{"id":"96234d5b-9ac6-4fe5-a49f-5fd79dba9ffb","arxiv_id":"2505.10459","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"SMIET synthesizes air-shower radio pulses in seconds from one simulated shower, matching CoREAS amplitudes within about 4-6% when the shower maximum is within 100 g/cm2.","lead":"This paper introduces SMIET, a software method that synthesizes radio signals of cosmic-ray air showers in seconds by rescaling a single detailed simulation. It could replace slow full Monte-Carlo simulations for many analyses, enabling faster reconstructions and machine-learning fits.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Headline 4%/6% scatter is measured in-sample; the transferability of spectral functions to 'any atmosphere, observation level, magnetic field' rests on only three 20-shower scenarios, so the accuracy claim is not yet independently established.","rationale":"The reader's verdict is CONDITIONAL, and my analysis supports that: the method is plausible, the code is public, and the independent tests in Section 5 are encouraging, but the headline accuracy claim rests on an in-sample benchmark and a limited set of transferability tests. The most load-bearing concern is therefore that the 4%/6% figure has not been demonstrated for genuinely different atmospheres, magnetic fields, or observation levels, despite the abstract's 'any' wording. The Appendix A log/exp discrepancy is a real typographical error but does not affect the core argument, since the main-text formula is the correct inversion of the exponential atmosphere profile. I agree with the reader's identification of the weakest assumption, and I would keep the verdict CONDITIONAL rather than moving it, because the proposed out-of-sample test could plausibly confirm the method's generality; nothing in the paper yet warrants rejection. I would not call it UNVERDICTED because the available evidence, while incomplete, is substantive and reproducible.","tokens_in":29763,"tokens_out":8947,"duration_ms":92854,"concrete_test":"Generate 50 CoREAS showers with a different atmospheric model (e.g., CORSIKA model 10, tropical) and a different magnetic field, at zenith angles between 20 and 50 degrees and energies between 1e17 and 1e19 eV. Use the published SMIET package with its provided spectral functions to synthesise each shower, and compute the peak-ratio metric Speak (Eq. 7) against the corresponding CoREAS traces over [30, 500] MHz for |ΔXmax| ≤ 100 g/cm2. If the GEO scatter exceeds 4% or the CE scatter exceeds 6%, the transferability claim fails. Also apply the interpolated synthesis to the same set and report the residual bias and scatter, to quantify whether the bias-correction procedure remains accurate outside the original library.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that SMIET reproduces CoREAS-level radio pulses to within 4% (GEO) and 6% (CE) scatter when |ΔXmax| ≤ 100 g/cm2. The primary benchmark in Section 4 uses the same ~800-shower library from which the spectral functions were extracted. Because the spectral functions are fits to the mean of that library, this benchmark cannot detect a systematic offset common to all library showers, such as one induced by the US-standard atmosphere, the LOFAR magnetic field, or the observation level. The abstract's claim that the package 'can be used with any atmosphere, observation level and magnetic field' is supported only by Section 5, which tests (i) the AERA site with a different magnetic field and observation level but the same atmosphere, (ii) a 10%-denser artificial atmosphere with the same geometry and magnetic field, and (iii) geometry changes up to ~3 degrees. Each scenario uses only 20 showers and the same spectral functions. If the scaling relations of Section 2.4 (e.g., the 1/ρ scaling for the geomagnetic component, which the cited literature [18] already shows breaking down below ~600 g/m3) do not hold for a genuinely different atmosphere, the claimed scatter would not transfer. Additionally, the unexplained bias of up to 5% is corrected via interpolated synthesis, but this correction is demonstrated on a single antenna (Figure 8) and its residual error is not quantified statistically. These issues do not refute the method; the independent tests are encouraging, and the code is publicly available. But the headline accuracy is not yet established outside the specific conditions used to fit the spectral functions.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents SMIET, a template-based method for fast synthesis of radio pulses from extensive air showers. Starting from a single sliced CoREAS simulation (the origin shower), the method rescales the emission from each atmospheric slice using semi-analytical spectral functions that depend on the shower age and the normalised viewing angle, and applies explicit arrival-time corrections to allow small changes in the shower geometry. The authors extract the spectral functions from a library of almost 800 showers covering zenith angles of 20–50 degrees, then benchmark the method by synthesising every shower onto every other shower in the same library. The headline result is that, for |ΔXmax| ≤ 100 g/cm², the scatter on the peak amplitudes is at most 4% for the geomagnetic component and smaller than 6% for the charge-excess component. A position-dependent bias of up to 5% is reported and a weighted-average ('interpolated synthesis') correction is proposed. The paper also describes the open-source SMIET package with NumPy and JAX implementations, and presents three smaller validation scenarios: a different site (AERA), a 10%-denser atmosphere, and geometry changes of up to a few degrees.","tokens_in":30052,"tokens_out":6869,"duration_ms":64806,"significance":"If the claimed accuracy holds for genuinely different sites, atmospheres, and observation levels, SMIET would be a valuable tool for the radio-detection community, enabling large simulation sets for reconstruction, machine-learning analyses, and information-field-theory applications. The paper is transparent about its approach, releases the code publicly, and includes machine-checked benchmarks and a fully differentiable JAX implementation, which are notable strengths. The method is novel in generalising template synthesis beyond a fixed geometry, and the observation that the spectral functions are universal in ΔXmax is a useful physics result. However, the central accuracy claim rests primarily on an in-sample benchmark, and the independent validation is limited to a small number of scenarios, so the quantitative headline should be treated with caution until out-of-sample evidence is provided.","major_comments":[{"comment":"The headline scatter values (4% for GEO, 6% for CE at |ΔXmax| ≤ 100 g/cm²) are computed by synthesising every shower in the same simulation library from which the spectral functions were extracted in Section 3.1. This makes the main benchmark partly in-sample: a systematic offset common to all library showers (for example, induced by the US-standard atmosphere, the LOFAR magnetic field, or the sea-level observation level) would be invisible to this test. The independent checks in Section 5 use only 20 showers per scenario and still apply the same spectral functions, so they constrain the transferability only weakly. I ask the authors to add an out-of-sample validation, for example by withholding a subset of showers from the spectral-function fit and benchmarking on that held-out set, or to explicitly state in the abstract and conclusions that the 4%/6% scatter is demonstrated only for the training library and that site-to-site transfer is supported only by the limited Section 5 tests.","section":"Section 4.1 vs Section 3.1"},{"comment":"The conclusion that zenith-angle changes up to about 3 degrees keep the peak ratio within one standard deviation is based on manual inspection, as the authors state, without a statistical analysis over the full antenna set. The outliers that motivated plotting the median instead of the mean are not quantified, and the antenna-distance dependence visible in Figure 12 is not characterised. Provide a quantitative criterion, for example the fraction of synthesis cases within a given tolerance as a function of Δθ, so that the 'up to 3 degrees' claim is reproducible. As written, the arrival-direction reconstruction use case is not firmly established.","section":"Section 5.3 (Figures 11 and 12)"},{"comment":"The interpolated-synthesis bias correction is demonstrated on a single antenna, and the residual bias and scatter after correction are not quantified over the benchmark set. Because the paper recommends interpolated synthesis for practical applications, please report the mean and standard deviation of the corrected peak ratios as a function of ΔXmax and antenna distance, or otherwise provide statistical evidence that the correction removes the bias without inflating the scatter.","section":"Section 4.2 (Figure 8)"}],"minor_comments":[{"comment":"The displayed formula for the emission time contains a typographical error: the expression after 'd·cos(θ) =' should involve a natural logarithm, not an exponential, and the argument of the logarithm should be (Xslice·cosθ − a_i)/b_i. Please correct the equation in the main text and the corresponding derivation in Appendix A.","section":"Appendix A and Section 2.7"},{"comment":"The limitation of the 1/ρ scaling at low densities is discussed in Section 6.4, but Section 2.4 states the scaling without this caveat. Adding a sentence noting that the scaling is verified in the literature only down to about 600 g/m³ would prevent readers from applying the method to very inclined showers or high-altitude sites beyond its validated range.","section":"Section 2.4"},{"comment":"The 20-degree zenith simulation set has a geomagnetic angle of only 2.3 degrees, and the GEO component is excluded from the main benchmark for this reason. It would be helpful to state in the conclusions that the quoted GEO accuracy assumes a sufficiently large geomagnetic angle, not just zenith angle.","section":"Section 3.1 and Section 4.1"},{"comment":"The universality tests use CORSIKA v7.7550, while the spectral functions were extracted with v7.6400. Please state explicitly whether the simulation-code update affects the radio-emission calculation, so that readers can rule out a version dependence as a confounding factor.","section":"Section 5"},{"comment":"The text says the peak ratio drops quickly with increasing zenith-angle difference, but the quantitative relation (for example, a linear fit of the score versus Δθ) is not given. Adding such a fit would make the 'up to 3 degrees' statement more precise and easier to test.","section":"Section 5.3 and Figure 12"}],"recommendation":"major_revision","confidential_remarks":"The paper is well-written and the code release is a positive feature. The main weakness is that the headline accuracy is measured on the same library used to fit the spectral functions, so the 4%/6% numbers are partly self-referential. The authors are transparent about this in the text, but the abstract overstates the universality ('any atmosphere, observation level and magnetic field'). I would ask for either a proper out-of-sample split or a clearly qualified abstract before publication. The other two major points (quantification of the geometry-variation performance and of the interpolated-synthesis correction) are practical but fixable. If the authors provide the additional validation or temper the claims, the paper would be suitable for Astroparticle Physics."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline is right but the asterisks matter. SMIET does what it claims—synthesizes CoREAS-level radio pulses in seconds for zenith angles up to 50°—but the headline 4%/6% scatter is measured on the same library used to fit the spectral functions, so it demonstrates self-consistency more than independent accuracy. The authors are upfront about this and add three independent 20-shower tests (AERA site, 10% denser atmosphere, geometry changes up to ~3°) that support transferability without nailing it down.\n\nWhat's genuinely new: the generalization from the fixed vertical geometry of [14] to arbitrary arrival directions, via viewing-angle normalization, universal spectral functions, and explicit phase-delay corrections. The phase treatment is a real step forward and is what makes the up-to-3° zenith variation possible. The open-source SMIET package, with NumPy and differentiable JAX implementations, is practical and will likely get used. The paper is well written, the motivation is clear, and the authors are candid about the residual bias and about the fact that the main benchmark is in-sample.\n\nSoft spots, in proportion. The in-sample benchmark cannot catch a systematic offset common to all library showers—say, one tied to the US-standard atmosphere, LOFAR field, or sea-level altitude. The claim that the package works with 'any atmosphere, observation level and magnetic field' rests on only three 20-shower scenarios: the AERA test changes field and altitude but not atmosphere; the dense-atmosphere test keeps geometry and field fixed; the geometry test is limited in zenith range and shows outliers beyond ~3°. So the universal transferability is plausible but not established. The interpolated bias correction is demonstrated on a single antenna and its residual error is not statistically quantified. Minor: Appendix A uses an exponential where the main text uses a logarithm in the depth inversion—worth a fix.\n\nThe central argument holds up. The scaling relations are physically motivated, the paper cites the relevant literature and even notes where the density scaling breaks down. For a reader who wants a fast forward model for LOFAR/Auger/SKA-style analyses, this is likely a useful tool now, with the caveat that they should validate within their own parameter range.\n\nRecommendation: send it to peer review. A good referee should ask for either a larger independent validation set or a softened 'any atmosphere' claim, and some statistical quantification of the bias-correction residual. But this is a serious, honest methodological contribution.","headline":"Genuinely useful generalization of template synthesis, honestly presented, but headline accuracy is in-sample and the 'any atmosphere' claim outruns the independent tests.","tokens_in":30661,"tokens_out":4681,"would_cite":true,"duration_ms":39569,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"One simulated shower can produce any target air shower's radio pulse in seconds, matching full simulation within 4%.","keywords":["template synthesis","extensive air showers","radio emission","CoREAS","forward model","X_max reconstruction","differentiable simulation"],"falsifier":"Use the precomputed spectral functions to synthesise the pulse of a shower at a third site, with a magnetic field strength and observation level clearly different from the two already tested, and compare to a full simulation; if the peak-ratio scatter for |ΔX_max| ≤ 100 g/cm² exceeds 4% for the geomagnetic trace or 6% for the charge-excess trace, the claimed universality of the spectral functions is refuted.","tokens_in":29508,"feed_emoji":"📡","tokens_out":12600,"duration_ms":117520,"temperature":0.7,"pith_summary":"The paper aims to replace Monte-Carlo simulations of air-shower radio emission, which take hours to days per event, with a forward model that computes the radio pulse of any target shower in seconds. The method takes one microscopically simulated 'origin' shower, slices it by atmospheric depth, and rescales the emission of each slice using semi-analytic spectral functions that isolate how the pulse shape depends on shower age and viewing angle. The central quantitative claim is that when the origin and target shower maxima are within 100 g/cm², peak amplitudes of synthesised geomagnetic traces scatter by at most 4% relative to CoREAS, and charge-excess traces by less than 6%, close to the intrinsic shower-to-shower fluctuation level. A symmetric bias of up to 5% is removed by interpolated synthesis, averaging two origins that bracket the target. If this holds, high-statistics event reconstruction and machine-learning or Bayesian-field analyses of radio data become computationally feasible.","feed_headline":"Air-shower radio pulses synthesised in seconds at 4% accuracy","feed_subtitle":"Slice-by-slice rescaling turns one simulated shower into any target geometry, cutting runtimes from days to seconds.","key_machinery":"The load-bearing object is the 'spectral function', a semi-analytic parametrisation of a slice's amplitude frequency spectrum after removing geometrical scalings. The spectral parameters a, b, and c are fitted as parabolas in ΔX_max, one set per normalised viewing angle, where the viewing angle is the opening angle between the shower axis and the antenna divided by the local Cherenkov angle. This machinery, together with explicit phase correction based on geometric arrival times, lets the method rescale the origin shower to any target geometry in seconds while retaining the full phase information from the microscopic simulation.","core_discovery":"The central claim is that the radio emission of an extensive air shower is a universal function of the longitudinal profile: once the emission of each atmospheric slice is normalised by distance, particle count, geomagnetic-angle sine, air density for the geomagnetic component, and local Cherenkov angle for the charge-excess component, the remaining amplitude spectrum depends only on the slice's distance to the shower maximum and on the viewing angle expressed as a fraction of the local Cherenkov angle. These dependencies are encoded in fitted 'spectral functions', parabolas in ΔX_max, one set per viewing angle, extracted from a library of about 800 CORSIKA/CoREAS showers. To synthesise a new shower, the origin template's slice spectra are rescaled by the inverse factors evaluated at target parameters, and the phases are shifted by the geometric arrival-time difference between origin and target geometry. The benchmark over the [30, 500] MHz band gives a peak-amplitude scatter of at most 4% for the geomagnetic component and 6% for the charge-excess component when |ΔX_max| ≤ 100 g/cm², with a bias symmetric about ΔX_max = 0 that is correctable by interpolating between two origin showers. The method is verified for zenith angles up to 50°, at a second observation level with a different magnetic field, under a 10% denser atmosphere, and for modest geometry changes of a few degrees during synthesis.","pith_inferences":["The paper does not push the density rescaling to its stated validity limit; a test with an atmosphere well below 600 g/m³ would show whether per-slice cancellation absorbs the breakdown of the inverse-density scaling.","Because the observed bias is symmetric in ΔX_max, its origin is likely an even function of the profile difference, for instance slice-to-slice phase coherence; isolating that quantity could replace interpolation with a single-origin analytic correction.","If the residual phase dispersion beyond the linear arrival-time term were parametrised, the demonstrated few-degree zenith-angle range during synthesis might extend substantially, enabling direct direction fits without resimulating origin showers.","A consolidated library of a few origin showers and one set of p-parameters per frequency band could serve all radio experiments, removing the need for each detector to build its own Monte-Carlo library; the paper provides the parts but leaves this consolidation unstated."],"forward_implications":["With a template bank of origins spaced roughly 100 g/cm² in X_max and the interpolated synthesis correction, pulses can be synthesised for any target X_max with scatter at or below the benchmarked 4–6% level.","Event-level analyses can afford thousands of synthesised showers per recorded event, making chi-squared or likelihood-based reconstruction of X_max, energy, and geometry routine.","Because the JAX version is fully differentiable, gradient-based and Bayesian information-field-theory inversions can reconstruct the full longitudinal profile, not just X_max.","The benchmarked universality across primary types, energies, atmospheres, observation levels, and magnetic fields means one set of spectral functions can be reused at many sites, provided the site parameters are entered into the scaling relations.","The method applies up to about 50° zenith angle; for more inclined showers it is designed to hand off to Radio Morphing, together covering the full range of arrival directions."],"supporting_citations":[{"why":"Supplies the CoREAS Monte-Carlo code that produces the reference traces for every benchmark in the paper.","marker":"[7, 8]"},{"why":"The earlier proof-of-concept of template synthesis for vertical showers, which this paper generalises in geometry and phase treatment.","marker":"[14]"},{"why":"Radio Morphing, the alternative rescaling method for inclined showers, whose Cherenkov-angle-based framing motivates the normalised viewing angle.","marker":"[16]"},{"why":"Provides the sine-of-geomagnetic-angle scaling used to normalise the geomagnetic emission per slice.","marker":"[17]"},{"why":"Supplies the air-density and Cherenkov-angle amplitude scalings for the geomagnetic and charge-excess components that the synthesis applies per slice.","marker":"[18]"},{"why":"The Fourier interpolation method used to place antennas at fixed normalised viewing angles when extracting the spectral functions.","marker":"[19]"},{"why":"Gives the atmospheric profile parametrisation used to compute the emission and arrival times of each slice.","marker":"[20]"},{"why":"The JAX framework that makes the synthesis pipeline fully differentiable in the provided software.","marker":"[25]"}],"fun_headline_variants":["Radio pulses from air showers in seconds, not days","4% accurate air-shower radio synthesis in seconds","Template synthesis: air-shower radio from one shower","Slice-by-slice rescaling speeds air-shower radio","From one shower to any: radio synthesis in seconds"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that a slice's radio emission, after simple rescalings, depends only on the slice's distance to the shower maximum and on the viewing angle, and not on what kind of particle started the shower, how energetic it was, or which atmosphere, magnetic field, or site geometry it is in.","fun_headline_variants_meta":{"raw":{"variants":["Radio pulses from air showers in seconds, not days","4% accurate air-shower radio synthesis in seconds","Template synthesis: air-shower radio from one shower","Slice-by-slice rescaling speeds air-shower radio","From one shower to any: radio synthesis in seconds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000246,"raw_usage":{"total_tokens":1664,"prompt_tokens":1195,"completion_tokens":469,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":811,"completion_tokens_details":{"reasoning_tokens":392}},"tokens_in":811,"tokens_out":469,"duration_ms":4455,"temperature":1.0,"reasoning_tokens":392,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:08:38.723486+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Use the precomputed spectral functions to synthesise the pulse of a shower at a third site, with a magnetic field strength and observation level clearly different from the two already tested, and compare to a full simulation; if the peak-ratio scatter for |ΔX_max| ≤ 100 g/cm² exceeds 4% for the geomagnetic trace or 6% for the charge-excess trace, the claimed universality of the spectral functions is refuted.","supporting_citations":[{"cited_title":"Proof of principle for template synthesis approach for the radio emission from vertical extensive air showers","cited_arxiv_id":"2307.02939","evidence_quote":"The earlier proof-of-concept of template synthesis for vertical showers, which this paper generalises in geometry and phase treatment."},{"cited_title":"Radio Morphing: towards a fast computation of the radio signal from air showers","cited_arxiv_id":"1811.01750","evidence_quote":"Radio Morphing, the alternative rescaling method for inclined showers, whose Cherenkov-angle-based framing motivates the normalised viewing angle."},{"cited_title":"Ammerman-Yebra, J","cited_arxiv_id":null,"evidence_quote":"Supplies the air-density and Cherenkov-angle amplitude scalings for the geomagnetic and charge-excess components that the synthesis applies per slice."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the atmospheric profile parametrisation used to compute the emission and arrival times of each slice."},{"cited_title":"URLhttp://github.com/google/jax","cited_arxiv_id":null,"evidence_quote":"The JAX framework that makes the synthesis pipeline fully differentiable in the provided software."}],"review_version":1}