{"id":"d1bfb1a1-ba95-41a5-9bed-91789308648d","arxiv_id":"2608.10066","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The last-crossing scale, the smallest scale where a region's own photon budget still balances, separates externally ionized underdense regions with central source deficits, linking source geometry to Lyα transmission.","lead":"This paper introduces a new statistic, the last crossing of the photon-counting barrier, to describe how ionized bubbles in cosmic reionization are sustained by sources on different scales. It shows that regions ionized by distant sources are underdense and source-poor at their centers, which may explain Lyα transmission spikes in quasar spectra.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The resolved/unresolved R_ell split is defined relative to the simulation's minimum smoothing scale, not the physical source-halo scale, so the empirical source-environment contrasts in Fig. 9 are resolution-dependent until shown otherwise.","rationale":"Reading the paper in good faith, the central contribution is a new last-crossing statistic for the FZH04 barrier. The analytic derivation in Sections 2–3 is careful, the finite-endpoint correction is physically motivated, and the Monte Carlo validation in Appendix A gives independent support for the solver. The empirical analysis in Sections 4–6 is where the claim becomes vulnerable. The paper itself flags that the resolved/unresolved classification is defined relative to the minimum resolved smoothing scale, not the source-halo scale, which is exactly the point on which the empirical source-environment contrasts depend. If the endpoint were moved, the class labels would change, and Fig. 9 is a comparison between those labels. The reader identified this same weakest assumption, and I agree it is the load-bearing concern. I do not see an internal inconsistency in the analytic crossing equations; the disagreement with the FZH04 analytic shape in Fig. 4 is presented as a physical consequence of nonlocal photon transport, not as a flaw. The observational Ly-alpha prediction is explicitly speculative and testable, so it does not undermine the central claim. The appropriate verdict remains CONDITIONAL: the analytic formalism is credible, but the empirical claims need a resolution robustness test, a specified matching procedure, and public code/data before they can be accepted as established. My pass does not move the reader's verdict.","tokens_in":16377,"tokens_out":11331,"duration_ms":117925,"concrete_test":"Recompute the resolved/unresolved classification and the Fig. 9 matched source-density ratios with the trajectory endpoint modified to a coarser minimum smoothing scale, e.g., truncating every trajectory at 2 R_min (and, if a higher-resolution CROC run or a rerun with a smaller smoothing scale is available, at 0.5 R_min). If the central-shell deficit and intermediate-shell excess shift by more than the paired-bootstrap 16th–84th percentile intervals, the empirical claim is resolution-dependent. As a complementary check, repeat the analysis using only resolved regions with R_ell > 2 R_min, which are less sensitive to the precise endpoint placement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's empirical claim — that resolved-R_ell regions show a central deficit and intermediate-radius excess in ionizing-luminosity and galaxy-number density relative to matched unresolved regions — rests on a classification whose boundary is not a physical scale. Section 4 states this explicitly: \"The endpoint of the simulation trajectory is the minimum resolved smoothing scale in the trajectory construction, not the source-halo scale, and the resolved/unresolved R_ell classification below is defined with respect to this scale.\" Consequently, the unresolved class contains two populations that cannot be separated at the current resolution: regions whose photon-counting condition genuinely holds to the halo scale, and regions that would show an interior last crossing if the trajectory were continued below R_min. A higher-resolution trajectory set would move some currently unresolved regions into the resolved class, changing both the composition of the samples and the matched ratios in Fig. 9. The direction and magnitude of the change are unknown; the reported factor-of-two central deficit could be diluted or enhanced. This concern targets only the simulation-based source-environment claims, not the analytic Volterra derivation, which is independently Monte Carlo validated in Appendix A and appears sound. The matching step for Fig. 9 also lacks stated tolerances, so residual differences in the matched variables could contribute to the contrast, but the resolution dependence is the more fundamental issue.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a new statistic for excursion-set models of cosmic reionization: the last crossing of the FZH04 photon-counting barrier, denoted R_ℓ, defined as the smallest smoothing scale on which the enclosed photon budget still ionizes the enclosed gas. After the first crossing gives the enclosing ionized-bubble scale, the last crossing identifies the smallest aperture on which the same condition holds; trajectories that remain above the barrier down to the minimum source-halo scale contribute a discrete endpoint population. The author derives an analytic last-crossing distribution using a finite-endpoint Volterra equation of the first kind, validates the solver against Monte Carlo sharp-k random walks in Appendix A, and then applies the same crossing statistics to an empirical barrier extracted from CROC radiative-transfer simulations following Kaurov (2016). The empirical analysis finds that resolved-R_ℓ regions are predominantly underdense, reionize later, and, when matched in local density, first-crossing scale, and reionization redshift, show a central deficit and intermediate-radius excess in ionizing-luminosity and galaxy-number density relative to unresolved regions. The author interprets these regions as externally ionized and suggests they may preferentially host Lyα transmission spikes.","tokens_in":16685,"tokens_out":6009,"duration_ms":60716,"significance":"The analytic part of the paper is a genuine contribution: the last-crossing formalism is new for the FZH04 barrier, the finite-endpoint treatment is explicit, and the Volterra solution is independently checked with Monte Carlo realizations. These strengths should be credited. If the empirical claims hold, the statistic connects the internal source geometry of ionized regions to testable galaxy–IGM measurements around transmission-selected locations. However, the empirical conclusions currently rest on a resolution-dependent classification and on an empirical barrier derived from the same ionization field whose crossings are then measured. The central analytic derivation appears sound; the empirical part needs additional robustness work before the source-environment contrasts can be regarded as physical rather than operational.","major_comments":[{"comment":"The resolved/unresolved R_ℓ classification is defined with respect to the minimum resolved smoothing scale in the CROC trajectory construction, not the physical source-halo scale, as stated in Sec. 4. The unresolved sample therefore contains all regions whose photon budget closes on scales below R_min, and a higher-resolution trajectory set would reclassify some of them as resolved. Because the source-environment contrasts in Fig. 9 are ratios between samples defined by this resolution boundary, the reported factor-of-two central deficit and intermediate-radius excess are not yet shown to be stable physical properties of externally ionized regions. Please add a convergence test, for example by repeating the crossing classification with the trajectory endpoint deliberately coarsened to several larger R_min values and checking whether the matched ratios in Fig. 9 persist; if this is not computationally feasible, the physical claims should be explicitly restricted to the operational, resolution-dependent definition.","section":"Sec. 4, text after Eq. (34); Sec. 6, Fig. 9"},{"comment":"The matching procedure for Fig. 9 is described only as holding fixed local density, first-crossing scale R_f, and reionization redshift, with no tolerances, binning scheme, or algorithm given. If the matching is not sufficiently tight, residual differences in these variables between the resolved and unresolved samples could contribute to the central deficit and intermediate excess. Please specify the exact matching construction (e.g., nearest-neighbor in log-density, log-R_f, and z_rei with stated tolerances or kernel widths), show the distributions of the matched variables before and after matching, and include a null control using randomly paired samples to calibrate the expected contrast.","section":"Sec. 6, Fig. 9 and matching description"},{"comment":"Because B_sim(R) is defined from the same CROC ionization field whose crossings are subsequently measured, and because the analytic ζ in Fig. 4 is chosen to match the simulated ionized fraction, the comparison in Fig. 4 is not an independent validation of the analytic model. The extended large-R tails in CROC are consistent with nonlocal photon transport, but they could also reflect the particular C_neutral=0.5 barrier definition, the Gaussianization of the density field, or the sharp-k smoothing filter. Please add a sensitivity test varying the barrier threshold (for example using the C_neutral=0.25 and 0.75 edges of the transition region) and refitting ζ, to show that the qualitative shape comparison and the resolved/unresolved split are robust to the empirical-barrier construction.","section":"Sec. 4, Eq. (34) and Fig. 4"}],"minor_comments":[{"comment":"The phrase 'prior-weighted neutral and ionized distributions' is ambiguous; please define the priors explicitly (presumably the volume fractions of each class at the snapshot redshift).","section":"Sec. 4, paragraph after Eq. (34)"},{"comment":"The 'approximately self-similar' evolution of the last-crossing distribution is asserted qualitatively; consider quantifying it with, for example, the ratio of the median or characteristic scales at different ionized fractions.","section":"Sec. 3 and Fig. 3"},{"comment":"The volume fractions 50%, 34%, and 16% are for one snapshot; please state in the caption or text that these values are specific to z=8.0 and will evolve with redshift.","section":"Fig. 6 caption"},{"comment":"The Brownian-bridge correction approximates the barrier as linear across each interval; since the FZH04 barrier curvature is strongest near S_min, please state explicitly that the u-grid (or equivalent) keeps the linear approximation accurate near the endpoint.","section":"Appendix A, Eq. (A4)"},{"comment":"When first citing the two papers by Zhu and coauthors, the text should disambiguate H. Zhu et al. (2024) from Y. Zhu et al. (2026) to avoid reader confusion, since both are relevant to the Lyα transmission discussion.","section":"Sec. 6, first paragraph"}],"recommendation":"major_revision","confidential_remarks":"The analytic derivation and its Monte Carlo validation are the strongest part of the paper and appear sound. The empirical sections are suggestive but currently overstate the robustness of the resolution-dependent classification. I recommend major revision rather than rejection because the analytic core is a genuine contribution and the empirical claims are potentially correct, but they need the convergence tests and matching details described in the major comments."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The analytic core is solid and genuinely new: the last-crossing statistic for the FZH04 barrier has not been done before, and the finite-endpoint Volterra first-kind equation is a clean adaptation of Hopkins (2012). The Monte Carlo validation in Appendix A is real evidence. The empirical part is the soft spot: the resolved/unresolved R_ell split is defined relative to the simulation's minimum smoothing scale, not a physical source-halo scale, so the Fig. 9 contrasts are resolution-dependent until shown otherwise.\n\nThe paper does well at separating the analytic and empirical claims. The derivation is internally consistent, with a normalization check that q_int + p_end equals the first-crossing probability. The numerical solver in u = sqrt(S_i - S) space is sensible. Measuring both first and last crossings in CROC with an empirical barrier following Kaurov (2016) is a reasonable use of the simulation.\n\nThe main weaknesses are in the simulation analysis. The empirical barrier is constructed from the same ionization field whose crossings are then counted, and zeta is fitted to the ionized fraction, so the CROC comparison is not an independent test. The stress-test note is right: Section 4 states plainly that the endpoint is the minimum resolved smoothing scale, not the source-halo scale. That means 'unresolved' includes two populations that higher resolution would separate. The direction of the change in Fig. 9 is unknown; the factor-of-two deficit could be diluted or enhanced. This does not undermine the analytic derivation, but it does mean the empirical claims should be treated as suggestive until a resolution test is done. Also, no code or data is released, and the matching tolerances for Fig. 9 are not stated.\n\nWho gets value: anyone working on excursion-set reionization models, the galaxy-IGM connection, or Ly-alpha transmission environments. The last-crossing statistic is a useful addition to the toolkit, and the prediction for stacked galaxy profiles around transmission spikes is testable.\n\nRecommendation: send to peer review. A serious referee should focus on the resolution dependence of the empirical classification and ask for a robustness test or explicit caveat, plus code/data. The analytic result deserves publication regardless.","headline":"The last-crossing statistic is a genuinely new analytic result, well validated by Monte Carlo, but the simulation-based empirical claims rest on a resolution-dependent classification that needs a robustness test.","tokens_in":17146,"tokens_out":2120,"would_cite":true,"duration_ms":19426,"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":"The paper introduces the last crossing of the photon-counting barrier as a new statistic for reionization, separating externally ionized underdense regions from internally ionized ones and predicting a central galaxy deficit around Lyα…","keywords":["last crossing","excursion-set theory","cosmic reionization","FZH04 barrier","photon counting","intergalactic medium","Lyα transmission","radiative transfer"],"falsifier":"Run the same empirical-barrier analysis on a higher-resolution radiative-transfer simulation, or on CROC with a finer smoothing grid, and measure the radial source-density profile around $R_\\ell$: if the central deficit and intermediate excess diminish or reverse once more small-scale structure is resolved, the central empirical claim is wrong. Independently, stack galaxy catalogs around observed Lyα transmission spikes; the predicted central galaxy deficit with an excess at finite separation would be refuted by a flat or monotonically increasing profile.","tokens_in":16192,"feed_emoji":"🔭","tokens_out":8183,"duration_ms":67445,"temperature":0.7,"pith_summary":"The paper introduces the last-crossing scale $R_\\ell$ of the FZH04 photon-counting barrier: the smallest smoothing radius at which the enclosed galaxies still produce enough ionizing photons to ionize the enclosed gas. It derives the analytic distribution of this scale, including an endpoint contribution, and validates it against Monte Carlo random walks. Applied to a radiative-transfer simulation through an empirical barrier, the formalism separates externally ionized regions, whose photon budget closes only on large scales, from internally ionized regions that satisfy the condition to the resolution scale. These externally ionized regions are predominantly underdense, reionized late, and show a central deficit and intermediate-radius excess of ionizing sources relative to matched controls. The author argues they are natural hosts for Lyα transmission spikes, giving a testable prediction for galaxy densities around transmission-selected IGM locations.","feed_headline":"Reionization's last crossing tells which gas is lit from outside","feed_subtitle":"The R_ℓ scale separates externally lit underdense gas from internally lit gas, with a testable Lyα prediction.","key_machinery":"The load-bearing object is the FZH04 photon-counting barrier $B(S,z) = \\delta_c(z) - \\sqrt{2}\\,K(\\zeta)\\sqrt{S_{\\rm min} - S}$, where $K(\\zeta) = \\operatorname{erfc}^{-1}(1/\\zeta)$, and the conjugate crossing rule $S_\\ell = \\max\\{S : \\delta(S) \\geq B(S,z)\\}$. The analytic machinery is a Volterra integral equation of the first kind, $A(S) = \\int_{S}^{S_i} f_\\ell(S')\\,K(S,S')\\,dS'$, where $A(S)$ is a bivariate Gaussian probability that the trajectory lies above the barrier at $S$ and below it at the endpoint $S_i = S_{\\rm min}$, and $K(S,S')$ is the sharp-$k$ conditional probability of being above at $S$ given a crossing at $S'$. The equation is inverted by forward substitution on a grid linear in $u = \\sqrt{S_i - S}$, plus the endpoint atom $p_{\\rm end} = 1 - \\Phi(\\delta_c(z)/\\sqrt{S_i})$. The same crossing definitions applied to the empirical neutral/ionized conditional barrier from CROC yield the resolved/unresolved $R_\\ell$ classification that drives the environmental analysis.","core_discovery":"The central claim is that the last crossing of the FZH04 photon-counting barrier, defined by $S_\\ell = \\max\\{S : \\delta(S) \\geq B(S,z)\\}$ with $B(S,z) = \\delta_c(z) - \\sqrt{2}\\,K(\\zeta)\\sqrt{S_{\\rm min} - S}$, is a physical statistic rather than a mathematical artifact. Because trajectories end at a finite source-halo endpoint $S_{\\rm min}$, the last-crossing distribution splits into a smooth interior density and a discrete endpoint atom, and the interior density obeys a Volterra integral equation that the paper solves deterministically. Measured in CROC via the Kaurov empirical barrier, resolved last-crossing regions form coherent structures tracing the edges of ionized regions around neutral pockets, are predominantly underdense, and, at fixed local density, first-crossing scale, and reionization redshift, differ from matched unresolved regions in source environment: roughly half the central ionizing-luminosity density and a one-third deficit in galaxy number density, with excesses of about 75% and 65% at intermediate radius. The paper interprets this as gas that is externally ionized—sustained by sources on larger scales rather than by its own small-scale photon budget.","pith_inferences":["The same finite-endpoint Volterra construction could be transported to other excursion-set problems with a physically motivated minimum scale, such as the star-formation and IMF fragmentation problem that motivated Hopkins's equation, to separate internally from externally fed fragments.","The resolved/unresolved split is resolution-relative by construction; a finer grid would convert some endpoint-atom regions into interior crossings, so the class fractions and the Figure 9 contrasts are predictions of the analysis scale, not just of reionization physics.","If the Lyα prediction survives mock-spectra tests, $R_\\ell$ could become a bridge statistic connecting 21 cm bubble morphology to transmission-spike statistics, two observables currently modeled in separate communities.","The analytic first- and last-crossing pair could be used as a fast two-population likelihood for inferring $\\zeta$ and $M_{\\rm min}$ from observations of bubble sizes and transmission-spike environments jointly."],"forward_implications":["Ionized gas splits into two observable classes: unresolved regions whose own photon budget closes at the smallest resolved aperture, and resolved regions that stay ionized only through photons counted on larger apertures.","Among resolved regions, larger $R_\\ell$ selects lower gas density and later reionization, so the statistic reads off a region's position in the density–reionization-time plane.","$R_\\ell$ adds information beyond $\\delta$, $R_f$, and $z_{\\rm rei}$: matched regions differ by about a factor of two in central luminosity density and by one third in central galaxy number density, with intermediate-radius excesses near 75% and 65%.","If transmission spikes preferentially inhabit resolved regions, stacked galaxy profiles around Lyα transmission-selected IGM locations should show a central deficit and a finite-separation excess relative to matched control locations.","The extended large-$R_\\ell$ tail relative to the analytic FZH04 model quantifies the importance of nonlocal photon transport in sustaining ionized regions."],"supporting_citations":[{"why":"Defines the photon-counting barrier and the first-crossing bubble scale that this paper conjugates into the last-crossing statistic.","marker":"S. R. Furlanetto et al. (2004)"},{"why":"Supplies the absorbing-barrier first-crossing construction and the Markov sharp-k random-walk statistics used in the analytic derivation.","marker":"J. R. Bond et al. (1991)"},{"why":"Derives the Volterra last-crossing equation in a different physical setting, which this paper adapts to the FZH04 barrier with a finite endpoint.","marker":"P. F. Hopkins (2012)"},{"why":"Provides the empirical-barrier method used to measure crossings in CROC trajectories.","marker":"A. A. Kaurov (2016)"},{"why":"Provides the CROC radiative-transfer simulation whose trajectories carry the empirical analysis.","marker":"N. Y. Gnedin (2014)"},{"why":"Establishes that low gas density is the primary condition for transmission spikes in CROC, motivating the proposed Lyα association.","marker":"H. Zhu et al. (2024)"},{"why":"Reports galaxy underdensities and a tentative finite-separation transmission correlation that this paper's prediction can test.","marker":"Y. Zhu et al. (2026)"}],"fun_headline_variants":["Last crossing of photon barrier spots externally ionized gas","External vs internal reionization told by last crossing scale","Last crossing scale reveals source geometry around ionized bubbles","Where Lyα spikes arise: last crossing of reionization pinpointed","New statistic: last crossing separates outside-lit from inside-lit gas"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification of a region as resolved or unresolved depends on how finely the simulation grid is smoothed, and a finer grid could move regions from one class to the other, changing the reported source-density contrasts.","fun_headline_variants_meta":{"raw":{"variants":["Last crossing of photon barrier spots externally ionized gas","External vs internal reionization told by last crossing scale","Last crossing scale reveals source geometry around ionized bubbles","Where Lyα spikes arise: last crossing of reionization pinpointed","New statistic: last crossing separates outside-lit from inside-lit gas"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000211,"raw_usage":{"total_tokens":1508,"prompt_tokens":1135,"completion_tokens":373,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":751,"completion_tokens_details":{"reasoning_tokens":290}},"tokens_in":751,"tokens_out":373,"duration_ms":4031,"temperature":1.0,"reasoning_tokens":290,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:14:20.537036+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same empirical-barrier analysis on a higher-resolution radiative-transfer simulation, or on CROC with a finer smoothing grid, and measure the radial source-density profile around $R_\\ell$: if the central deficit and intermediate excess diminish or reverse once more small-scale structure is resolved, the central empirical claim is wrong. Independently, stack galaxy catalogs around observed Lyα transmission spikes; the predicted central galaxy deficit with an excess at finite separation would be refuted by a flat or monotonically increasing profile.","supporting_citations":[],"review_version":1}