Multimessenger Astronomy with the Pierre Auger Observatory: Recent Results

uni-siegen.de 20. März 2023 Multimessenger Astronomy with the Pierre Auger Observatory: Recent Results Marcus Niechciol1on behalf of the Pierre Auger Collaboration2 1 Center for Particle Physics, University of Siegen, Germany 2 Observatorio Pierre Auger, Malargüe, Argentina XXI Workshop on Neutrino Telescopes Padova, Italy, 1 October 2025 Introduction •Key question: what is the origin and nature of ultra-high-energy cosmic rays (UHECRs)? •UHE (𝐸 ≳ 10!" eV) neutrinos and photons play a crucial role in this endeavor •Neutral particles: point right back at their source •Intimate connection between UHE neutrinos/photons and UHECRs: Can be produced either directly at the sources of UHECRs or during their propagation through the Universe, e.g., in interactions with the CMB Multimessenger Astronomy 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 3 Source Earth Cosmic Ray Neutrino /Photon •Neutrinos and photons entering the Earth’s atmosphere can initiate extensive air showers, just like charged cosmic rays – making indirect detection possible •Main challenge: distinguishing neutrinoand photon-induced air showers from the vast background of showers initiated by cosmic protons and heavier nuclei •In a nutshell: •Searching for UHE photons means looking for deeper (vertical) showers with fewer muons Needles in the Cosmic Haystack 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 4 Top of the atmosphere EM component Hadronic component Cosmic Ray (Vertical) Muons Earth Top of the atmosphere EM component Hadronic component Photon (Vertical) Muons Earth •Neutrinos and photons entering the Earth’s atmosphere can initiate extensive air showers, just like charged cosmic rays – making indirect detection possible •Main challenge: distinguishing neutrinoand photon-induced air showers from the vast background of showers initiated by cosmic protons and heavier nuclei •In a nutshell: •Searching for UHE photons means looking for deeper (vertical) showers with fewer muons •Searching for UHE neutrinos means looking for inclined showers with an electromagnetic component –or slightly upgoing showers coming from Earth-skimming 𝜏neutrinos Needles in the Cosmic Haystack 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 4 Top of the atmosphere EM component Hadronic component Cosmic Ray (Inclined) Earth Muons Top of the atmosphere EM component Hadronic component Neutrino (Inclined) Muons Earth Top of the atmosphere EM component Hadronic component Earth-Skimming tau Neutrino Muons Earth tau lepton •Located near Malargüe, Argentina •Surface detector array (SD) •~1660 water Cherenkov detectors (WCDs) covering a total area of ~3000 km2 •Measuring secondary particles on ground •Fluorescence detector (FD) •4 stations with 27 telescopes, overlooking the SD •Measuring the longitudinal development in the atmosphere •Further detector systems complementing the main SD and FD (e.g., buried muon counters, radio antennas) •Currently ongoing: detector upgrade AugerPrime •Main goal: increase composition sensitivity Pierre Auger Observatory 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 5 [Pierre Auger Coll., NIM A 798 (2015) 172] Searches for UHE Photons and Neutrinos 7 Search for Photons Between 1018 and 1019 eV 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) •The observable Fµis used as a proxy for the muon content, calculated using a model based on air-shower universality •Overall background rejection ~99.9% •Data period: 1 Jan 2005 –31 Dec 2017; ~1000 km2sr yr •22 candidate events, consistent with the expectation of 30±15, estimated using data [Pierre Auger Coll., PRD 110 (2024) 062005] •Analysis uses data from the Underground Muon Detectors (UMD) and the SD-433 •Key observable Mbis based on the measured (high-energy) muon densities; expected background contamination smaller than 10-5 assuming pure-proton background •Data set spans 16 months, exposure: ~0.6 km2sr yr •Total sample split in six categories, mainly depending on the number of available detector stations •No event passes the photon candidate cut Search for Photons Between 5×1016 and 2×1017 eV 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) PoS(ICRC2023)238 Photon search at 50 PeV with Auger Nicolás González angle ranges is presented in Fig. 2. The energy-integrated trigger efficiency, weighted by a spectrum ⇢3 W, is convolved with the aforementioned contributions to adjust the exposure, as shown in Eq. 1. This correction is approximately 12% (less than 2%) above ⇢W=1016.7eV (1017 eV). To ensure a good trigger efficiency, the data-set above ⇢W=1016.7eV (refer to Sec. 3for the definition of ⇢W) and up to a zenith angle of 52 is used in this study. /eV) γ (E 10 log 16 16.2 16.4 16.6 16.8 17 17.2 17.4 Array trigger efficiency 0 0.2 0.4 0.6 0.8 1 = 0 - 23°/θ = 23 - 34°/θ = 34 - 43°/θ = 43 - 52°/θ Figure 2: The trigger efficiency as a function of the simulated energy for photon primaries and different zenith angle intervals. Solid lines correspond to fitted sigmoid functions. The selected data-set begins on December 17, 2020, when the central UMD stations of the three hexagons were deployed and operational. It extends until March 31, 2022, resulting in 15,919 recorded events during ⇠15.5 months. The accumulated exposure in the full-triggerefficiency regime is nW=0.629 km2sr yr. 3. Event-based muon estimate with Sb The number of muons detected at ground level serves as a well-established observable for identifying air showers characterized by a dominant electromagnetic development. The highenergy muon footprint is directly measured by the UMD stations. An example of an acquired event is illustrated in Fig. 3(left). We define the event-based observable "1using the measured muon density d8 `at a distance A8from the shower axis [12]: "1=log10 ’ 8 d8 ` dp `⇥⇣A8 200 m⌘1!(2) The choice of 1=1has been found to provide the best separation power between photons and protons at tens of PeV. The discrimination observable "1is characterized with the simulated events described before. A reference distance of 200 m is chosen based on the fluctuations in the number of muons. These relative fluctuations exhibit a minimum plateau between 50 and 300 m in the energy range relevant to this work. Beyond this range, the uncertainties increase rapidly due to the inherent Poissonian fluctuations associated with the decreasing number of muons. The normalization factor dp `is defined as the average muon density at 200 m from the shower axis in simulated proton events. This quantity is parameterized using an unbinned maximum likelihood approach, as illustrated in Fig. 3(right). The likelihood model accounts for the zenith-dependent atmospheric attenuation of the muon component. As a result, events from hadronic origin would typically yield positive values of "1, while events with photon origin exhibit negative values (Fig. 4, left). The energy reconstruction method optimized for hadronic showers tends to underestimate the energy in the case of photon primaries due to their delayed shower development and the lack of a significant muon component. Still, a single energy scale needs to be employed for all events, irrespective of the primary particle species. As a solution, a photon energy scale is customized. It consists of an attenuation function, 5att, parameterized using simulated proton events, and an energy 4 [Pierre Auger Coll., JCAP 05 (2025) 061] JCAP05(2025)061 1 M 2.5−2−1.5−1−0.5−0 0.5 1 1.5 Events 1 10 2 10 3 10 Photon Proton b 1−0.5−0 0.5 1 1.5 2 2.5 3 3.5 4 Background contamination 8− 10 7− 10 6− 10 5− 10 4− 10 3− 10 2− 10 [16.7,16.9]∈/eV) ,eqγ lg(E [16.9,17.1]∈/eV) ,eqγ lg(E [17.1,17.3]∈/eV) ,eqγ lg(E Figure 9. Left: The Mb=1 distributions for simulated photonand proton-initiated events for energies lg (Eω,eq/eV)→ (16 . 7 , 16 . 9). The uncertainty in bins with fewer than ten entries is the asymmetric confidence belt calculated with the Feldman-Cousins method at 95% confidence level [ 67 ]. The fit to the proton tail employed to estimate the background contamination is displayed as a black band. Right: The average background contamination for the quoted energy bins in terms of b for a fixed signal e!ciency of 50%. Signal efficiency 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Background contamination 10− 10 9− 10 8− 10 7− 10 6− 10 5− 10 4− 10 3− 10 2− 10 1− 10 [16.7,16.9]∈/eV) ,eqγ lg(E [16.9,17.1]∈/eV) ,eqγ lg(E [17.1,17.3]∈/eV) ,eqγ lg(E [17.3,17.5]∈/eV) ,eqγ lg(E Figure 10. The background contamination in terms of the signal e!ciency for the quoted energy bins in the photon-equivalent scale. We note that the background contamination and signal e!ciency shown in figure 10 are obtained assuming a complete UMD hexagon, e.g., the north-west hexagon in figure 1, left. We verified that the choice of b =1remains e"ective even when considering missing UMD stations in the first ring of neighbors. However, the discrimination power, i.e. both the background contamination and signal e!ciency, varies depending on the real configurations of the detector, as discussed in the next section. 5 Selection of photon events in data In this section, we study the application of the discrimination method to data, considering the actual configuration of the detector. Given that the fraction of photon primaries expected to be present in the cosmic-ray flux may be well below 10 →5 , the standard event reconstruction, fine-tuned to hadronic events, is employed for the data. The event selection follows the – 14 – 8 Gravitational Wave Follow-Up Searches •Routine follow-up of gravitational-wave (GW) event alerts •Search for associated neutrinos in the SD dataset with a latency of at most 15 minutes, in the Earthskimming and down-going channels – so far, no neutrinos identified •One example: GW170817 (binary neutron star merger) •Source perfectly within the FoV of the ES channel at the time of the event •Auger limits complement those of IceCube and ANTARES Follow-Up of Gravitational-Wave Events 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 15 [ANTARES, IceCube, Pierre Auger, LIGO Scientific and Virgo Colls., ApJL 850 (2017) L35] > / •Stacking analysis of all binary black hole mergers observed by LIGO/Virgo in runs O1-O3 to probe the UHE neutrino luminosity of such mergers •Assumptions: constant luminosity for all mergers with an 𝐸$ %& spectrum and constant emission during two hypothetical emission periods of 24 hours and 60 days after each merger Limits on UHE Neutrinos from Binary Black Hole Mergers 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 16 PRELIMINARY [Pierre Auger Coll., in preparation], [M. Schimp (Pierre Auger Coll.), PoS (ICRC 2021) 968] •Stacking analysis of all binary black hole mergers observed by LIGO/Virgo in runs O1-O3 to probe the UHE neutrino luminosity of such mergers •Limit on the energy emitted in UHE neutrinos per source (at 90% C.L.): ~2.3 × 10'( erg (about 5% of the total energy typically emitted in a GW) Limits on UHE Neutrinos from Binary Black Hole Mergers 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 17 [Pierre Auger Coll., in preparation], [M. Schimp (Pierre Auger Coll.), PoS (ICRC 2021) 968] PRELIMINARY •Photon detection efficiency better than for neutrinos, but identification more difficult •Only follow-up selected GW alerts to reduce the rate of false-positive detections •Focus on close and/or well-localized GW events measured by LIGO/Virgo: 10 GW events pass the selection; look at time windows of ±500 sand +1 day around the time of the merger •No coincident photons were identified in the SD data set (above 1019 eV) for any of the 10 GW events: determine upper limits on the spectral fluence •Closer look at GW170817: less than 20% of the total energy transferred into photons above 40 EeV Follow-Up Search for UHE Photons 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 18 [Pierre Auger Coll., ApJ 952 (2023) 91] GW150914 GW190517 GW190728 -2 spectral fluence upper limit / MeV cm 6 6.5 7 7.5 8 8.5 directional uncertainty of GW source [-2.3,-1.7]∈ α-flux spectral index γ GW170817 GW170818 GW190701 GW190814 GW200208 GW200224 GW200311 -2 spectral fluence upper limit / MeV cm 25 30 35 40 45 50 55 60 directional uncertainty of GW source [-2.3,-1.7]∈ α-flux spectral index γ +1 day ±500 s Further Multimessenger Studies •The Pierre Auger Collaboration takes part in several multimessenger astronomy networks (GCN/TAN, AMON, ACME) •Enables direct follow-up studies to transient events by reacting to the corresponding alerts, one example: search for neutrinos from the flaring blazar TXS 0506+056 Follow-Up Studies to Transient Events 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) corresponds to 7.5 yr, the whole observation time that the IceCube detector had been in operation at the time of detection. We here address similar scenarios of half a year and the whole observation period of the Pierre Auger Observatory, which is 15 yr from 2004 January 1 to 2018 August 31. We note that periods over which the SD was unstable have been removed from the analysis and that during the first four years of operation the effective area was a rapidly growing function of time because the Observatory was under construction until 2008 June. The average spectral fluxes of UHE neutrinos with a fixed spectral index (~gE)that would produce a single event at the Observatory for these two periods are displayed in Figure 3for a spectral index of γ=2.0, assumed to hold in the energy range between 100 PeV and 10 EeV and to be constant in time during the corresponding time period. In this plot they are compared to the fluxes obtained from the neutrino detected in 2017 September 22 and inferred to have energy of order few hundred TeV, considering a period of half a year and 7.5 yr. The plot also displays the average VHE gamma-ray flux detected with Fermi-LAT and MAGIC over periods within a couple of weeks around the neutrino detection date of 2017 September 22 (Aartsen et al. 2018b). These gamma-ray fluxes correspond to the reported flaring activity and have not been corrected for absorption in the extragalactic background light. They are considerably larger than the average gamma-ray Figure 2. Hours per day a source is visible in each of the search channels as a function of decl. The decl. of TXS 0506+056 is marked with an arrow. Figure 3. UHE flux reference that would give one expected neutrino event at the Pierre Auger Observatory over a period of half a year (2017 March 22–September 22) for a spectrum µd NdE E2in comparison to the flux that would produce on average one detection like the IceCube-170922 A event over the same period (solid red and black lines). Flux references are also shown for the Pierre Auger Observatory for a period of ∼15 yr during which it has taken data (2004 January 1–2018 August 31)and for a period of 7.5 yr for IceCube (Aartsen et al. 2018b; dashed red and black lines). The average VHE and UHE photon fluxes measured with Fermi-LAT and MAGIC around 2017 September 22 (Aartsen et al. 2018b), and the archival photon measurement from Fermi-LAT (Acero et al. 2015), as well as the UHE photon flux from this direction that would give one expected photon event in half a year at the Pierre Auger Observatory, are also shown for comparison. 6 The Astrophysical Journal, 902:105 (8pp), 2020 October 20 Aab et al. fluxes that had been recorded to date from this source, and which are also illustrated for comparison (Acero et al. 2015). The sensitivity of the Auger Observatory to UHE neutrinos is about an order of magnitude below extrapolations with E −2 spectra, partly due to the nonoptimal position of the source. We have also compared the sensitivity of the Observatory to the neutrino flux observed by IceCube between 2014 October 19 and 2015 February 6. The analysis of this period resulted in constraints for the normalization and spectral index of the observed fluence (Aartsen et al. 2018a). This period of increased neutrino flux in IceCube was not coincident with a VHE gamma-ray flare from the same source, although a hardening of the spectrum in the GeV region was reported (Padovani et al. 2018). In Figure 4we display the 1σand 2σbands of the average flux obtained from the fluence reported assuming an activity period of 110 days as obtained from the IceCube data analysis using a Gaussian window. The bands are calculated using the whole parameter space allowed at 68% and 95% confidence levels in the IceCube analysis. The extreme values of the spectral index are γ∼1.75 and γ∼2.45 (∼1.5 and ∼2.7)for the 68% (95%)CL contour plot (Aartsen et al. 2018a). The figure also displays the average gamma-ray flux obtained for this period illustrating the reported hardening (Padovani et al. 2018). The results obtained indicate that the Pierre Auger Observatory could only be expected to have detected a signal if the flux extrapolated to the EeV regime with spectral indices harder than γ∼1.5. With the Pierre Auger Observatory it is also possible to search for UHE photons (Aab et al. 2016,2019c; Niechciol et al. 2017). For a source as distant as TXS 0506+056, any UHE photon flux that could have been produced is expected to be strongly attenuated through interactions with the cosmic photon-background fields, unless new physics would occur. The data have been searched for UHE photons between 10 and 300 EeV in coincidence with IceCube-170922 A over a period of half a year and also in coincidence with the 110 day interval interpreted by Aartsen et al. (2018a)as a burst of neutrinos. No event has been found with an angular distance to the source below 2°. The shower with closest angular distance to the source (2°.1)was observed for the latter period and the corresponding value of the Principal Component (PC)for photon discrimination (Rautenberg et al. 2019)is very low, so that less than 0.1% of the simulated photons have a smaller PC value. As a result, the probability of this event to be a correlating photon is less than 6×10 −5 . Assuming an E −2 spectrum, the photon energy flux that would give one expected photon event at the Observatory is ´- 1.8 10 12 ergcm −2 s −1 . For the half a year period in 2017 the closest event, at an angular distance of 3°.0, has an even lower probability to be a photon, and the reference energy flux for one detected photon becomes 1.0×10 −12 ergcm −2 s −1 . In summary, we have studied the implications of the nonobservation of UHE neutrinos with the Pierre Auger Observatory. The source is not located at one of the preferential declinations for observation so the flux constraints that can be obtained are rather limited. The neutrino flux from TXS 0506 +056 at hundreds of TeV sampled by IceCube with event IceCube-170922 was converted to a flux using a half a year period (Aartsen et al. 2018b). If the flux from the source had an E −2 spectrum extending to the EeV and if it had remained constant over the lifetime of the Observatory with the same normalization, one neutrino event could be expected to have been observed at the Pierre Auger Observatory. We have also shown that the Observatory could have a chance to detect UHE neutrinos produced between 2014 October and 2015 February only in a case in which the spectrum extended to the EeV range with a spectral index harder than γ∼1.5 (Aartsen et al. 2018a). Figure 4. UHE neutrino flux sensitivities for the Pierre Auger Observatory (one event expected)assuming a constant flux during a period of 110 days from 2014 October 19 to 2015 February 6 in comparison to the measured photon flux (Padovani et al. 2018)and to the neutrino flux inferred with IceCube during the same period with a spectral index of γ=2.1±0.2 (Aartsen et al. 2018a). The band shown for IceCube is obtained using the extreme values of γ(∼1.75, ∼2.45)from the given 1σcontour plot and (∼1.5 and ∼2.7)from the 2σcontour. 7 The Astrophysical Journal, 902:105 (8pp), 2020 October 20 Aab et al. [Pierre Auger Coll., ApJ 902 (2020) 105] 20 [https://gcn.nasa.gov/] [https://www.amon.psu.edu/] [https://www.acme-astro.eu] •Use the absence of neutrinos in our dataset to place limits on the neutrino fluence from GRBs •Select GRBs with known spectral data measured by gamma-ray satellites within the Auger FoV (ES and DGH channels): 140 + 430 GRBs selected •For each GRB, simulate the prompt neutrino emission according to the NeuCosmA model (NB: internal shock model, only taking into account accelerated protons) and compare this to the stacking limit for 570 GRBs UHE Neutrinos and GRBs [Y. Lema Capeans (Pierre Auger Coll.), PoS (ICRC 2025) 1093] 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 21 [S. Hümmer et al., ApJ 721 (2010) 1538] •Use the absence of neutrinos in our dataset to place limits on the neutrino fluence from GRBs •Select GRBs with known spectral data measured by gamma-ray satellites within the Auger FoV (ES and DGH channels): 140 + 430 GRBs selected •For each GRB, simulate the prompt neutrino emission according to the NeuCosmA model (NB: internal shock model, only taking into account accelerated protons) and compare this to the stacking limit for 570 GRBs UHE Neutrinos and GRBs 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 21 [S. Hümmer et al., ApJ 721 (2010) 1538] For more information, see Therese‘s talk tomorrow in the parallel session 10°8 10°7 10°6 10°5 10°4 10°3 10°2 10°1 F(E∫) [GeV cm°2] 1013 1014 1015 1016 1017 1018 1019 E∫[eV ] 10°14 10°13 10°12 10°11 10°10 10°9 10°8 10°7 ©(E∫)£GeV ·cm°2·s°1·sr°1§ fp 1 10 100 1000 Pierre Auger average GRB flux Pierre Auger 90% C.L. Limit (2004 - 2021) : 570 GRBs Auger Uncertainty flux ANTARES average flux ANTARES 90% C.L. Limit (2007-2017) : 784 GRBs ANTARES Uncertainty flux IceCube 90% C.L. Limit 1172 GRBs IceCube IS Model Pierre Auger sensitive energies [Y. Lema Capeans (Pierre Auger Coll.), PoS (ICRC 2025) 1093] 10°8 10°7 10°6 10°5 10°4 10°3 10°2 10°1 F(E∫) [GeV cm°2] 1013 1014 1015 1016 1017 1018 1019 E∫[eV ] 10°14 10°13 10°12 10°11 10°10 10°9 10°8 10°7 ©(E∫)£GeV ·cm°2·s°1·sr°1§ fp 1 10 100 1000 Pierre Auger average GRB flux Pierre Auger 90% C.L. Limit (2004 - 2021) : 570 GRBs Auger Uncertainty flux ANTARES average flux ANTARES 90% C.L. Limit (2007-2017) : 784 GRBs ANTARES Uncertainty flux IceCube 90% C.L. Limit 1172 GRBs IceCube IS Model Pierre Auger sensitive energies •ANITA reported two “anomalous” events consistent with upward-going showers •Debate about interpretation: Observational artefact? Background? BSM physics? •Search for upward-going air showers with 14 years of FD data •1 event has been found, compatible with the expected background (0.27 ± 0.12) •Overall strong disagreement with the interpretation of the anomalous events as upward-going showers •Tabulated exposure allow to use the resulting upper limits to provide constraints to various BSM scenarios Search for Upward-Going Air Showers 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) [ANITA Coll., PRL 117 (2016) 071101], [ANITA Coll., PRL 121 (2018) 161102] [Pierre Auger Coll., PRL 134 (2025) 121003] [B. Yue (Pierre Auger Coll.), PoS (ICRC 2023) 1095] 22 on lower Monte Carlo statistics and using different energy ranges have been presented in [45,47]. To relate the nonobservation of upward-going showers at the Auger Observatory to the observation of the anomalous events in ANITA, we calculate 2D exposure maps for both observatories as a function of shower energy and shower starting point in the atmosphere. This is done for three zenith angle ranges: 110°–130°, 130°–145°, and 145°–180° and is depicted in Fig. 3(top) for the first angular bin that largely overlaps with the ANITA anomalous events. The acceptance of the ANITA detector to upward-going showers has been calculated using an analytical approach integrating the surface area and solid angle over a spherical surface concentric with the Earth at altitude h. The accepted solid angle is approximated by the maximum and minimum off-axis angles (of the radio pulse relative to the shower axis) within which the recorded peak amplitude is seen at the detector above a fixed value. These angles are obtained with a set of proton simulations at 0.1 EeV starting at h between 0 and 9 km (above the ice surface) for zenith angles in the range 90°–130° [6]. The pulses are assumed to scale linearly with shower energy. Threshold values of 446 and 284 μVm −1have been used respectively for ANITA I and III instruments [6]. The exposure for the 110°–130° range in zenith angle has been obtained by multiplying the effective area by the effective flight time, which is 17 (7) days for the ANITA I (III) flight [3,48]. It is displayed in Fig. 3(bottom) for ANITA III and has been cross-checked modifying a Monte Carlo simulation developed for the calculation of the exposure to tau neutrinos in the ANITA IV flight [49]. (Several authors of this Letter are members of the ANITA Collaboration and have been involved in these calculations.) The two exposures increase with the energy of upward-going showers. That of the Auger Observatory can only be calculated for showers starting at low altitudes, h, because it falls very rapidly with h, and the required simulation statistics become unfeasible to produce, particularly at lower energies. The sums of the exposure bins in hillustrate that the sensitivity of the Auger Observatory exceeds that of the ANITA III flight by factors rising with energy from about 2 to 2000 for a uniform h distribution. Assuming that the anomalous ANITA events are indeed produced by upward-going showers, we can then calculate the number of upward-going showers expected in the data of the Auger Observatory by convolving a given spectral flux and an hdistribution of the showers, with the two 2D exposure maps. Three power law spectra, E−γ, with γ¼2, 3, 5 have been assumed, and for each case, we consider both a uniform distribution in hand that expected for exiting taus that decay in the atmosphere. In the latter case, expected from tau-neutrino interactions and in many of the proposed beyond the SM scenarios, the energy Eis that of the tau leptons. The starting altitude of the shower strongly depends on zenith angle and tau energy. The required distribution in Eand his obtained from a convolution of the tau flux, the decay-length distribution and the distribution of shower energy for all tau decays as obtained with TAUOLA [50], with an average of ∼50% of the tau energy. A given energy spectrum can be normalized to the anomalous observations demanding one expected event after folding with the ANITA I or III exposures. Normalizing to ANITA I (III) observation under the assumption of an E−3spectrum and a uniform hdistribution, we expect 59 (69) events at the Auger Observatory. Using an hdistribution compatible with tau decay reduces expectations to 37 (34) events. Assuming a very conservative spectrum, E−5, to the event of flight I (III) results in 11.7 (8.1) expected events for a uniform h distribution and 18 (11) events for the hdistribution expected from tau decay. These numbers are to be compared with one observed event compatible with background. We note that, given a spectral index, the expected number of events obtained using normalizations 1 4 16 33 59 156 208 363 440 714 859 1101 5 9 23 40 55 198 333 461 749 977 2 15 4 32 125 190 379 529 782 24 30 72 82 273 418 771 4 5 43 57 212 377 604 9 60 123 298 558 4 15 70 237 463 10 60 126 266 4 56 144 192 8 15 77 154 5 9 42 116 14 62 55 5 38 4 19 4 5 8 17 17.2 17.4 17.6 17.8 18 18.2 18.4 0 1 2 3 4 5 6 7 8 9 h (km) Pierre Auger Observatory ! (km2 sr yr) log(E/eV) 10–1 1 10 102 103 to quantify exposure 1 4 21 44 101 229 325 819 1208 2381 3923 6113 0.01 0.04 0.09 0.12 0.15 0.18 0.20 0.22 0.25 0.27 0.30 0.34 0.01 0.05 0.09 0.12 0.14 0.17 0.19 0.21 0.23 0.26 0.29 0.33 0.02 0.06 0.09 0.11 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.30 0.02 0.06 0.09 0.11 0.13 0.15 0.16 0.18 0.20 0.22 0.24 0.27 0.03 0.06 0.08 0.10 0.12 0.14 0.15 0.17 0.18 0.20 0.22 0.25 0.03 0.06 0.08 0.10 0.11 0.13 0.14 0.16 0.17 0.19 0.21 0.24 0.03 0.06 0.08 0.09 0.10 0.12 0.13 0.14 0.16 0.17 0.19 0.22 0.04 0.06 0.07 0.08 0.09 0.11 0.12 0.13 0.14 0.15 0.16 0.18 0.04 0.05 0.07 0.08 0.09 0.10 0.11 0.11 0.12 0.13 0.15 0.16 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.10 0.11 0.12 0.13 0.15 0.04 0.05 0.06 0.06 0.07 0.08 0.09 0.09 0.10 0.11 0.12 0.13 0.03 0.04 0.05 0.06 0.06 0.07 0.08 0.08 0.09 0.10 0.11 0.12 0.03 0.03 0.04 0.05 0.06 0.06 0.07 0.07 0.08 0.09 0.09 0.10 0.02 0.03 0.04 0.04 0.05 0.05 0.06 0.06 0.07 0.08 0.08 0.09 0.02 0.02 0.03 0.04 0.04 0.05 0.05 0.06 0.06 0.07 0.08 0.08 0.01 0.02 0.03 0.03 0.03 0.04 0.04 0.05 0.05 0.06 0.07 0.08 0.01 0.02 0.02 0.02 0.03 0.03 0.04 0.04 0.05 0.05 0.06 0.07 0.01 0.01 0.02 0.02 0.02 0.03 0.03 0.03 0.04 0.04 0.05 0.05 0 1 2 3 4 5 6 7 8 9 h (km) 17 17.2 17.4 17.6 17.8 18 18.2 18.4 log(E/eV) ! (km2 sr yr) 10–1 10–2 ANITA III 0.43 0.77 1.07 1.31 1.52 1.72 1.92 2.11 2.32 2.55 2.82 3.16 FIG. 3. Exposure of the Auger Observatory (top) and the ANITA III flight (bottom) as a function of shower energy, E, and injection altitude, h, integrated over the zenith angle range 110°≤θ≤130°, for an isotropic distribution of arrival directions. The gray area indicates insufficient statistics. In the white cells on top of the exposure plots, we display the sums of the h bins to facilitate comparison (see text). PHYSICAL REVIEW LETTERS 134, 121003 (2025) 121003-8 60°–100° (where all the background was found in the first batch; cf. [36]. The energy distribution of the full background sample is weighted to mimic the measured UHECR spectrum [30,42]. To study the signal, an isotropic distribution of 6×107 upward-going proton showers has been similarly simulated in the range 0.03–10 EeV. We note that, due to shower universality [43], protons stand in for arbitrary primaries with minimal or no loss of generality, since the first interaction points are directly set and all calculations use the shower energy rather than the energy of the primary particle. The combined efficiency for triggering and reconstructing such showers was found to be negligible for energies below 0.03 EeV. The simulated showers are forced to develop at a uniformly distributed altitude above the ground (the altitude of the observatory is taken to be 1400 m above sea level), h, in the range 0<h<9km with zenith angles, measured at the exit point on the Earth, between 110° and 180°. Altitudes h>9km are not considered because the exposures of Auger and ANITA fall down rapidly here. This is mostly for geometrical reasons (maximum elevation angles of the Auger FD telescopes and a narrowing Cherenkov cone in case of ANITA). The flat distribution of shower starting points used in the simulations is no restriction, as any distribution can be generated from this by applying corresponding weights to the altitude bins (see below). The ground impact (exit) points have been sampled in a square area of 100 ×100 km2centered on the SD array. This area extends up to ∼20 km behind each FD site to include simulated trajectories with exit points behind the field of view of a telescope [36]. Additionally, to increase statistics in the low energy region, 5×106proton showers have been simulated between 0.1 and 0.3 EeV with impact points contained in a circle with a radius increasing with energy from 12 to 23 km around HEATs but otherwise with the same distributions. The simulation of the FD signals and trigger, and the subsequent event reconstruction, are done within OFFLINE to study the performance of the reconstruction algorithms. The reconstructed zenith angles correlate well with their true value. However, as no cuts targeted to directional reconstruction precision were applied, a tail in θrec −θsim is present leading to a 68% central interval of ½−1.1°;11°"(see Ref. [36] for further details). The selection of candidates compatible with upwardgoing showers that exit the Earth’s surface was performed making use of simulations to reduce the large background to a minimal level. After deciding the entire selection strategy, it was applied to the aforementioned 10% burn sample for verification (see Fig. 1) before it was finally applied blindly to the full FD dataset (7.6×106events). In the first step, the aforementioned laser cuts reduce the data sample to 4.7×106events. To guarantee a minimum data quality [31], only time periods with a clean atmosphere and low cloud coverage [44], and only events with at least six camera pixels contributing to the time-geometry fit of the shower axis [28] are considered, leaving ∼600 k events. Out of these, 165 000 events can be reconstructed as upward-going in a simple time-geometry fit [28], if only the time sequence and pointing direction of the triggered pixels are considered. The GF reconstruction is used to check whether the pixel intensity is consistent with a showerlike dE=dX-profile, eliminating many events where the dominant signal is from Cherenkov light. Because of their time-compressed structure they are misreconstructed as upward-going by the simple time fit; cf. [36]. Only 2774 events survive this step. The GF is then also applied in the downward mode and it is found that the majority of the selected events allow both upward and downward reconstructions. Only 986 events are left when two further quality constraints are applied ensuring that the interval of atmospheric slant depth, over which the shower profile is observed, exceeds 80 g cm−2and the reconstructed shower maximum is above ground. To eliminate background from misreconstructed and Earth-missing showers, a cut of θ>110° is applied leaving 928 events. Finally, events with χ2 up ≥1.2χ2 down are removed as the downward reconstruction is clearly preferred over the upward reconstruction (the final search criterion, discussed below, is based on comparing the more precise likelihood ratios), reducing the sample to 255 events (cf. Fig. 1). The effects of cuts on data and simulation are compared in the supplemental material. A search criterion is finally needed to optimize discrimination between upward-going showers (signal) and cosmicray events (background). For convenience, we use a function 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Discrimination Variable l Number of Events / bin 10–2 10–1 1 10 102Burn sample Full data Background simul. Signal simulation dN dE-CR spectrum dN dE E3-arbritary normalization Cut value lc Signal region FIG. 1. Distributions of the discriminating variable l, as defined in Eq. (1), for (i) a simulated isotropic background weighted and normalized with the measured UHECR spectrum in the energy range 1017 to 1020 eV [42] (red histogram with an exponential fit and its uncertainty band); (ii) the signal simulation with energy 1016.6to 1018.5eV weighted with an E−3spectrum and arbitrarily normalized to one event (blue histogram); and (iii) the data distributions, both for the 10% burn sample and the full dataset (open and filled symbols). The cut value lcdiscriminating signal and background is indicated by the vertical dashed line. PHYSICAL REVIEW LETTERS 134, 121003 (2025) 121003-6 [Pierre Auger Coll., PRL 134 (2025) 121003] Search for Photons Between 2×1017 and 1018 eV 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) ] -2 [g cm max X 500 600 700 800 900 1000 1100 Entries (normalized) 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 Data Photon Sim. (Test) Proton Sim. (Test) Photon Sim. (Training) Proton Sim. (Training) [VEM]) b (S 10 log 3−2−1−0 1 2 3 Entries (normalized) 0 0.2 0.4 0.6 0.8 1 1.2 Data Photon Sim. (Test) Proton Sim. (Test) Photon Sim. (Training) Proton Sim. (Training) β 0.4−0.2−0 0.2 0.4 0.6 Entries (normalized) 0 1 2 3 4 5 6 7 8Data Photon Sim. (Test) Proton Sim. (Test) Photon Sim. (Training) Proton Sim. (Training) 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0.25 stations N 0 2 4 6 8 10 12 14 16 18 20 Entries (normalized) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Data Photon Sim. (Test) Proton Sim. (Test) Photon Sim. (Training) Proton Sim. (Training) Observables MVA (BDT) inside the geometrical field of view of the fluorescence telescopes and gaps in the recorded tracks, which can appear, for example, for air showers crossing several telescopes, amount to less than 30% of the total observed track length. Finally, it is required that the uncertainty on the reconstructed photon energy E γ , defined as the calorimetric energy taken from the integration of the profile plus a missing-energy correction of 1% appropriate for primary photons (Aab et al. 2017a), is less than 20%. Since the precise knowledge of the atmospheric conditions is crucial for the hybrid reconstruction, events recorded during periods without information on the aerosol content of the atmosphere are not taken into account. To exclude events where the recorded profile may be distorted due to clouds over the Observatory, only events from known cloud-free periods are accepted. Events where no information on the cloud coverage is available from either the Lidar system installed at the FD site Coihueco (BenZvi et al. 2007)or infrared data from the GOES-12 satellite (Abreu et al. 2013)are excluded. Finally, the last selection criterion removes events where fewer than four of the six SD stations in the first 750 m hexagon around the station with the largest signal are active. Such cases can occur, e.g., in the border region of the array or when individual SD stations are temporarily offline and not taking data. In this case, the discriminating observables S b and N stations (see Section 4)can be underestimated, mimicking air showers initiated by photons. The numbers of events after each level of the event selection and the associated selection efficiencies are given in Table 1, excluding the burnt sample as mentioned before. The largest reduction occurs already at the geometry level. Here, the main contribution comes from the restriction of the acceptance to the area of the 750 m SD array, followed by the requirement that the events have to be reconstructed using the hybrid procedure. After all cuts, 2,204 events remain with a photon energy E γ above 2 ×10 17 eV. A large sample of simulated events has been used to study the photon/hadron separation by the observables used in this analysis, to train the multivariate analysis, and to evaluate its performance. Air-shower simulations have been performed with CORSIKA (Heck et al. 1998), using EPOS LHC (Pierog et al. 2015)as the hadronic interaction model. About 72,000 photon-induced and 42,000 proton-induced air showers in six bins of equal width in ([ ])Elog eV 10 between 10 16.5 and 10 19.5 eV, following a power-law spectrum with spectral index −1 within each bin, have been used. Zenith and azimuth angles of the simulated events were drawn from an isotropic distribution between 0°and 65°and from a uniform distribution between 0°and 360°, respectively. Although they do not have a significant impact on the development of photon-induced air showers at the target energy range below 10 18 eV, preshowering (Erber 1966; McBreen & Lambert 1981; Homola et al. 2007), and LPM effects (Landau & Pomeranchuk 1953; Migdal 1956)were included in the simulations. Only protoninduced air showers are used as background, as these are the most photon-like compared to air showers induced by heavier nuclei such as helium. Even though there are indications that the composition of UHECRs is getting heavier with energy (see, e.g., Yushkov & Pierre Auger Collaboration 2019), the assumption of a pure-proton background in the context of a search for UHE photons can be taken as a conservative worstcase assumption, since including heavier nuclei would always lead to a smaller estimate for the contamination in the final sample of photon candidate events. All simulated air-shower events are processed with the Auger Offline Software Framework (Argiro et al. 2007)for a detailed simulation of the detector response. In these simulations, the actual detector status of both the SD and the FD as well as the atmospheric conditions at any given time during the aforementioned data period are taken into account, leading to a realistic estimate of the detector response. Each simulated air shower is used five times, each time with a different impact point on the ground, randomly taken from a uniform distribution encompassing the region of the 750 m SD array, and with a different event time, which was randomly determined according to the on-time of the Coihueco and HEAT telescopes during the data period used in this analysis. All simulated events are finally passed through the same event selection as the events from the data sample. After the event selection stage, the simulated samples contain about 55,000 photon-induced events and about 35,000 proton-induced events. 4. Analysis The search for primary photons presented in this work exploits the well-known differences in air-shower development for photon-induced and hadron-induced air showers: on the one hand, air showers initiated by photons develop deeper in the atmosphere than those initiated by hadrons, and on the other hand, they exhibit a smaller number of muons at ground level (Risse & Homola 2007). The first difference can be quantified through Xmax, which can be directly measured with the FD. To complement the FD observable Xmax, we use another quantity determined from the data of the 750 m SD array, called S b , which is defined as follows (Ros et al. 2011): ()SS R 1000 m ,1 b i iib ⎛ ⎝⎞ ⎠  q where S i denotes the measured signal in the ith SD station at a perpendicular distance R i to the shower axis. The parameter S b has been chosen here as b=4 to optimize the photon-hadron separation in accordance with Aab et al. (2017a). By construction, S b is sensitive to the lateral distribution, which in turn depends on the depth of the air-shower development in the atmosphere and the number of muons. Hence, S b can be used to distinguish photonand hadron-induced air showers. In addition to Xmax and S b , the number of triggered SD stations N stations is also used in the analysis, as it has been shown in Aab et al. (2017a)that it can significantly improve the overall performance of the analysis. The distributions of Xmax,S b , and Table 1 Numbers of Events from the Data Sample (Excluding the Burnt Sample) Passing the Different Event Selection Levels and the Associated Selection Efficiencies Relative to the Preceding Level Total number of HeCo events: 557,944 ... After geometry level: 20,545 3.7% After profile level: 12,129 59.0% After atmosphere level: 4373 36.1% After S b level: 3873 88.6% E γ 2×10 17 eV: 2204 56.9% Note. See the text for explanations. 5 The Astrophysical Journal, 933:125 (11pp), 2022 July 10 Abreu et al. •Photon candidate cut chosen to ensure 50% signal efficiency, leading to ~99.9% background rejection •Data period: 1 Jun 2010 –31 Dec 2015 •Exposure to photons (from simulations): ~2.5 km2sr yr •No events pass the candidate cut [Pierre Auger Coll., ApJ 933 (2022) 125] •Analysis uses two observables, both based on ”benchmarks” obtained from data •No assumptions needed on the cosmic-ray composition •Combination in a Fisher analysis trained with photon simulations and a fraction of the data sample •Data period: 1 January 2004 – 30 June 2020, exposure: ~17000 km2sr yr •16 events pass the candidate cut, consistent with the background expectation Search for Photons Above 1019 eV with the SD 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) 19 19.1 19.2 19.3 19.4 19.5 19.6 19.7 19.8 19.9 20 /eV) γ (E 10 log 0.8− 0.6− 0.4− 0.2− 0 0.2 0.4 0.6 0.8 LDF L photon MC (no preshower) photon MC (preshower) Data - burn sample 19 19.1 19.2 19.3 19.4 19.5 19.6 19.7 19.8 19.9 20 /eV) γ (E 10 log 5− 0 5 10 15 20 ∆ photon MC (no preshower) photon MC (preshower) Data - burn sample 4−2−0 2 4 6 8 Fisher 2− 10 1− 10 1 10 2 10 Events photon MC (no preshower) photon MC (preshower) Data - burn sample Data - search sample [Pierre Auger Coll., JCAP 05 (2023) 021] Version September 27, 2022 submitted to Universe 11 of 20 the SD, the lack of a corresponding fluorescence measurement for the bulk of the data poses 233 some challenges. For example, there is no direct measurement of Xmax available. Also the 234 primary energy can only be accessed indirectly, using S( 1000 ) —the interpolated signal in 235 the SD stations at a perpendicular distance of 1000 m from the shower axis—as a proxy. 236 237 Two observables are used in this analysis, one related to the thickness of the shower 238 front at ground and one based on the steepness of the lateral distribution. These two 239 properties of an air shower depend on the type of the primary particle initiating the shower, 240 hence they can be used for photon-hadron separation. The first observable, D , is based 241 on the risetime t1/2 in the individual SD stations, which is defined as the time at which 242 the integrated signal in the FADC time trace rises from 10 % to 50 % of its total value. For 243 showers of the same primary energy and zenith angle, t1/2 is expected to be larger for 244 primary photons with respect to primary nuclei due to the reduced muonic content, which 245 implies larger scattering and attenuation of secondary particles, and to Xmax being closer 246 to the ground. Dis defined as: 247 D=1 NÂ i (ti 1/2 tbench 1/2 ) st1/2 , (3) which can be taken as the average deviation of the measured rise-times from a data bench248 mark tbench 1/2 , describing the average risetime of all of the SD data (assumed to be overwhelm249 ingly constituted by primary nuclei) [ 32 ], in units of sampling fluctuations st1/2 . Details 250 on the selection criteria for the SD stations can be found in [ 25 ]. By construction, D is 251 expected to average to zero for data and to be significantly positive for photon-induced 252 air showers. As photon-induced air showers are also expected to have a steeper lateral 253 distribution of the signals in the SD stations than the average of data, a second observable 254 LLDF is introduced to quantify the departure of the observed LDF from the average of all 255 SD data (see also [33]): 256 LLDF =log10 1 N N Â i=1 Si fLDF(ri)!, (4) where Si is the total signal measured in the i -th selected station and fLDF(ri) gives the 257 average signal, obtained from all SD data, for a station at the same distance ri from the 258 shower axis. The photon energy Eg is determined for each measured air-shower event 259 taking into account S( 1000 ) and the reconstructed zenith angle. For this purpose, a look-up 260 table has been constructed using a large simulation sample. Only non-preshowering photon 261 events are used, which are weighted according to a reference spectrum µE2 . In Fig. 5, 262 the distributions of the two observables are shown as a function of the photon energy. In 263 particular D shows a good separation between photons and data. Finally, the two variables 264 are combined using a Fisher discriminant analysis with the burnt sample representing the 265 background and photon simulations the signal. 266 267 The analysis is applied to SD data collected between 1 January 2004 and 30 June 2020. 268 Only air-shower events with a reconstructed zenith angle between 30  and 60  are taken 269 into account to ensure that the majority of possible selected photon-induced showers reach 270 their maximum development before reaching ground level. A number of selection criteria 271 are applied to ensure a reliable reconstruction of the two observables. These criteria are 272 described in detail in [ 25 ]. Overall, the data sample (search sample) consists of 48 , 061 273 selected events with a photon energy Eg 10 19 eV , excluding the burnt sample which 274 consists of 886 events (1.8 % of the total number of selected events). The results of the 275 analysis are shown in Fig. 5, bottom. 16, 2 and 0 events above energy thresholds of 1, 2 and 276 4 ⇥ 10 19 eV , respectively, had a value of the Fisher discriminant above the photon-candidate 277 cut, which was fixed to the median of the Fisher distribution for non-preshowering primary 278 photons (shown as the solid black line in Fig. 5, bottom). The number of observed candidate 279 events is in statistical agreement with what is expected from the fit of an exponential to 280 Version September 27, 2022 submitted to Universe 11 of 20 the SD, the lack of a corresponding fluorescence measurement for the bulk of the data poses 233 some challenges. For example, there is no direct measurement of Xmax available. Also the 234 primary energy can only be accessed indirectly, using S( 1000 ) —the interpolated signal in 235 the SD stations at a perpendicular distance of 1000 m from the shower axis—as a proxy. 236 237 Two observables are used in this analysis, one related to the thickness of the shower 238 front at ground and one based on the steepness of the lateral distribution. These two 239 properties of an air shower depend on the type of the primary particle initiating the shower, 240 hence they can be used for photon-hadron separation. The first observable, D , is based 241 on the risetime t1/2 in the individual SD stations, which is defined as the time at which 242 the integrated signal in the FADC time trace rises from 10 % to 50 % of its total value. For 243 showers of the same primary energy and zenith angle, t1/2 is expected to be larger for 244 primary photons with respect to primary nuclei due to the reduced muonic content, which 245 implies larger scattering and attenuation of secondary particles, and to Xmax being closer 246 to the ground. Dis defined as: 247 D=1 NÂ i (ti 1/2 tbench 1/2 ) st1/2 , (3) which can be taken as the average deviation of the measured rise-times from a data bench248 mark tbench 1/2 , describing the average risetime of all of the SD data (assumed to be overwhelm249 ingly constituted by primary nuclei) [ 32 ], in units of sampling fluctuations st1/2 . Details 250 on the selection criteria for the SD stations can be found in [ 25 ]. By construction, D is 251 expected to average to zero for data and to be significantly positive for photon-induced 252 air showers. As photon-induced air showers are also expected to have a steeper lateral 253 distribution of the signals in the SD stations than the average of data, a second observable 254 LLDF is introduced to quantify the departure of the observed LDF from the average of all 255 SD data (see also [33]): 256 LLDF =log10 1 N N Â i=1 Si fLDF(ri)!, (4) where Si is the total signal measured in the i -th selected station and fLDF(ri) gives the 257 average signal, obtained from all SD data, for a station at the same distance ri from the 258 shower axis. The photon energy Eg is determined for each measured air-shower event 259 taking into account S( 1000 ) and the reconstructed zenith angle. For this purpose, a look-up 260 table has been constructed using a large simulation sample. Only non-preshowering photon 261 events are used, which are weighted according to a reference spectrum µE2 . In Fig. 5, 262 the distributions of the two observables are shown as a function of the photon energy. In 263 particular D shows a good separation between photons and data. Finally, the two variables 264 are combined using a Fisher discriminant analysis with the burnt sample representing the 265 background and photon simulations the signal. 266 267 The analysis is applied to SD data collected between 1 January 2004 and 30 June 2020. 268 Only air-shower events with a reconstructed zenith angle between 30  and 60  are taken 269 into account to ensure that the majority of possible selected photon-induced showers reach 270 their maximum development before reaching ground level. A number of selection criteria 271 are applied to ensure a reliable reconstruction of the two observables. These criteria are 272 described in detail in [ 25 ]. Overall, the data sample (search sample) consists of 48 , 061 273 selected events with a photon energy Eg 10 19 eV , excluding the burnt sample which 274 consists of 886 events (1.8 % of the total number of selected events). The results of the 275 analysis are shown in Fig. 5, bottom. 16, 2 and 0 events above energy thresholds of 1, 2 and 276 4 ⇥ 10 19 eV , respectively, had a value of the Fisher discriminant above the photon-candidate 277 cut, which was fixed to the median of the Fisher distribution for non-preshowering primary 278 photons (shown as the solid black line in Fig. 5, bottom). The number of observed candidate 279 events is in statistical agreement with what is expected from the fit of an exponential to 280 A Very Deep Event 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) [Pierre Auger Coll., in preparation, arXiv:2406.07439] 𝐸 = 1.73 ± 0.16 EeV 𝑋max =1245 ±57 g cm%& Exposure to Neutrinos 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) PRELIMINARY [M. Niechciol (Pierre Auger Coll.), EPJ Web Conf. 283 (2023) 04003] Upper Limits on the Diffuse Flux of UHE Neutrinos 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) [Pierre Auger Coll., JCAP 10 (2019) 022] [M. Niechciol (Pierre Auger Coll.), EPJ Web Conf. 283 (2023) 04003] IceCube, PRL 135 (2025) 031001 ANITA, PRD 98 (2019) 022001 KM3NeT, Nature 638 (2025) 376 Constraints on the Characteristics of UHECR Sources [C. Petrucci (Pierre Auger Coll.), PoS (ICRC 2023) 1520] 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) PoS(ICRC2023)1520 Constraints on UHECR characteristics from cosmogenic neutrino limits Camilla Petrucci Figure 2: Constraints on UHECR source evolution models parameterized as /(1+I)<for sources distributed homogeneously up to a maximum redshift Imax and emitting protons following a power-law (⇢)/⇢2.5 up to ⇢max =6⇥1020 eV. The cosmogenic neutrino fluxes for each combination of <and Imax were obtained with SimProp. The color code shows different levels of C.L. exclusion reported on the I-axis. The solid and dashed lines represent the contours of 90% and 68% C.L. exclusion, respectively. Top panel: exposure from Jan 04 to Dec 21; Bottom panel: exposure from Jan 04 to Dec 35 (considering running of the observatory until Dec 35). has been fixed, and the parameters that describe the UHECR spectrum, e. g. spectral index and rigidity cutoff, have been obtained; in Sec. 3.3, we study the presence of a subdominant component at energies above the ankle and, for this reason, we scale the proton component in the primary mass composition; finally a mixed composition scenario has been analyzed in Sec. 3.4. We refer to the energy spectrum and mass composition data of ICRC 2019 reported in [10]. 3.1 Pure-proton scenario with fixed parameters In this first analysis, only a pure-proton scenario was considered in which the spectral index Wof the UHECR proton spectrum at the sources is fixed to 2.5, and the maximum energy is cut at ⇢max =6⇥1020 eV. We performed a comprehensive scanning of the parameters Imax (in the range [1.0,5.0]with steps of 0.1) and <(in the range [2.0,5.0]with steps of 0.1) of the source evolution function: for each pair, the cosmogenic neutrino flux was obtained and, consequently, the expected number of neutrino events considering the Eq. 3. In Fig. 2(top), the obtained exclusion plot is reported, corresponding to the exposure shown in Fig. 1(solid line). We observe that the parameter 4 GW Event Selection for the Photon Follow-Up 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) / Mpc) L (D 10 log 1.5 2 2.5 3 3.5 ) 2 / deg 50% Ω( 10 log 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 photon horizon estimateVery close sources ) 2 < 20degΩWell localized events ( crit Ω 2 < 720degΩ GWTC-1 GWTC-2.1 GWTC-3 [Pierre Auger Coll., ApJ 952 (2023) 91] UHE neutrinos and GRBs [Y. Lema Capeans] 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) Internal Shock Model sketch Neutrino Prompt emission from GRBs 4 / 10 Two approaches to neutrino fluences Neutrinos from CosmicRay Accelerators (NeuCosmA) ApJ, 721, 1538 (2010) Lia & Tamborra JCAP, 10, 054 (2024) Only accelerated protons Includes Nuclear Cascade Simulations per GRB Not simulated Only Internal Shock (IS) 3 different models : oICMART oPhotospheric oIS Uses measured and inferred parameters Fixed parameters oГ=300 otv= 0.5 s Higher normalization Lower normalization Multi-Hybrid Measurements of Air Showers 1 October 2025Marcus Niechciol (Pierre Auger Collaboration) / XXI Workshop on Neutrino Telescopes (Padova, Italy) AugerPrime – multi-hybrid measurements (David Schmidt) (Lukas Nellen) 32