{"id":"553ba9ca-89e7-4e05-9c5d-87f7ee97c090","arxiv_id":"2501.06257","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In IceCube Upgrade simulations, in-ice photon scattering and hadronic shower spread, not module resolution or noise, set the practical floor on cascade zenith-angle reconstruction resolution.","lead":"Using idealized simulations of the IceCube Upgrade, this paper finds that in-ice photon scattering and the transverse spread of hadronic shower particles are the dominant limits on low-energy cascade zenith angle reconstruction, while module resolution and noise matter little. This gives detector physicists a quantitative target for where reconstruction improvements will pay off in neutrino oscillation analyses.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The unquantified claim that IceCube systematics dominate the neutrino–shower kinematic opening angle is load-bearing; at 10 GeV the paper's own α=12°, β=0.3 fit gives ~6° mean offset, comparable to the Fig. 10 median cosθ resolutions.","rationale":"The reader's weakest_assumption correctly identifies the unsupported exclusion of the neutrino–shower kinematic opening angle. The paper flags this exclusion in Section 3, then asserts in Section 5.1 without quantitative support that IceCube systematics dominate it. Because the paper's motivation is neutrino zenith reconstruction for oscillation analyses, this omission directly affects the interpretation of every quoted resolution. The paper's own parameterization, α = 12° and β = 0.3, implies a mean offset of roughly 6° at 10 GeV, which near the horizon is comparable to the median resolutions shown in Fig. 10. This is therefore not a stylistic or presentational issue but a load-bearing assumption that could change the meaning of the central result. Other potential concerns, such as the loose wording of the 20% per-module-PDF improvement in Section 5.2 or the lack of release of code and data, are real but secondary: the ranking of scattering and shower spread over module resolution and noise is internally consistent within the simulation, and the per-module-PDF claim does not threaten the central ranking. The reader's CONDITIONAL verdict already captures the appropriate level of caution, so no verdict adjustment is needed.","tokens_in":16789,"tokens_out":5125,"duration_ms":54873,"concrete_test":"Add an opening-angle step to the existing simulation: for each 1–20 GeV event, sample a shower-axis direction from a GENIE-based opening-angle distribution around the true neutrino direction (or, as a minimal check, from the paper's own Ψ ≈ 12° (E/GeV)^−0.3 fit), then propagate photons and reconstruct exactly as in Sections 3–5. Report the 68% containment of |cosθ_ν^true − cosθ_shower^reco| as a function of photon hits and compare it with Fig. 10. If the added opening-angle contribution is ≤0.02 in cosθ or ≤20% of the quoted medians across the 5–80 hit range, the Section 5.1 dominance claim is supported; if it is comparable to the quoted values, the quoted resolutions are cascade-axis limits, not neutrino-zenith resolutions, and the conclusions require revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the assertion in Section 5.1: 'These systematics in IceCube dominate over the kinematic opening angle (Section 3; first paragraph).' Section 3 explicitly excludes the neutrino–shower opening angle from the simulation, fitting Ψ ≈ 12° (E/GeV)^−0.3. At 10 GeV this gives a mean opening angle of about 6°; for near-horizontal events (sin θ ≈ 1) that corresponds to Δcos θ ≈ 0.1, comparable to the median resolutions reported in Fig. 10 at moderate photon counts. The paper's stated physics motivation, in the abstract and Section 1.3, is reconstructing the neutrino zenith angle for oscillation analyses, where L ≈ 2R|cos θ|. If the unsupported dominance claim fails, the quoted resolutions bound the cascade-axis reconstruction, not the neutrino-direction reconstruction needed for oscillation and mass-ordering studies, so the practical conclusions do not follow from the simulation as presented. The ranking of in-ice scattering versus shower spread may survive this issue, but the claimed 'intrinsic zenith resolution limits' for neutrino events are not established without quantifying the opening-angle contribution.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses the standalone Photon Propagation Code (PPC) to simulate low-energy (1–20 GeV) hadronic cascades in an idealized IceCube Upgrade geometry with homogeneous ice and only mDOMs, then reconstructs the shower-axis zenith angle with an extended unbinned likelihood based on photon direction, timing, and per-module charge. The central method is a controlled ablation: individual information-loss processes (in-ice photon scattering, transverse hadronic shower spread, module directional resolution, and module noise) are toggled on or off one at a time, always with matching PDFs generated from the same simulation chain. The authors find that in-ice photon scattering is the dominant resolution limiter, followed by transverse shower spread, while module resolution and noise have negligible impact; they also report that per-module photon PDFs would improve resolution by about 20% relative to the vertex-averaged integrated PDFs used in the main reconstruction. The paper frames these results as approaching the intrinsic zenith-angle resolution limits for the IceCube Upgrade and as guidance for the development of low-energy reconstruction algorithms.","tokens_in":17092,"tokens_out":3563,"duration_ms":34921,"significance":"If the conclusions hold, the paper provides a useful, quantitative ranking of information-loss processes for GeV-scale cascade reconstruction in IceCube Upgrade, with direct implications for where algorithmic and hardware effort should be invested. The ablation design is a genuine strength: it isolates each process with matching simulation and reconstruction PDFs, and the use of the open-source PPC chain makes the study reproducible in principle. The true-versus-averaged PDF comparison in Section 4.3 is a commendable attempt to quantify modeling error, even if it covers only two extreme geometries. Two caveats temper the significance: the kinematic opening angle between the incoming neutrino and the hadronic shower axis is excluded by construction and then asserted to be subdominant without a quantitative comparison, and the absolute resolution values carry an unquantified modeling-error component that is comparable in size to the reported median resolutions in the two test cases. As a result, the paper's practical conclusions about neutrino zenith-angle reconstruction are defensible in direction but not yet quantitatively established as stated.","major_comments":[{"comment":"The assertion that \"These systematics in IceCube dominate over the kinematic opening angle\" is load-bearing for the paper's stated goal of reconstructing neutrino zenith angles, but it is not quantified anywhere. Section 3 deliberately simulates only the hadronic shower axis and excludes the neutrino–shower opening angle, whose fitted parameterization Ψ ≈ 12° (E/GeV)^−0.3 implies a mean offset of about 6° at 10 GeV. For near-horizontal events this corresponds to Δcos θ ≈ 0.1, which is comparable to the median resolutions in Fig. 10 at moderate photon counts. Without a quantitative comparison (or a simulation that includes the opening angle), the quoted resolutions bound cascade-axis reconstruction, not neutrino-direction reconstruction, and the abstract's claim about zenith-angle resolution limits for neutrino events is not established.","section":"Section 5.1, paragraph after Fig. 10; Section 3, first paragraph"},{"comment":"The check of vertex-averaged versus true integrated PDFs is carried out for only two extreme geometries. The reported RMS spreads of 0.06 and 0.09 are comparable to the median resolutions of about 0.1 and 0.13 in Fig. 10, so the conclusion that \"the vertex-averaged PDFs do not introduce an additional bias on average\" is not sufficient to show that the modeling error is negligible for the absolute resolutions. The paper should either quantify the modeling error over a distribution of vertex positions and directions or explicitly restate the resolutions as upper bounds that include this modeling uncertainty; as written, the absolute resolution values in Figs. 8–11 have an unquantified systematic component.","section":"Section 4.3, Figs. 5 and 6"},{"comment":"The toy simulation replaces per-photon observables with samples from the averaged integrated PDF, which the paper itself describes as unphysical because scattering becomes distance-independent. The resulting ~20% improvement is therefore not a direct estimate of the gain from per-module PDFs; it is an estimate of the combined effect of removing both the modeling error of the averaged PDF and the per-module distance information. The caption and text should distinguish these two interpretations, and the brown line in Fig. 11 should be labeled as an idealized bound rather than \"a more accurate estimation of the achievable resolutions.\"","section":"Section 5.2, Fig. 11"}],"minor_comments":[{"comment":"The equation for the opening angle appears with broken notation (\"Ψ ˆ𝜈,®𝑢≈ 𝛼 ...\"); please typeset it properly and define Ψ explicitly as the mean angle between the incoming neutrino and the shower axis.","section":"Section 3, opening-angle formula"},{"comment":"The paper states that 2D histograms are averaged over 10000 events and 3D histograms over 50000 events, but it does not report whether the fitted KDE bandwidth and window-size parameters are stable across independent simulation samples; a brief cross-validation statement would strengthen confidence in the PDFs.","section":"Section 4.3, PDF averaging"},{"comment":"The caption reads \"with the same MC as in Fig.10 before photon generation and propagation,\" which is self-referential; it should refer to the MC setup of Figs. 8 and 9.","section":"Section 5.1, Fig. 10 caption"},{"comment":"The abstract restricts the conclusions to neutral-current events while Section 6 argues the results apply equally to νe and ντ charged-current cascades; harmonize the wording so the scope is stated consistently.","section":"Abstract and Section 6"},{"comment":"The assumption that all modules are 24-PMT mDOMs is acknowledged as optimistic; since resolutions are shown versus photon hit counts rather than energy, please add a sentence noting how excluding pDOMs and DEggs could shift the hit-count distribution in real Upgrade events.","section":"Section 2, instrumentation assumption"}],"recommendation":"major_revision","confidential_remarks":"This is a well-structured simulation study with a clear ablation design, and the central ranking of scattering versus shower spread is credible. The main blocker is the unsupported kinematic-opening-angle claim, which is load-bearing for the neutrino-zenith interpretation; it is fixable by adding a quantitative estimate (or a short simulation including the opening angle) and by softening the 'intrinsic limits' language where the modeling-error estimates in Figs. 5–6 are comparable to the quoted resolutions. The paper fits JINST well, and I see no circularity concern in the PDF generation, provided the modeling-error caveat is properly stated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the ablation study is solid and the qualitative ranking of resolution limiters is believable; the weak point is the one-sentence dismissal of the neutrino–shower kinematic opening angle, which matters more than the paper lets on.\n\nWhat's new: this transfers the KM3NeT/ORCA intrinsic-limit analysis to the IceCube Upgrade and to flavor-blind hadronic NC cascades. The controlled toggling of individual information-loss processes—with PDFs regenerated for each case—is a clean way to separate effects, and the true-vs-averaged PDF comparison gives an honest estimate of modeling error. The rough 20% gain from per-module PDFs is new and actionable. On the quantitative side, the conclusion that in-ice scattering and transverse shower spread dominate, while module resolution and noise are nearly irrelevant, is well supported by the ablation curves.\n\nSoft spots, in order of importance. First, the kinematic opening angle is excluded from the simulation (Section 3) and then asserted to be dominated by IceCube systematics (Section 5.1) without a number. Using the paper's own fit α=12°, β=0.3, the mean opening angle at 10 GeV is about 6°, which for near-horizontal events is Δcosθ≈0.1—comparable to the median resolutions in Fig. 10 at moderate photon counts. So the quoted resolutions bound the shower-axis direction, not necessarily the neutrino direction needed for oscillation analyses. That doesn't undermine the relative ranking, but it does mean the phrase 'intrinsic zenith resolution limits' is too strong unless the opening-angle contribution is added to the error budget. This is fixable with a short quantitative paragraph.\n\nSecond, there is no code or data release, and no comparison with existing IceCube reconstructions (e.g., RETRO), so the 'achievable resolutions' cannot be independently checked or placed against current algorithms. Minor but relevant. The other idealizations—homogeneous ice, no PMT electronics, confined events, uniform mDOMs—are acknowledged and appropriate for an optimistic-limit study.\n\nWho should read this: anyone working on low-energy cascade reconstruction in IceCube/DeepCore/Upgrade or on likelihood-based methods in ice/water Cherenkov detectors. It deserves a serious referee. My recommendation: send to peer review; require the opening-angle quantification and, ideally, release the reconstruction code and configuration so the numbers can be reproduced.","headline":"Solid ablation study with a believable ranking, but the unquantified dismissal of the neutrino–shower opening angle makes the 'intrinsic zenith limit' claim too strong as stated.","tokens_in":17553,"tokens_out":3005,"would_cite":true,"duration_ms":28132,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["95.55.Vj","29.40.Ka"],"model":"deepseek-v4-flash","headline":"The IceCube Upgrade's cascade zenith-angle resolution is capped by in-ice photon scattering and shower spread, not by the optical modules.","keywords":["IceCube Upgrade","cascade reconstruction","zenith angle resolution","in-ice photon scattering","hadronic shower spread","module resolution","neutrino oscillations","Cherenkov detectors"],"falsifier":"Simulate low-energy NC events with the full neutrino–shower opening angle included in the truth, reconstruct them with the same likelihood, and compare the residual between reconstructed and true neutrino zenith at the same photon-hit counts. If near-horizontal 10 GeV events, where the mean opening angle of about 6 degrees corresponds to $\\Delta\\cos\\theta \\sim 0.1$, show median residuals appreciably above the quoted resolution curves, then the paper's resolutions bound only the cascade axis, not the neutrino direction.","tokens_in":16544,"feed_emoji":"🧊","tokens_out":8620,"duration_ms":79862,"temperature":0.7,"pith_summary":"This paper asks what ultimately limits how precisely the IceCube Upgrade can reconstruct the zenith angle of low-energy hadronic-cascade events, a quantity that sets the neutrino oscillation baseline for the detector. By simulating the same events with one information-loss process switched off at a time, it argues that in-ice photon scattering is the dominant limiter, followed by the transverse spread of hadronic shower particles, while the finite directional resolution of the multi-PMT modules and module noise degrade the resolution only slightly. Removing scattering improves the median resolution by roughly 50 percent and removing shower spread by about 30 percent. The study therefore identifies where reconstruction effort and detector design choices can actually buy better physics, and where they cannot.","feed_headline":"In-ice scattering sets the ceiling on cascade direction resolution","feed_subtitle":"IceCube Upgrade study: photon scattering and shower spread, not module precision, dominate zenith-angle error.","key_machinery":"The argument is carried by a controlled information-loss simulation: a photon propagation code generates Cherenkov photons from hadronic cascades in homogeneous ice, and each resolution-limiting process — in-ice scattering, transverse shower spread, module PMT resolution, and module noise — is toggled on or off independently. The shower-spread model uses the empirical angular distribution $dl/dx \\approx \\exp(-2.61\\,x^{0.39})\\,x^{-0.61}$ with $x = 1 - \\cos\\delta$ for the angle $\\delta$ between a charged-track element and the emitted photon, and the ice is described by scattering and absorption coefficients $a_e = 0.0013\\ \\mathrm{m}^{-1}$ and $b_e = 0.013\\ \\mathrm{m}^{-1}$. Reconstruction uses an extended unbinned likelihood over photon direction, residual arrival time, and per-module charge, with PDFs averaged over a randomized detector geometry to remove local geometry bias. That setup lets the authors attribute changes in resolution to individual physical effects.","core_discovery":"The central discovery is a ranking of the processes that destroy directional information before it reaches the photosensors. The paper claims that for GeV-scale hadronic cascades, in-ice photon scattering sets the intrinsic resolution ceiling for zenith-angle reconstruction, with the transverse momentum spread of the shower as the second-largest contribution; module PMT pointing and module noise are nearly irrelevant. The evidence is a controlled simulation campaign in which each process is included or excluded while the others stay fixed, yielding about a 50 percent resolution gain when scattering is turned off and about 30 percent when shower spread is turned off, versus negligible gains from perfect module resolution or noiseless modules. A separate toy simulation indicates that replacing the vertex-averaged integrated photon PDFs with perfectly modeled per-module PDFs would improve resolutions by about 20 percent.","pith_inferences":["Inference: The quoted resolutions are for the hadronic shower axis, not the neutrino direction; at 10 GeV the excluded neutrino–shower opening angle has a mean of about 6 degrees, which can shift $\\cos\\theta$ by roughly 0.1 for near-horizontal events, so oscillation analyses may need an extra smearing term before using these curves.","Inference: The homogeneous-ice idealization probably understates scattering's dominance, because real South Pole ice has layers, birefringence, and hole ice that add distance- and direction-dependent photon distortions; per-module PDFs fitted to real ice might yield more than the 20 percent gain seen in the toy model.","Inference: The same toggle-one-process approach could be applied to water Cherenkov detectors to separate medium scattering from shower physics, since the shower-spread limit should carry over while the scattering limit would shrink.","Inference: The 20 percent estimate is likely a lower bound because the toy simulation samples from the averaged PDF rather than constructing true per-module tables; a full per-module likelihood could recover additional distance-dependent information."],"forward_implications":["If in-ice scattering is the dominant limiter, then a detector medium with much less scattering, such as water, should yield median zenith resolutions roughly 50 percent better than ice at the same photon statistics.","Since module directional resolution is negligible, adding more or smaller PMTs per module would not improve low-energy cascade zenith reconstruction unless it increases the number of detected photons.","Reconstruction algorithms that model per-module photon arrival patterns rather than using vertex-averaged integrated PDFs can expect about a 20 percent improvement in zenith resolution.","Because NC, $\\nu_e$ CC, and $\\nu_\\tau$ CC cascade topologies are indistinguishable at these energies, these resolution limits apply to essentially all low-energy cascade analyses in the IceCube Upgrade.","Existing likelihood reconstructions are already close to the information ceiling set by the physics; the remaining gain from perfect per-module light modeling is about 20 percent."],"supporting_citations":[{"why":"Supplies the measured South Pole ice transparency parameters, including the scattering and absorption coefficients used in the photon propagation simulation.","marker":"[1]"},{"why":"The earlier detector-resolution study that established the method of isolating individual information-loss processes, which this paper adapts to IceCube Upgrade cascades.","marker":"[19]"},{"why":"Provides the measured mDOM PMT dark rate used to set the module-noise level that the paper finds negligible.","marker":"[27]"},{"why":"The photon propagation code used to simulate Cherenkov photon transport through the ice and into the modules.","marker":"[28]"},{"why":"The particle-tracking simulation that generates and propagates the charged shower particles emitting Cherenkov light.","marker":"[31]"},{"why":"The hadronic interaction package that determines the shower's particle content and transverse spread.","marker":"[32]"},{"why":"Provides the empirical parameterization of the Cherenkov photon angular distribution relative to the shower axis, the shower-spread model tested in Section 5.","marker":"[33]"}],"fun_headline_variants":["Photon scattering, not modules, caps cascade zenith resolution","Shower spread and scattering limit IceCube upgrade direction fits","Scattering and shower spread, not modules, set cascade zenith limit","IceCube upgrade: scattering rules, module precision negligible","Cascade angle resolution capped by in-ice scattering, study finds"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the angle between the incoming neutrino direction and the hadronic shower axis is small enough to ignore, so the reconstructed shower axis can be used as the neutrino direction.","fun_headline_variants_meta":{"raw":{"variants":["Photon scattering, not modules, caps cascade zenith resolution","Shower spread and scattering limit IceCube upgrade direction fits","Scattering and shower spread, not modules, set cascade zenith limit","IceCube upgrade: scattering rules, module precision negligible","Cascade angle resolution capped by in-ice scattering, study finds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000167,"raw_usage":{"total_tokens":1209,"prompt_tokens":847,"completion_tokens":362,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":463,"completion_tokens_details":{"reasoning_tokens":278}},"tokens_in":463,"tokens_out":362,"duration_ms":4217,"temperature":1.0,"reasoning_tokens":278,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:18:43.339405+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate low-energy NC events with the full neutrino–shower opening angle included in the truth, reconstruct them with the same likelihood, and compare the residual between reconstructed and true neutrino zenith at the same photon-hit counts. If near-horizontal 10 GeV events, where the mean opening angle of about 6 degrees corresponds to $\\Delta\\cos\\theta \\sim 0.1$, show median residuals appreciably above the quoted resolution curves, then the paper's resolutions bound only the cascade axis, not the neutrino direction.","supporting_citations":[{"cited_title":"Measurement of South Pole ice transparency with the IceCube LED calibration system","cited_arxiv_id":null,"evidence_quote":"Supplies the measured South Pole ice transparency parameters, including the scattering and absorption coefficients used in the photon propagation simulation."},{"cited_title":"Intrinsic limits on resolutions in muon- and electron-neutrino charged-currenteventsintheKM3NeT/ORCAdetector","cited_arxiv_id":null,"evidence_quote":"The earlier detector-resolution study that established the method of isolating individual information-loss processes, which this paper adapts to IceCube Upgrade cascades."},{"cited_title":"Acceptance Tests of more than 10 000 Photomultiplier Tubes for the multi-PMT Digital Optical Modules of the IceCube Upgrade","cited_arxiv_id":null,"evidence_quote":"Provides the measured mDOM PMT dark rate used to set the module-noise level that the paper finds negligible."},{"cited_title":"Fesefeldt","cited_arxiv_id":null,"evidence_quote":"The hadronic interaction package that determines the shower's particle content and transverse spread."},{"cited_title":"The Detection of faint light in deep underwater neutrino telescopes","cited_arxiv_id":null,"evidence_quote":"Provides the empirical parameterization of the Cherenkov photon angular distribution relative to the shower axis, the shower-spread model tested in Section 5."}],"review_version":1}