from Marek
• comments from Spencer
  • reply from Aya 1
  • reply from Aya 2
• question from Paolo
  
  • reply from Aya

FAQ

The BASIS on the EHE 9 string analysis

• Unblinding Proposal on the EHE analysis of the 9-strings data ( version 2006-9-7 PDF )
• The documentation of the online filtering process
• The documentation of the Level0 and Level1 processing
• The python script to generate the EHE level1 data file By Dusan/Phil (July 11 2006)
• Proposal on the EHE working data sample of the 9-strings data ( version 2006-5-22 PDF )

The other useful Information

Special Devices
Golden DOM (Chiba U.)
Standard Candle (Andres Morey)
Real Data
Verification Team
Chris's archive (Jan 06)
South Pole Logs
Mark's Page