From cosmic messengers to a physical picture

Where are high-energy neutrinos produced in AGN?

IceCube has established a diffuse flux of astrophysical neutrinos, but the sources responsible for most of this emission remain unidentified. Active galactic nuclei (AGN) are among the leading candidates, yet one fundamental question is still open: where within an AGN do hadronic interactions become efficient enough to produce high-energy neutrinos?

We test whether the dominant neutrino-production sites are located in relativistic jets or in compact environments close to the supermassive black hole, such as the accretion-disk corona. By combining neutrino, hard X-ray, gamma-ray and radio observations, we aim to identify the physical conditions that turn AGN into efficient cosmic particle accelerators.

The hard X-ray–neutrino connection

A key clue comes from hard X-rays. In compact regions close to the black hole, dense photon fields can make proton–photon interactions efficient while absorbing the accompanying high-energy gamma rays. The resulting electromagnetic cascades can reprocess part of this energy into the hard X-ray band.

We found observational evidence for a correlation between the unabsorbed 15–55 keV hard X-ray luminosity and the 0.3–100 TeV neutrino luminosity of IceCube-associated AGN. Remarkably, the relation extends across both radio-quiet Seyfert galaxies and jet-dominated blazars, suggesting that compact photon-rich environments near the central black hole may provide a common neutrino-production channel.

We are extending this test using archival and targeted NuSTAR observations of IceCube candidate blazars. This includes fields in which several AGN may contribute to the same neutrino hotspot, allowing hard X-ray observations to help disentangle candidate counterparts beyond the angular resolution of the neutrino data.

Time-domain multimessenger diagnostics

Neutrino production does not necessarily coincide with the brightest gamma-ray flares. Several IceCube-associated blazars have instead shown neutrino activity during gamma-ray faint or suppressed states, suggesting that the highest-energy particles can be produced in regions that are opaque to gamma rays.

We combine IceCube neutrino histories with adaptively binned Fermi-LAT gamma-ray light curves, gamma-ray spectra, radio monitoring and VLBI observations. By identifying flares, measuring time delays and following spectral changes, we test whether neutrino activity originates close to the black hole or farther downstream in the relativistic jet.

VLBI jet kinematics

Very long baseline interferometry (VLBI) provides the sharpest direct view of relativistic AGN jets, resolving structures on parsec and sub-parsec scales. Tracking individual jet features over many years allows us to reconstruct the geometry and dynamics of the flow.

Using long-term monitoring programmes such as MOJAVE and other VLBI datasets, we analyse jet ridgelines, component trajectories, apparent superluminal motion, core brightness temperatures and changes in jet position angle. These measurements constrain Doppler factors, viewing angles and the physical distance between the radio core and the high-energy emission region.

Combining these geometric constraints with gamma-ray and neutrino variability allows us to distinguish activity produced close to the central engine from disturbances propagating through the parsec-scale jet.

Population searches with IceCube

Individual source associations provide important clues, but establishing a source population requires statistical tests across many AGN. We use physically motivated source samples to test whether particular classes of AGN contribute significantly to the diffuse astrophysical neutrino flux.

A central goal is to use intrinsic hard X-ray emission as a predictive tracer of neutrino production. We perform IceCube stacking analyses in which candidate AGN are weighted by their unabsorbed hard X-ray flux and compare correlation-consistent sources with carefully constructed control samples.

As an independent consistency test in the local Universe, we also investigate whether nearby hard X-ray AGN are preferentially associated with ultra-high-energy cosmic-ray arrival directions. Propagation calculations with CRPropa allow us to account for energy losses and deflections in Galactic and extragalactic magnetic fields.

Supermassive black hole binaries and precessing jets

Some AGN may host inspiralling supermassive black hole binaries (SMBHBs). Their orbital dynamics and spin–orbit coupling can induce long-term jet precession, producing characteristic changes in VLBI morphology and quasi-periodic high-energy variability.

We model jet variability using orbital motion, spin–orbit precession and relativistic Doppler boosting, fitting VLBI morphology, component kinematics and radio variability simultaneously. This allows us to constrain the underlying binary parameters and distinguish purely geometric variability from physical changes in particle acceleration.

If jet precession periodically changes the interaction of the relativistic flow with surrounding matter or radiation fields, neutrino production may also become phase dependent. We therefore develop phase-resolved IceCube searches and connect candidate systems to the nanohertz gravitational-wave sky and to PTA and LISA sensitivity.

Methods and statistics

Multimessenger coincidences become physically meaningful only when the statistical analysis properly accounts for measurement uncertainties, selection effects and the large number of possible source associations.

Our toolkit includes likelihood-based IceCube stacking analyses, weighted permutation tests, regression with measurement uncertainties, adaptive and Bayesian-block time-series analysis, spectral modelling, multiwavelength time-lag measurements and Monte Carlo simulations. Rather than relying only on positional coincidences, we build physically motivated observables directly into the statistical tests.

Gravitational-wave multimessenger searches

In parallel with the core AGN–neutrino programme, we contribute to gravitational-wave multimessenger research with the LIGO Scientific Collaboration. One question of particular interest is whether a fraction of stellar-mass binary black hole mergers occurs in the dense environments of AGN.

We work on targeted searches connecting gravitational-wave events with AGN environments and on strategies for combining gravitational-wave information with electromagnetic and high-energy neutrino observations. This provides a complementary route toward understanding how active galaxies influence compact-object populations and high-energy particle acceleration.

International collaborations

Our research is carried out through a broad international network spanning neutrino astronomy, high-energy astrophysics, radio interferometry, cosmic rays and gravitational-wave science.

Key collaborations include Ruhr University Bochum (Germany), the University of Florida (USA), the University of Wisconsin–Madison (USA), the Max Planck Institute for Radio Astronomy in Bonn (Germany), and Columbia University (USA), together with colleagues in the IceCube and LIGO scientific communities and international VLBI programmes.

Data, observatories and numerical tools

IceCube Neutrino Observatory · Fermi Large Area Telescope (Fermi-LAT) · Nuclear Spectroscopic Telescope Array (NuSTAR) · VLBA / MOJAVE and other VLBI networks · radio monitoring programmes · ultra-high-energy cosmic-ray catalogues · CRPropa · pulsar timing array data and sensitivity curves · LISA forecasts · LIGO–Virgo–KAGRA gravitational-wave data and catalogues.