Evidence For A Correlation Between Astrophysical Neutrinos and Radio Flares

TL;DR

This paper investigates whether variable radio sources detected in VLASS are physically associated with high-energy astrophysical neutrinos observed by IceCube. By defining spatially and temporally constrained radio-neutrino samples (full variables, flaring, and lagged-flaring) and comparing to random backgrounds via Monte Carlo trials, the authors identify a >2σ excess of associations for the flaring and lagged-flaring samples, implying radio flares contribute about $13\%$ of IceCube's astrophysical neutrinos. The associated radio sources mostly lack gamma-ray or X-ray counterparts, but infrared colors indicate AGN/blazar hosts, consistent with a population that may extend to higher redshifts than high-energy counterparts alone. The results motivate continued VLASS observations and anticipate stronger significance with future IceCat data (IceCat-2) and additional VLASS epochs, potentially reaching >$5\sigma$ for radio-neutrino associations and enabling a more complete census of neutrino sources.

Abstract

We use data from the first two epochs of the Very Large Array Sky Survey (VLASS) and the IceCube Neutrino Observatory to search for evidence of a correlation between radio variability and the detection of astrophysical neutrinos. We find an excess number of associations between flaring radio sources and neutrinos that were detected between the first and second VLASS observations at $>2σ$ confidence. This excess is consistent with radio flares contributing $\sim13\,\%$ of the astrophysical neutrinos observed by IceCube. Notably $>80\,\%$ of the radio flares associated with neutrinos are not detected at either $γ$-ray or X-ray wavelengths, highlighting the importance of radio observations for identifying potential electromagnetic counterparts to astrophysical neutrinos. No excess in the number of associations between the wider radio-variable population and the IceCube neutrinos is seen when no time constraint is placed on the neutrino detection. We predict that data from future VLASS epochs will see an excess number of associations between radio flares and neutrinos at the $>3σ$ level, and expected improvements to the positional constraints on the neutrinos may increase that confidence to $>5σ$, should our results be representative.

Figures (7)

• Figure 1: The on-sky distribution of VLASS variables (blue dots) and IceCube neutrino events from IceCat-1 (red circles) used in this work shown on an Aitoff projection in equatorial coordinates. Note that the size of the red circles is arbitrary and does not correspond to the positional uncertainty of the neutrinos.
• Figure 2: Example IceCube events consistent with VLASS sources. In all panels the red cross shows the most likely origin coordinates of the neutrino, the black ellipse shows the $90\,$% positional confidence, gray dots show VLASS sources, light green circles are variable VLASS sources, and blue stars highlight flares in VLASS that are consistent with the neutrino event time. Note that the apparent relative sparsity of VLASS sources in panel (c) is an effect of the panel being zoomed in on a neutrino event with a smaller positional uncertainty than in panels (a) and (b).
• Figure 3: The number of associations between VLASS radio sources and IceCube neutrinos for our three test samples. In each panel the blue histogram shows the distribution of the association count in the 1000 run Monte Carlo simulation, and the red dashed line shows the observed number of associations.
• Figure 4: The expected background distributions of radio source and neutrino event counts for the two test samples with a higher than expected number of radio-neutrino associations. Panel (a) shows the flaring sample, and panel (b) shows the lagged-flaring sample. In both panels the contours show the $95$th, $75$th, $50$th, $25$th, and $5$th percentile from the PDF estimated by the Monte Carlo simulation. The red cross and dashed lines show the position of the real numbers of neutrino events and radio sources associated.
• Figure 5: Distributions of angular separations between associated radio sources and neutrino events. The solid blue histogram shows random associations from Monte Carlo simulations. Angular separations for real associations from the flaring and lagged-flaring sample are shown by the yellow dot-dashed and red solid lines, respectively. The dashed black vertical line shows the 95th percentile of the positional errors exhibited by IceCat-1 events, $\sigma_{\nu}$ and the dotted black vertical line represents an angle of $1^{\circ}$.