[Paper Review] AMON Searches for Jointly-Emitting Neutrino + Gamma-Ray Transients

A. Keivani, D. B. Fox+3 more|arXiv (Cornell University)|Aug 6, 2015

Astrophysics and Cosmic Phenomena9 references3 citations

TL;DR

This paper presents an archival search for coincident high-energy neutrino and gamma-ray transients using public data from IceCube (IC40) and Fermi LAT, applying unbinned log-likelihood analysis and multiple vetting tests. Despite identifying 2,138 coincidences with a signal-like likelihood distribution suggesting ~70 potential signals, all subsequent tests—multiplicity, time difference distribution, and spatial clustering—show no significant excess over background, indicating no statistically significant joint emission from cosmic sources in the dataset.

ABSTRACT

We present the results of archival coincidence analyses using public neutrino data from the 40-string configuration of IceCube (IC40) and contemporaneous public gamma-ray data from Fermi LAT. Our analyses have the potential to discover statistically significant coincidences between high-energy neutrino and gamma-ray signals, and hence, possible jointly-emitting neutrino/gamma-ray transients. This work is an example of more general multimessenger studies that the Astrophysical Multimessenger Observatory Network (AMON) aims to perform. AMON is currently under development and will link multiple running and future high-energy neutrino, cosmic ray and follow-up observatories as well as gravitational wave facilities. This single network will enable near real-time coincidence searches for multimessenger astrophysical transients and their electromagnetic counterparts. We will present the component high-energy neutrino and gamma-ray datasets, the statistical approaches that we used, and the results of analyses of the IC40+LAT datasets.

Motivation & Objective

• To search for statistically significant coincidences between high-energy neutrinos from IceCube and gamma-ray photons from Fermi LAT as evidence of jointly-emitting astrophysical transients.
• To test the feasibility of multimessenger coincidence analysis using public datasets within the AMON framework.
• To evaluate statistical methods for identifying potential neutrino-gamma ray source associations in archival data.
• To assess whether any observed coincidences exceed background expectations through multiple vetting procedures.

Proposed method

• Utilized unbinned log-likelihood function (λ) to assess the probability of coincidence between neutrino and photon events based on angular separation and time difference.
• Applied energy-dependent point spread functions (PSFs) for IceCube (Gaussian with energy-dependent width) and Fermi LAT to model directional uncertainties.
• Set angular separation threshold <10° and temporal window |Δt| < 50 s to define candidate doublets.
• Used background rejection term B(γ̂) and signal injection simulations to calibrate the likelihood metric and assess significance.
• Performed three vetting tests: multiplicity (number of photons per neutrino), time difference distribution (Δt), and spatial clustering (within 2° of another pair).
• Calculated signal-to-noise ratio (SNR) to determine optimal λ cut-off at 11, selecting only high-likelihood events for further analysis.

Experimental results

Research questions

• RQ1 Is there a statistically significant excess of neutrino-gamma ray coincidences in the IC40 and Fermi LAT datasets beyond background expectations?
• RQ2 Do the observed coincidences exhibit temporal, directional, or multiplicity patterns consistent with astrophysical sources?
• RQ3 Can the likelihood metric effectively distinguish signal from background in multimessenger coincidence searches?
• RQ4 Do spatial clustering or time distribution anomalies suggest the presence of bright or repeating sources?
• RQ5 What is the significance of the observed signal-like distribution with ~70 injected signals in the likelihood analysis?

Key findings

• A total of 2,138 neutrino-gamma ray coincidences were identified within 10° angular separation and ±50 s time window between IC40 and Fermi LAT data.
• The Anderson-Darling test indicated a signal-like likelihood distribution consistent with ~70 injected signals, suggesting a potential signal excess.
• The multiplicity test showed a mean of 2.17 photons per high-λ neutrino in data, compared to 2.08 in null hypothesis, with a p-value of 0.25, indicating no significant excess.
• The Δt distribution was consistent with a flat uniform distribution (reduced χ² ≈ 0.5), showing no temporal clustering or preferred emission window.
• Only six high-likelihood ν-γ pairs were found within 2° of each other, fewer than the 12.9 expected from null hypothesis simulations, indicating no significant spatial clustering.
• All three vetting tests—multiplicity, time distribution, and clustering—showed no significant deviation from background, leading to the conclusion that no statistically significant joint emission was detected.