An improved method to search for flares from point sources of ultra-high-energy photons

J. Stasielak 1, N. Borodai 1, D. Góra 1, M. Niechciol 2 1 Nuclear Physics Polish Academy of Sciences, PL-31342 Krakow, Poland 2 Center for Particle Physics Siegen, University of Siegen, Siegen, Germany · 2024

Read An improved method to search for flares from point sources of ultra-high-energy photons by J. Stasielak 1, N. Borodai 1, D. Góra 1, M. Niechciol 2 1 Nuclear Physics Polish Academy of Sciences, PL-31342 Krakow, Poland 2 Center for Particle Physics Siegen, University of Siegen, Siegen, Germany (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-13804.txt.

Introduction

Stasielak and co-authors improve a method for hunting brief flares of ultra-high-energy photons from points in the sky, in a 2024 paper. If certain cosmic objects flare in these photons, their arrivals should bunch in both place and time against a background of ordinary cosmic rays. Their stacking method joins a time-clustering step with a likelihood analysis, and adds a tag that scores each event as photon-like or hadron-like.

Tests show the method can pull a true photon flare out of background with only a handful of events, and can recover both the flare's length and its photon count. The authors aim it at searches for cosmic ray sources and at tightening limits on how many such photons reach Earth.

Catalog record

CREATORJ. Stasielak 1, N. Borodai 1, D. Góra 1, M. Niechciol 2 1 Nuclear Physics Polish Academy of Sciences, PL-31342 Krakow, Poland 2 Center for Particle Physics Siegen, University of Siegen, Siegen, Germany
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-13804

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

An improved method to search for flares from point sources of ultra-high-energy photons J. Stasielak 1, N. Borodai 1, D. Góra 1, M. Niechciol 2 1 Nuclear Physics Polish Academy of Sciences, PL-31342 Krakow, Poland 2 Center for Particle Physics Siegen, University of Siegen, Siegen, Germany Abstract Flares produced by certain classes of astrophysical objects may be sources of some ultra-high-energy particles, which, if they are photons, would group into clusters of events correlated in space and time. Identification of such clustering in cosmic-ray data would provide important evidence for possible existence of ultra-high-energy (UHE) photons and could potentially help identify their sources. We present an analysis method to search for space-time clustering of ultra-high-energy extensive air showers, namely the stacking method, which combines a time-clustering algorithm with an unbinned likelihood study. In addition, to enhance the capability to discriminate between signal (photon-initiated events) and background (hadron-initiated) events, we apply a photon tag. This involves using relevant probability distribution functions to classify each event as more likely to be either a photon or a hadron. We demonstrate that the stacking method can effectively distinguish between events initiated by photons and those initiated by hadrons (background). The number of photon events in a data sample, as well as the flare(s) duration can also be retrieved correctly. The stacking method with a photon tag requires only a few events to identify a photon flare. This method can be used to search for the cosmic ray sources and/or improve limits on the fluxes of UHE photons. keywords space-time clustering of UHECRs, unbinned likelihood method, point sources, UHE photons 1 Introduction Some fraction of ultra-high-energy cosmic rays (UHECRs) may originate from astrophysical flaring events such as AGN flares or gamma-ray bursts. This could potentially lead to the formation of clusters in the observational data collected by various cosmic-ray experiments, meaning the formation of groups of events in the data that are correlated both temporally and spatially. Space-time correlations of events (clustering) are most likely to occur for neutral particles, such as photons, that are not deflected by magnetic fields and therefore follow a straight path to Earth, pointing to their sources. Thus, the observation of clustering in the cosmic-ray data may indicate the existence of UHE photons, i.e. with energy larger than 10^{17} eV. In this paper, we propose an improved search algorithm for space-time clustering in UHECR (air-shower) data, a slight modification ( Góra et al., 2011 ) of the commonly used unbinned likelihood method ( Braun et al., 2008 ) , which can be applied to a time-dependent search for weak multiple flares from point sources of UHE photons. The proposed algorithm has already been applied to investigate correlations between the arrival directions of neutrinos detected by the IceCube Neutrino Observatory and the directions to their possible sources ( Góra et al., 2011 ) . We demonstrate that this improved approach, so called stacking method, is capable of effectively detecting time-space-correlated UHECR clusters. Moreover, we calculate its efficiency. This method may prove useful in identifying cosmic ray sources and in establishing stronger constraints on the fluxes of UHE neutral particles such as UHE photons. 2 Standard space-time clustering analysis The standard approach to search for space-time clustering in UHECR (air-shower) data is based on the unbinned likelihood method ( Braun et al., 2008 ) with an addition of search for the time correlation between events. In this method, finding a point source of photons in the sky (flaring event) means to locate an excess of space-time correlated events from a particular direction over the background (a cluster of events). The overall procedure for such a search is as follows. First, we take all consecutive multiplets in the data, i.e. doublets, triplets, quadruplets, and so on. Subsequently, for each multiplet j we construct likelihood \mathcal{L} and the corresponding test statistic \rm{TS} . Next, we maximize \rm{TS} to obtain its maximum value \rm{TS}_{\max} . As an outcome, we get a set of \rm{TS}_{\max} values, one for each multiplet. We choose the most significant multiplet, i.e. the one with the largest \rm{TS}_{\max} . The flare duration \Delta T is estimated to be the length of the time window of this most significant multiplet. The statistical significance of the results is assessed by comparing the value of \rm{TS}_{\max} calculated for the most significant multiplet of the data with a distribution of maximum values of the test statistic derived from many randomly generated scrambled maps mimicking the original dataset. Details of the described method are given below. A signal probability distribution function (PDF) of an event i is given by s_{i}=s_{i}^{\mathrm{space}}s_{i}^{\mathrm{time}}\rm{,} (1) a combination of the contribution from space and time part. The spatial probability s_{i}^{\mathrm{space}} is a Gaussian function s_{i}^{\rm{space}}=\frac{1}{2\pi\sigma^{2}}\exp\left(-\frac{\left|\vec{r}_{i}-\vec{r}_{s}\right|^{2}}{2\sigma_{i}^{2}}\right)\rm{,} (2) where the angular reconstruction uncertainty of considered event and its direction are given by \sigma_{i} and \vec{r}_{i} , respectively. The source direction is denoted by \vec{r}_{s} . For each time window of a multiplet tested \Delta t_{j}=t^{\rm{max}}_{j}-t^{\rm{min}}_{j} , a temporal PDF is given by s_{i}^{time}=\frac{H\left(t_{j}^{max}-t_{i}\right)H\left(t_{i}-t_{j}^{min}\right)}{\Delta t_{j}}, (3) where H is the Heaviside step function and t_{i} is the arrival time of event. Note, that by using this definition, we count only events that fall within the considered time window \Delta t_{j} , therefore s_{i}=0 outside of it. Similar to the signal PDF, the background PDF also contains space and time part, so it can be written as b_{i}=b_{i}^{\mathrm{space}}b_{i}^{\mathrm{time}}. (4) To define the spatial and temporal background PDFs we use the total solid angle \Delta\Omega subtended by the considered part of the sky, and the length of the entire data taking period, or more precisely, the uptime \Delta T_{\mathrm{data}} . The resulting PDFs are b_{i}^{\mathrm{space}}=1/\Delta\Omega and b_{i}^{\mathrm{time}}=1/\Delta T_{\mathrm{data}} . Both the signal and background PDFs are combined to evaluate the likelihood over all observed events (numbered by index i ) such that \mathcal{L}(n,\Delta t_{j},\vec{r}_{s})=\prod_{i=1}^{N}\Bigg(\frac{n}{N}s_{i}+(1-\frac{n}{N}){b}_{i}\Bigg)\rm{,} (5) where n is the assumed number of signal events present in the cluster and N is the number of all events in the considered data sample. Following this equation, a likelihood representing absence of signal events \mathcal{L}(0,\Delta t_{j},\overrightarrow{r_{s}}) , i.e. containing only the background, can also be calculated. Then by comparing these two likelihoods, a test statistic evaluating the significance of a given multiplet, can be constructed: \rm{TS}(n)=-2\cdot log\Bigg[\frac{\mathcal{L}\left(0,\Delta t_{j},{\vec{r}_{s}}\right)}{\mathcal{L}(n,\Delta t_{j},\vec{r}_{s})}\Bigg]. (6) The \rm{TS}(n) should be then maximized against n to obtain an estimate of the number of signal events, denoted by n_{s} . Note that by construction n_{s} is a real number and not necessarily an integer, because TS is maximized over the entire range of real numbers rather than over a discrete set of integers. However, we can expect integers to occur more frequently, thus forming multi-peak distributions of n_{s} . The TS distribution obtained from many scrambled maps is used to evaluate the significance level of the results. 3 The stacking method We propose an improved method of the space-time clustering analysis ( Góra et al., 2011 ) , which involves a slight modification of the standard likelihood approach ( Braun et al., 2008 ) described in the previous section. This improved method is based on doublet stacking analysis, offering significantly faster performance than the standard method, and making it sensitive to multiple weak flares of arbitrary shapes.

Read the complete text