Abstract Since the start of the gravitational wave observation era, no joint high-energy neutrino and gravitational wave event has been found. These non-detections can be used to set upper bounds on the neutrino emission properties for gravitational wave events individually or for a set of them. Although previous joint searches have established upper limits, there has been a lack of a consistent method for these calculations. This paper addresses the problem of finding limits for astrophysical events that are poorly localized in the sky—where the sensitivities of neutrino detectors change significantly—and may emit neutrinos, such as gravitational wave detections. We describe methods for assigning limits for expected neutrino count, emission fluence, and isotropically equivalent emission based on maximum likelihood estimators (MLE). These methods are applied to the three gravitational wave detections from aLIGO's first observing run (O1) to establish robust upper limits.
Introduction
As multi-messenger astronomy matures, the search for high-energy neutrinos coincident with cataclysmic events like binary black hole mergers has become a priority. However, the current reality of the field is defined by the "non-detection." While the absence of a signal might seem like a dead end, it actually provides critical constraints on the underlying physics of cosmic accelerators.
This paper, published in the Journal of Cosmology and Astroparticle Physics (JCAP), provides a rigorous statistical framework to transform these null results into scientifically valuable upper limits. By standardizing how we calculate the maximum possible neutrino emission from a source that went undetected, researchers can more effectively narrow the phase space for theoretical models.
The Challenge: Blurry Maps and Changing Sensitivity
Calculating upper limits for gravitational wave (GW) events is complicated by two primary factors:
- Large Sky Localizations: Unlike optical telescopes, gravitational wave interferometers like LIGO often provide "blurry" localization maps, sometimes spanning hundreds of square degrees.
- Directional Sensitivity: Neutrino telescopes, such as IceCube, do not have uniform sensitivity across the sky. Their ability to detect a signal depends heavily on the declination of the source and the amount of Earth the neutrinos must traverse before reaching the detector.
Previous studies often used inconsistent methods to account for these variables. This research seeks to standardize the process using a frequentist approach that remains robust even when the background noise in the detector fluctuates.
Methodology: Maximum Likelihood Estimators
The core of the proposed framework is the use of Maximum Likelihood Estimators (MLE) to derive upper limits.
The Likelihood Function
The authors construct a model based on the probability of observing a specific Test Statistic (TS) given a hypothesized number of signal neutrinos. They specifically model cases for 0, 1, and 2 signal neutrinos ($P_0, P_1, P_2$). This approach is particularly useful because it "restores" physicality to the results; direct use of a TS can sometimes lead to empty confidence intervals if the observed background is lower than expected. The MLE approach avoids this by using distributions derived from physical simulations.
Integrating Sky Maps
To handle the uncertainty of GW localization, the method integrates the neutrino detector's effective area over the entire probability distribution of the GW sky map. This ensures that the resulting limit reflects the weighted average of the detector's sensitivity across the entire region where the event likely occurred.
The Power of Ensemble Analysis
One of the paper's most significant contributions is the "ensemble method." By assuming that a group of similar events (e.g., all binary black hole mergers) shares the same isotropically equivalent emission energy ($E_{iso}$), researchers can stack multiple non-detections. This combined analysis allows for much tighter global limits than could be achieved by looking at any single event in isolation.
Results from LIGO’s First Observing Run (O1)
The authors validated their framework using the three confirmed detections from LIGO’s O1 run: GW150914, GW151012, and GW151226, alongside public data from the IceCube Neutrino Observatory.
| Event | Signal Neutrino Count Limit | Fluence Limit ($GeV/cm^2$) | $E_{iso}$ Limit (ergs) |
|---|---|---|---|
| GW150914 | 2.3 | 0.80 | $5.1 \times 10^{53}$ |
| GW151012 | 2.2 | 0.40 | $1.1 \times 10^{54}$ |
| GW151226 | 2.3 | 0.45 | $2.6 \times 10^{53}$ |
| Combined | - | - | $6.1 \times 10^{52}$ |
The results highlight how location impacts sensitivity. GW150914, located entirely in the southern hemisphere where IceCube is less sensitive to high-energy neutrinos, had emission limits roughly twice as high as the other two events. Most importantly, the combined $E_{iso}$ limit was an order of magnitude lower than the individual limits, demonstrating the efficiency of the ensemble approach.
Discussion and Future Outlook
While the framework provides a major step toward standardization, it currently relies on a few simplifying assumptions, such as a standard $E^{-2}$ neutrino energy spectrum and a maximum of two signal neutrinos. Future iterations could expand this to include multiple detectors (such as ANTARES or KM3NeT) and more complex spectral models.
By providing a consistent way to "squeeze" the theoretical possibilities through non-detections, this MLE-based framework ensures that every observation—even those that see nothing—contributes to our understanding of the high-energy universe.
Summary updated