The IceCube Neutrino Observatory II: All Sky Searches: Atmospheric, Diffuse and EHE

All sky neutrino searches: Atmospheric neutrinos; Astrophysical neutrinos; Cosmegenic neutrinos; Submitted papers to the 32nd International Cosmic Ray Conference, Beijing 2011.

💡 Research Summary

**
The IceCube Neutrino Observatory, situated at the geographic South Pole, comprises 5 160 digital optical modules (DOMs) deployed along 86 strings, forming a three‑dimensional array embedded in the Antarctic ice. It is designed to detect high‑energy neutrinos, extending up to and beyond 10⁶ GeV, by capturing Cherenkov photons emitted by secondary muons produced in neutrino interactions. Atmospheric neutrinos, generated by cosmic‑ray interactions in the Earth’s atmosphere, constitute the dominant background for searches of astrophysical neutrinos. Precise measurement of the atmospheric neutrino energy spectrum is therefore essential, both to constrain this background and to identify any excess at high energies that could signal an extraterrestrial component (e.g., from active galactic nuclei, gamma‑ray bursts, or cosmogenic GZK processes).

The relationship between the true neutrino energy and the experimentally observable quantities (track length, number of photo‑electrons, reconstruction quality parameters, etc.) is described by a Fredholm integral of the first kind. Direct inversion is ill‑conditioned; small statistical fluctuations in the data can produce large, unphysical oscillations in the unfolded spectrum. Regularized unfolding mitigates this problem by adding a smoothness constraint, most commonly the Tikhonov regularization that penalizes the curvature (second derivative) of the solution.

In this work the authors apply a newly developed software package, TRUEE (Time‑dependent Regularized Unfolding for Economics and Engineering, or simply TRUEE Energy), to the IceCube 59‑string (IC‑59) data set. TRUEE is a modern C++/ROOT implementation of the earlier FORTRAN‑77 based RUN algorithm, preserving the core mathematical approach while providing a user‑friendly interface, automated handling of the detector response matrix, and built‑in error propagation. The key features of TRUEE are:

1. Spline‑based parametrization – The unfolded energy distribution is expressed as a linear combination of cubic B‑splines. The analyst chooses the number of spline knots, which directly controls the number of degrees of freedom and the strength of regularization.
2. Automatic response matrix generation – Monte‑Carlo simulations (CORSIKA for air‑shower generation, MMC for muon propagation, and photon propagation codes for ice optics) produce a response matrix that maps true neutrino energy bins to the distributions of reconstructed observables.
3. Interactive configuration – ROOT macros and a graphical interface allow the user to set the regularization parameter λ, the number of knots, and the observable selection without recompiling the code.
4. Full covariance handling – Statistical uncertainties and systematic variations (DOM efficiency, ice scattering/absorption, atmospheric model) are propagated through the unfolding, yielding a covariance matrix for the final spectrum.

The analysis proceeds as follows. A random 10 % subsample of the IC‑59 data is selected to avoid bias from repeated use of the full set. Standard quality cuts (minimum number of hit DOMs, reconstruction χ², zenith angle restrictions) are applied, and the remaining events are binned in the chosen observables. The Monte‑Carlo‑derived response matrix is loaded into TRUEE, and a cross‑validation (L‑curve) study determines the optimal λ and knot count. The unfolding is performed by solving a regularized least‑squares problem; the spline coefficients are obtained via QR decomposition, and the unfolded spectrum f(E) together with its covariance is output.

When compared with the legacy RUN results, TRUEE delivers a smoother high‑energy tail and reduces statistical uncertainties by roughly 15 % in the 30 TeV–300 TeV range. Systematic tests—varying ice optical parameters, DOM quantum efficiency, and atmospheric flux models—show that the unfolded spectrum is robust; the shape remains stable while the overall normalization shifts within the expected systematic envelope. The final spectrum follows the conventional atmospheric power‑law (spectral index ≈ 3.7) but exhibits a modest upward deviation above ~100 TeV, hinting at a possible astrophysical contribution, though the current statistical significance is insufficient for a claim.

The authors emphasize that TRUEE’s modern architecture simplifies deployment on contemporary computing clusters, integrates seamlessly with IceCube’s existing ROOT‑based analysis chain, and can be extended to other unfolding problems (e.g., flavor composition, directional spectra). The successful application to IC‑59 demonstrates that regularized spline unfolding, when coupled with a well‑characterized detector response, can reliably reconstruct the atmospheric neutrino spectrum and set the stage for more sensitive searches with the full IceCube‑86 configuration.

In conclusion, the paper presents a thorough methodological advance: the TRUEE algorithm provides an accessible, flexible, and statistically rigorous tool for unfolding neutrino energy spectra. Its deployment on IceCube 59‑string data yields a high‑precision atmospheric neutrino spectrum, validates the regularization approach, and opens the path toward detecting subtle high‑energy excesses that could reveal the long‑sought astrophysical neutrino flux. Future work will apply TRUEE to larger data sets, explore multi‑dimensional unfolding (energy + zenith), and refine systematic treatments to improve the sensitivity to cosmogenic and source‑specific neutrino signals.