Analysis / Verification / Simulation

IceCube Acronym Dictionary

LBNL has responsibilities in many aspects of IceCube offline analysis. We have physics interests in both electron and muon reconstruction, and have developed algorithms which are used for both of these channels. We have major responsibilities for hardware calibration and verification, mostly involving timing and amplitude calibrations. Here, we have used both muons and LED flasher data to verify the detector calibrations. We have also written the portions of the offline software and simulations that involve the electronics and calibrations, and that turn phototube signal waveforms into photon arrival times.

Neutrino Searches

Although there are 500,000 times as many downward going muons from cosmic ray air showers than upward going muons, with careful attention to background rejection, it is possible to find neutrino candidates with only a single string of DOMs. The figure (below) shows one of two strong candidates found in the 2005 data.


As the picture shows, the event follows the string for about 830 meters, hitting 49 DOMs.  The plot on the right shows the time residuals of each individual hit; most of the first hits on each DOM are direct hits, with residuals ~<20 nsec.  The muon track is reconstructed about 1 degree from vertically upwards; the geometry requires that a track hitting 50 DOMs on one string must be nearly vertical.


Cascade reconstruction

The following figure shows a simulated cascade:


One algorithm exploits the fact that in the ice the scattering length is much shorter than the absorption length; the width of the photon arrival time distribution at the digital optical module (DOM) therefore depends on the amount of light scattering and by extension the distance from the cascade. Signals at each DOM can be parameterized into a width, arrival time of first photon, and number of photons. The position of the cascade can be determined from the widths alone, independent of cascade energy. The 1st photon arrival times and number of photons then can be used for a large number of cross-checks. This redundancy should provide a robust algorithm that will aid in background (i.e. muon) rejection. Preliminary results in development of this idea can be found in this talk given at the September 2005 collaboration meeting in London.

The other algorithm uses a maximum-likelihood based approach that uses the arrival time of each photon at each DOM. The photon arrival times are determined via a FeatureExtractor. The algorithm folds in the photon arrival time probability, using the measured depth-dependent ice properties and the DOM response. It is discussed Talk presented at September 2005 London collaboration meeting. This figure shows how the FeatureExtractor deconvolutes a moderately complex ATWD waveform into 6 individual photon waveforms, determining their arrival times within a few nanoseconds.

Muon analyses

Muons can be used to determine the arrival directions of muon neutrinos far more accurately than is possible for electron neutrinos. This figure shows an early (Feb. 7, 2005) reconstructed muon from an air shower array. The shower is reconstructed at the surface, and a muon is seen in the InIce detector. After hitting the surface, the muon takes over 7 microseconds to reach the bottom DOM. The direction of the muon agrees with that of the air shower.

On the other hand, the backgrounds to muon neutrinos are much larger. Our initial physics goal is to use muon neutrinos to look for point sources of neutrinos. Gamma-ray bursters (GRBs) are an attractive first target of study.

The muon analysis uses a maximum likelihood based reconstruction algorithm similar to the second cascade analysis. The muon angular distributions, number of hit DOMs, etc. have been studied in detail, and are consistent with expectations. This is discussed in IceCube String 21 reconstruction from September 2005 London collaboration meeting.

Detector calibration, verification, and performance

We are resonsible for the pieces of the detector calibration and simulation that are connected to the electronics. The calibrations are determined from laboratory measurments (data taken at the Dark Freezer Lab in Madison, and also in-situ calibrations involving internal sources. These calibrations are then verified using muons and LED flasher data. We have demonstrated that the timing calibration is consistent to within 2 nsec across all of String 21 plus the four deployed IceTop tanks.

This histogram shows the timing resolution of all 60 DOMs, as measured using LED flasher data.


Amplitude calibration is also well understood. These calibrations are discussed in pages 64-67 of the collected IceCube contributions to the 29th International Cosmic Ray Conference in Pune, India during August 2005.

The detector calibration includes corrections for PMT gain, ATWD and fADC gain and timing and (coming soon) PMT saturation.  The 3 ATWD waveforms are combined into a single calibrated waveform.  The detector calibration is performed in DOMCalibrator, which is documented here.


DOM main board simulation

Each digital optical module (DOM) contains electronics that process and capture the signal produce by Cerenkov light in the photomultiplier tubes. It essentially acts as a small satellite, with independent triggering, data acquisition, and calibration circuitry. We have written a simulator for this board, incorporating all of the pieces that are relevant for physics.

The DOM main board electronics apply a threshold trigger to the PMT analog signals and digitize the ones above threshold. Fast waveform sampling and capture (128 samples at 300 MegaSamples/Second (MSPS) is provided by two analog transient waveform digitizers (ATWDs). Longer duration signals (up to 6.4 micoseconds) are sampled at 40 MHz by a fast analog-to-digital converter (FADC). Each ATWD has three channels with different gains, providing a 14-bit dynamic range. A slow (40 MSPS) 10-bit digitizer provides a 6.4 microsecond record of arrival time, but with coarser resolution. The simulation also covers the digital logic for triggering, local coincidence signalling between adjacent DOMS, and deadtime.

The simulation is discussed in this talk given at the March 2005 collaboration meeting in Berkeley.


Much of our offline computing is done at PDSF a 780 processor Linux farm operated by the NERSC (National Energy Research Supercomputer Center)

High pT muons in air showers

The LBNL group is also interested in the search for high pT muons in cosmic-ray air showers.   The production of these muons is calculable in perturbative quantum chromodynamics (pQCD), and so the presence of these muons can be used to probe the cosmic-ray composition. shows an event, recorded on May 23, 2007, containing an apparent well-separated track consistent with a high pT muon.   An air shower hits 11 IceTop stations, while a total of 96 IceCube DOMS are hit; 84 on 4 strings near the extrapolated air shower core, plus 12 DOMs on another string, about 400 m from the projection.  Files: highptchina.pdf and writeup2.pdf contain an extended discussion of this search.


Back to Top