Time-Dependent Searches for Point Sources of Neutrinos with the 40-String and 22-String Configurations of IceCube

This paper presents searches for flaring sources of neutrinos using the IceCube neutrino telescope. For the first time, a search is performed over the entire parameter space of energy, direction and time looking for neutrino flares of 20 microseconds to a year duration from astrophysical sources among the atmospheric neutrino and muon backgrounds. Searches which integrate over time are less sensitive to flares because they are affected by a larger background of atmospheric neutrinos and muons that can be reduced by the time constraint. Flaring sources considered here, such as active galactic nuclei, soft gamma-ray repeaters and gamma-ray bursts, are promising candidate neutrino emitters. We used mainly data taken between April 5, 2008 and May 20, 2009 by a partially completed configuration of IceCube with 40 strings. For the presented searches an unbinned maximum likelihood method is used with a time-dependent prior to test several different source hypotheses. An “untriggered” search covers any possible time-dependent emission from sources not correlated to any other observation using other astrophysical messengers such as photons. Moreover, a similar time scan is performed for a predefined catalogue of sources that exhibit intense photon flares. Searches triggered by multi-wavelength information on flares from blazars and soft gamma-ray repeaters are performed using the 40 string data and also the data taken by the previous configuration of 22 strings in operation between May 31, 2007 and April 5, 2008. Flares for which extensive and continuous monitoring is available from Fermi-LAT and SWIFT and flares detected by imaging Cherenkov telescopes with shorter time-scale monitoring are considered. The results from all searches are compatible with a fluctuation of the background.

💡 Research Summary

This paper reports four time‑dependent searches for astrophysical neutrino flares using data from the partially‑completed IceCube detector in its 22‑string (May 2007–April 2008) and 40‑string (April 2008–May 2009) configurations. The motivation is that many candidate sources—active galactic nuclei (AGN), soft gamma‑ray repeaters (SGR), and gamma‑ray bursts (GRB)—are expected to emit neutrinos in short, variable episodes. Traditional time‑integrated analyses suffer from a large atmospheric neutrino and muon background, whereas restricting the search to a narrow time window can dramatically improve sensitivity.

The analysis employs an unbinned maximum‑likelihood method that incorporates three PDFs: spatial, energy, and a time‑dependent prior. The time prior is taken to be log‑uniform between 20 µs and one year, allowing the algorithm to scan a wide range of flare durations without bias. Background PDFs are derived directly from data by randomizing event times (time‑scrambling), ensuring that the background model reflects the actual detector conditions. Trial factors arising from scanning many time windows, source positions, and flare durations are accounted for by performing the same likelihood maximization on many scrambled datasets; the post‑trial p‑value is then the fraction of scrambled trials that produce a test‑statistic equal to or larger than that observed.

The four searches are:

1. All‑sky untriggered scan – a blind search over the entire sky for clusters of events in space and time, with no external timing information.

2. Catalog‑based untriggered scan – a targeted search of 40 AGN identified by Fermi‑LAT as variable in the GeV band; each source is examined independently with the same time‑window scan.

3. Multi‑wavelength (MWL) triggered search using continuous monitoring – flare periods are defined from Fermi‑LAT and Swift light curves, assuming that neutrino emission follows the photon light curve.

4. MWL triggered search using sporadic high‑energy observations – flare windows are set from short‑duration observations by imaging atmospheric Cherenkov telescopes (MAGIC, VERITAS, H.E.S.S.) that typically monitor sources only during known high‑state episodes.

For flares of order one second, the time‑dependent method improves the discovery potential by a factor of four to five relative to a time‑integrated analysis. Sensitivities are presented as flux limits (E⁻² spectra) ranging from ~10⁻⁸ GeV cm⁻² s⁻¹ for the shortest windows to ~10⁻⁹ GeV cm⁻² s⁻¹ for year‑long integrations.

The results: in all four analyses, the most significant excesses correspond to post‑trial p‑values of a few percent (≈2–3 σ), fully compatible with background fluctuations. No evidence for neutrino emission correlated with any of the examined photon flares or with any untriggered cluster was found.

The paper concludes that while the 22‑ and 40‑string configurations lack the statistical power to claim a detection, the methodology demonstrates that time‑dependent searches are a powerful tool for reducing atmospheric background and enhancing sensitivity to short‑duration astrophysical neutrino transients. The authors advocate extending these techniques to the full 86‑string IceCube detector, increasing the live‑time, and integrating real‑time multi‑wavelength alerts to enable rapid follow‑up of promising transient events. This work thus lays the groundwork for future neutrino time‑domain astronomy.