Pith. sign in

REVIEW 3 major objections 4 minor 19 references

Search for ultra-high energy photons: observing the preshower effect with gamma-ray telescopes

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Pith's one-line read A simulation shows that preshower air showers from 40 EeV photons can be separated from the cosmic-ray background at a significance of 29.85 using next-generation gamma-ray telescopes.

desk verdict Novel but flawed proof-of-concept: the CTA preshower search is promising, but the headline BDT separation compares 40 EeV photons against <=3 PeV protons, so the significance is largely an energy artifact and the quoted sigma is not interpretable. read the letter →

arxiv 1908.08805 v2 pith:UB4HJ3U3 submitted 2019-08-23 astro-ph.IM

classification astro-ph.IM
keywords preshowereffectultra-high-energyphotonsgamma-rayastronomyCherenkovtelescopesboosteddecisiontreescosmic-raybackgroundgeomagneticfieldairshowers
topics Dark Matter
open problems Dark Matter
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that a screening effect in Earth's magnetic field could turn ultra-high-energy photons into detectable air showers, and that a next-generation gamma-ray telescope array could pick these out from the overwhelming cosmic-ray background. The authors simulate 40 EeV photons arriving from the geomagnetic north at an 80-degree zenith angle, where the conversion probability is highest. A boosted decision tree trained on Cherenkov image shapes separates simulated preshower showers from proton-initiated background showers with a maximum significance of 29.85, identifying 92.6% of preshower events while admitting 3.8% of background events. A point-source version performs even better. The catch is that known production models predict so few ultra-high-energy photons that a 30-hour run would most likely see zero events.

What carries the argument

The load-bearing mechanism is the preshower effect: a 40 EeV photon crossing the geomagnetic field has a conversion probability (here 0.67) of turning into an electron-positron pair far above the atmosphere, and the pair's bremsstrahlung emission produces a collimated cosmic-ray ensemble that later strikes the atmosphere as a single air shower. The classification is carried by a boosted-decision-tree classifier applied to four Hillas image parameters (size, width, length, distance). The high-zenith, north-pointing geometry is what makes the separation work, because the Cherenkov images at 80° zenith are dominated by muons, and the muon content differs between photon-initiated and hadron-initiated showers.

What would settle it

Train the same classifier on a background sample weighted to the true cosmic-ray intensity (the $E^{-2.7}$ proton spectrum at the chosen geometry and observation time) rather than on equal-sized Monte Carlo samples; if the signal efficiency at a fixed background rate drops below about 5σ significance, the claimed discrimination is not robust enough for real observations.

Watch

Extended reading notes

Core claim

The central claim is that preshower-induced air showers, produced when an ultra-high-energy photon converts to an electron-positron pair in the geomagnetic field and the pair radiates a cascade of bremsstrahlung photons, leave a distinguishable imprint in the cameras of CTA-La Palma. Using air-shower simulations combined with a detector-response simulation, the authors find that high-zenith (80°) observations pointed toward geomagnetic north, where the field is stronger, produce images dominated by muon Cherenkov light, and that the image-shape parameters (size, width, length, distance) separate preshower events from proton background. The multivariate classifier reaches $\sigma = 29.85$, $\varepsilon_{\mathrm{preshw}} = 0.926$, $\varepsilon_{\mathrm{CR}} = 0.038$ for a diffuse source and $\varepsilon_{\mathrm{preshw}} = 0.982$, $\varepsilon_{\mathrm{CR}} = 0.022$ for a point source. The effective aperture is $0.05\,\mathrm{km}^2$ (diffuse) and $3.40\,\mathrm{km}^2$ (point source); with the super-heavy dark matter, GZK, and current upper-limit photon fluxes, the expected number of events in 30 hours ranges from $10^{-8}$ to $10^{-4}$, and a flux of $8.73 \times 10^{3}\,\mathrm{km}^{-2}\,\mathrm{yr}^{-1}\,\mathrm{sr}^{-1}$ would be needed to obtain a single event.

Load-bearing premise

The results rest on one simulated geometry — a 40 EeV photon arriving at 80° zenith from the geomagnetic north — and on the assumption that the simulated Cherenkov image shapes match what the telescope array will actually record in that configuration.

Editorial extensions

If this is right

  • If preshower events are real and pass the boosted-decision-tree selection, gamma-ray telescopes would provide a new observational channel for GZK photons and super-heavy dark-matter decay products.
  • A 30-hour observation run with the specified pointing (80° zenith, toward geomagnetic north) would, under current flux models, expect far fewer than one event, so the practical outcome would be an improved upper limit rather than a detection.
  • The point-source scenario has an effective area roughly 70 times larger than the diffuse scenario, making targeted searches of candidate sources a more promising route.
  • Choosing a different cut on the classifier score can eliminate cosmic-ray background entirely at the cost of signal efficiency, enabling background-free searches if the photon flux is sufficiently high.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An extension not explored in the paper would scan photon energies and arrival directions other than 40 EeV and geomagnetic north; conversion probability and Cherenkov image morphology both vary strongly with geometry, so the claimed separation could map an acceptance-weighted region of the sky.
  • Because the reported significance comes from Monte Carlo samples with balanced signal and background counts, real-data application would need re-optimisation against a vastly more numerous cosmic-ray background; the quoted efficiencies and significance are best-case values.
  • If the muon-isolation mechanism is correct, the same high-zenith strategy may transfer to other imaging atmospheric Cherenkov arrays and to lower photon energies, where the preshower conversion probability is smaller but the background environment also differs.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The manuscript presents a simulation-based feasibility study of detecting ultra-high-energy (UHE) photons through the preshower effect with the future CTA-La Palma array. It uses the PRESHOWER code to generate electron-positron cascades from a 40 EeV photon converting in the geomagnetic field, CORSIKA for air-shower development, and sim_telarray for the CTA camera response, with a diffuse source simulated by a 5-degree viewing cone and a nearly-horizontal arrival direction (theta = 80 deg, phi = 180 deg). A TMVA boosted-decision-tree classifier is trained on four Hillas parameters to separate preshower-induced events from a proton background simulated in the 10^4 to 3x10^6 GeV range. The paper reports a maximum classification significance of sigma = 29.85 (epsilon_presh = 0.926, epsilon_CR = 0.038), an effective aperture of 0.05 km^2 for the diffuse source, and expected event numbers below 1.5x10^-6 in 30 hours for SHDM, Auger-limit, and GZK flux models. The authors conclude that preshower events may be distinguished from the cosmic-ray background but that very few events are expected.

Significance. If the discrimination claim held, this would be a useful first step for a novel UHE-photon search channel: the simulation chain is built from established tools (PRESHOWER, CORSIKA, sim_telarray, TMVA), the observation geometry is concrete, and the paper is honest about the very low predicted event rates and about the large flux required for a single event. The central limitation is that the headline separation does not currently support the physical claim of preshower/hadron discrimination, because the signal and background samples differ by about 1.3x10^4 in primary energy and because the quoted significance depends on an arbitrary simulated sample size. With energy-matched samples and a properly defined detection significance, the classification result could be made meaningful; as it stands, the paper demonstrates classifier separation on synthetic samples rather than observability.

major comments (3)
  1. [Section 2.1 and Section 3.1] The preshower primary is fixed at E = 40 EeV, while the proton background is restricted to 10^4 to 3x10^6 GeV, a factor of about 1.3x10^4 lower in energy. Because the Hillas size and related image parameters scale steeply with primary energy, the separation shown in Figs. 2 and 3 and the quoted sigma = 29.85 are largely an energy selection effect and do not demonstrate that preshower-induced air showers can be distinguished from proton-initiated showers of the same energy. The 3 PeV cap is justified by the expected background count falling below one event in 30 hours, but that addresses the expected rate, not the classifier's physical discrimination capability. The authors should either simulate a same-energy hadronic background or clearly state that the classifier is intended for a background population with a different energy distribution and quantify the resulting selection bias.
  2. [Section 3.1, Eq. (3.1)] The definitions of S and B in Eq. (3.1) are circular or ambiguous: S is described as the preshower efficiency multiplied by the number of true positives, and B as the background efficiency multiplied by the number of false negatives. True positives already encode the efficiency, and false negatives are not the background events that pass the cut. More importantly, because S and B are proportional to the total number of simulated test events, sigma = 29.85 depends on an arbitrary simulation sample size and is not an expected detection significance for a 30-hour observation. The authors should define S and B as counts after the cut (or use a standard Poisson/Li-Ma significance), report the test-sample sizes, and, if a detection significance is intended, convert the classifier efficiencies into expected signal and background counts for the stated exposure.
  3. [Section 3.2 and Table 1] All aperture and event-rate numbers are derived for a single arrival geometry (E = 40 EeV, theta = 80 deg, phi = 180 deg, fixed cone and impact radius). The preshower conversion probability depends strongly on the geomagnetic field along the trajectory, so the quoted A_eff and N_preshw values do not generalize to an all-sky diffuse flux without an explicit integration over arrival directions and energies, or at least a demonstration that the dependence is weak. In addition, the expected numbers in Table 1 are far below one even for the most optimistic SHDM model (1.03x10^-4 for the point-source case and 1.49x10^-6 for the diffuse case in 30 hours). The conclusions should therefore state unambiguously that no event is expected in 30 hours under current flux models and that sigma = 29.85 is a classification significance on simulated samples, not an observability claim.
minor comments (4)
  1. [Section 3.2] The aperture is quoted in km^2, but Eq. (3.2) includes the solid angle Omega; for the diffuse case Omega = 2pi(1-cos 5 deg) is approximately 0.024 sr, so A_eff has units of km^2 sr. Please clarify whether the quoted 0.05 km^2 and 3.40 km^2 include this solid-angle factor and give Omega explicitly for the point-source value.
  2. [Section 3.1] The phrase 'false negatives' for CR background events classified as preshower is nonstandard; these events are usually called false positives or the background efficiency. Please correct the terminology to avoid confusion with the definitions in Eq. (3.1).
  3. [Introduction] The symbol d_int is used before its meaning is fully specified; please define it explicitly in the text where it first appears.
  4. [Section 2.1] The claim that the 3x10^6 GeV cap follows from an expected CR count below one event in 30 hours should be accompanied by a formula or reference for the aperture and spectrum used in that estimate, since it is not derivable from the information given.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the simulation chain and event-rate predictions do not reduce to their inputs by construction.

full rationale

The paper's derivation chain is: simulate preshowers with PRESHOWER/CORSIKA and CR background with CORSIKA, extract Hillas parameters, train a BDT on a training sample and evaluate it on a test sample, compute an aperture by Eq. (3.2), and convert external UHE-photon flux models into event counts via Eq. (3.3). None of these steps is equivalent to its own input by construction. The aperture is a definitional ratio of triggered to simulated events, and the event-rate prediction is a multiplication of an external model flux by a fixed aperture, conversion probability, and observation time; no fitted parameter is later renamed as a prediction. The main self-citations, the PRESHOWER code [4] and the geomagnetic-direction discussion [10], are tool/geometry references from co-authors, not load-bearing uniqueness claims or ansatz smuggled in through citation. Two caveats are correctness concerns rather than circularity: (i) the preshower sample is fixed at 40 EeV while protons are capped at 3 PeV, and Sec. 3.1 itself attributes the size separation to 'the larger primary energy', so the BDT separation may be partly an energy-selection artifact rather than particle-type discrimination; (ii) the quoted sigma = 29.85 from Eq. (3.1) is a cut-optimized in-sample classifier metric whose S and B are scaled by simulated event counts, so it is not an expected detection significance. The conclusion also acknowledges the extremely low model fluxes (Table 1) and states that the required flux would be much larger than optimistic SHDM predictions. These limitations affect the strength of the astrophysical claim, but they do not make the derivation circular.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical entities. Its central claim rests on a chain of simulation codes and several externally chosen parameters (energy, direction, cone angle, impact radius, observation time, cut value). The flux inputs are from previous literature. The most significant assumptions are the validity of the simulation codes and the representativeness of the single simulated geometry.

free parameters (7)
  • Primary photon energy = 40 EeV
    Chosen as a compromise between e+/e- conversion rate and computational time; the central simulation results depend on this single energy.
  • Zenith angle = 80 degrees
    Chosen to increase conversion probability and isolate the muon component in Cherenkov images; all runs use this value.
  • Azimuth angle = 180 degrees (CORSIKA frame)
    Direction of the stronger geomagnetic field at La Palma; the photon conversion rate and all subsequent results depend on this orientation.
  • Viewing cone angle = 5 degrees (diffuse), 0 degrees (point source)
    Defines the simulated angular spread of the source; directly enters the solid angle in the aperture formula.
  • Impact radius = 1300 m
    Maximum impact distance for shower core randomization; sets the geometric area term in the aperture formula.
  • BDT cut value = -0.01
    Chosen by maximizing the significance computed from the training sample; determines the reported efficiencies and contamination.
  • Observation time = 30 hours
    Assumed exposure for the event rate calculation; affects the expected number of preshowers linearly.
assumptions (5)
  • domain assumption The PRESHOWER algorithm correctly models photon conversion and bremsstrahlung in the geomagnetic field.
    The paper uses PRESHOWER [4] to generate the preshower particle distributions; if the physical modeling is wrong, the classification study is invalid.
  • domain assumption CORSIKA with QGSJETII-03 and URQMD accurately describes air shower development in the 40 EeV regime.
    These interaction models are applied to energies beyond their most tested range; the resulting shower images may not be accurate.
  • domain assumption sim_telarray with production-I settings faithfully represents the CTA-La Palma camera response.
    The trigger and electronics parameters are left at defaults; the aperture and efficiencies rely on this detector model.
  • domain assumption The adopted UHE photon production models (SHDM, GZK) and Pierre Auger limits provide the input fluxes for event rate predictions.
    The expected number of preshowers is linearly proportional to these external fluxes taken from the literature.
  • ad hoc to paper The four Hillas parameters (size, width, length, distance) are sufficient to separate preshowers from cosmic-ray background.
    No systematic study of additional parameters or the effect of the viewing cone on these parameters is presented.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Search for ultra-high energy photons: observing the preshower effect with gamma-ray telescopes." pith.science (2026). https://pith.science/paper/UB4HJ3U3

@misc{pith2026190808805,
  author       = {Pith},
  title        = {Pith review of: Search for ultra-high energy photons: observing the preshower effect with gamma-ray telescopes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UB4HJ3U3}},
  note         = {Machine review of arXiv:1908.08805}
}
read the original abstract

Ultra-high energy photons constitute one of the most important pieces of the astroparticle physics problems. Their observation may provide new insight on several phenomena such as supermassive particle annihilation or the GZK effect. Because of the absence of any significant photon identification by a leading experiments such as the Pierre Auger Observatory, we consider a screening phenomenon called preshower effect which could efficiently affect ultra-high energy photon propagation. This effect is a consequence of photon interactions with the geomagnetic field and results in large electromagnetic cascade of particles several thousands kilometers above the atmosphere. This collection of particles, called cosmic-ray ensembles (CRE), may reach the atmosphere and produce the well-known air showers. In this paper we propose to use gamma-ray telescopes to look for air showers induced by CRE. Possible sources of ultra-high energy photons include the GZK effect and Super Heavy Dark Matter particles. Simulations involving the preshower effect and detectors response are performed and properties of these peculiar air showers are investigated. The use of boosted decision trees to obtain the best cosmic-ray ensemble/hadron separation, the aperture and event rate predictions for a few models of photon production are also presented.

Figures

Figures reproduced from arXiv: 1908.08805 by the authors.

Figure 1
Figure 1. Number of secondary particles reaching the top of the atmosphere as a function of the [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Distribution of Hillas parameters for simulated preshower effect (red) and CR background [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Left panel: BDT score distributions of the testing and training samples of CR background and preshowers. Righ panel: Efficiencies, purity of preshower sample and significance of the classification as a function of the cut value applied to the distributions shown on the left side. smaller. Preshower efficiency and CR background contamination were found at εpreshw = 0.982 and εCR = 0.022, respectively. 3.2 Aperture an… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 13 canonical work pages

  1. [1]

    McBreen, C

    B. McBreen, C. J. Lambert, Phys. Rev. D24 (1981), 21-23

  2. [2]

    CTA Consortium, Experimental Astronomy 32 (2011), 193 [astro-ph/1008.3703] 6 Search for ultra-high energy photons through preshower effect Kevin Almeida Cheminant

  3. [3]

    https://credo.science/publications/

  4. [4]

    Homola, et al., Computer Physics Communications 184 (2005), 1468 [astro-ph/0311442]

    P. Homola, et al., Computer Physics Communications 184 (2005), 1468 [astro-ph/0311442]

  5. [5]

    National Geophysical Data Center [ www.ngdc.noaa.gov]

    U. National Geophysical Data Center [ www.ngdc.noaa.gov]

  6. [6]

    Heck, et al., FZKA Report 6019 (1998)

    D. Heck, et al., FZKA Report 6019 (1998)

  7. [7]

    Ostapchenko, Nuc

    S. Ostapchenko, Nuc. Phys. B (Proc. Suppl.) 151 (2006), 143 [hep-ph/0412332]

  8. [8]

    S. A. Bass, Prog. in Part. and Nucl. Phys. 41 (1998), 255 [nucl-th/9803035]

Show all 19 references
  1. [9]

    Bernlöhr, Astropart

    K. Bernlöhr, Astropart. Phys. 30 (2008), 149 [astro-ph/0808.2253]

  2. [10]

    Almeida Cheminant for the CREDO Collaboration, Proceedings of 35th ICRC (2017) [astro-ph/1709.05180]

    K. Almeida Cheminant for the CREDO Collaboration, Proceedings of 35th ICRC (2017) [astro-ph/1709.05180]

  3. [11]

    Neronov et al., Phys

    A. Neronov et al., Phys. Rev D94 (2016) 123018, [astro-ph/1610.01794]

  4. [12]

    A. M. Hillas, Proc. of 19th ICRC 3 (1985), 445

  5. [13]

    Hoecker, P

    A. Hoecker, P. Speckmayer, J. Stelzer, J. Therhaag, E. von Toerne, H. V oss,PoS ACAT(2007), 40 [physics/0703039]

  6. [14]

    Ohm, and others, Astropart

    S. Ohm, and others, Astropart. Phys. 31 (2009), 383

  7. [15]

    Albert, and others, Nucl

    J. Albert, and others, Nucl. Instrum. Meth. 588 (2008), 424

  8. [16]

    Ellis, and others, Phys

    J. Ellis, and others, Phys. Rev. D74 (2006), 115003

  9. [17]

    Gelmini, O

    G. Gelmini, O. Kalashev, D. Semikoz, J. Exp. Theor . Phys.106 (2008), 1061

  10. [18]

    Bleve for the Pierre Auger Collaboration, PoS(ICRC2015) 1103 (2015) [arXiv 1509.03732 ]

    C. Bleve for the Pierre Auger Collaboration, PoS(ICRC2015) 1103 (2015) [arXiv 1509.03732 ]

  11. [19]

    Pierre Auger Collaboration, JCAP 009 (2017) 7

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.