IceCube, the telescope searching for the highest energy neutrinos created by nature, is also a huge muon track detector, where a few neutrino-induced muons hide within millions of muons created by the interaction of cosmic rays with the Earth´s atmosphere. The search for high-energy neutrino point sources relies on the ability of scientists to accurately reconstruct most of the muon tracks in IceCube in order to separate the background cosmic-ray muons from muons produced by the interaction of neutrinos in the Antarctic ice.
A new study by the IceCube Collaboration shows that the muon track reconstruction performed in the early stages of the analysis can be significantly improved by using robust statistical methods to estimate particle trajectories through the detector. New optimized filters and classical data analysis techniques have been introduced to detect and remove outlier hits before reconstructing the trajectory of the muon in IceCube. The new algorithm results in a 13% gain in the angular resolution of the muon track and a 98% accuracy rate in determining the number of muons in coincident events. The paper has just been submitted to Nuclear Instruments and Methods in Physics Research Section A.
IceCube data are analyzed in real time at the South Pole. Only those events that are shown to be interesting for physicists are sent through satellite connection to Madison, Wisconsin. There they are made available to an international collaboration of about 250 scientists, and, at this point, more complex analyses are performed looking for neutrino or cosmic ray signals.
Previous efforts to improve the reconstruction of the muon direction build upon sophisticated and demanding computational techniques that model the Cherenkov cone of the high-energy muon moving through IceCube as well as the scattering and absorption of photons in the ice. The collaboration now presents an improved algorithm for the early online track reconstruction performed in the IceCube Lab, located at the Amundsen-Scott South Pole station, which will benefit any later analysis.
“The results are impressive. A few changes in the initial reconstruction of the muon track turned into a significant improvement of the estimation of its direction. And for coincident events, our new algorithm is doing a 66% better job than the software previously used in IceCube,” explains Mark Wellons, a graduate student from the Department of Computer Sciences at UW-Madison and a member of the IceCube Collaboration.
In the old algorithm, the muon was modeled as a single point moving in a straight line. However, the Cherenkov emission and the photon scattering create hit signals at significant distances from this ideal line. On top of that, outlier noise hits had a significant impact in the model because they were penalized quadratically on their distance from this point.
In the new reconstruction algorithm developed by the IceCube Collaboration, the hits due to photon scattering are identified and disregarded for determining the direction of the muon, improving the accuracy by almost a factor of two. In order to marginalize noise outliers, hits in the near-hit regime are given quadratic weights, while those in the far-hit regime are given linear weights as they are more likely to be noise. In a further improvement of the reconstruction algorithm, hits identified as noise outliers are taken out of the fit. The 13% improvement of the angular resolution results in a 10% reduction of erroneously reconstructed atmospheric muons and a 1% increase in muons correctly reconstructed as upgoing, in other words, as neutrino-induced muons.
Info “Improvement in Fast Particle Track Reconstruction with Robust Statistics,” IceCube Collaboration: M.G. Aartsen et al. Nuclear Instruments and Methods in Physics Research Section A 736 (2014), arXiv.org:1308.5501