{"id":"31d7657d-6a12-41e9-8ff0-f7f724f3446a","arxiv_id":"2507.15942","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Stacking annuli around 88,000 HETDEX LAEs reveals Lyα absorption from neutral hydrogen out to ~350 kpc and yields an empirical Wλ(Lyα) radial profile at z~2.5.","lead":"The authors stacked millions of HETDEX spectra around 88,000 distant Lyman alpha emitting galaxies and detected faint hydrogen absorption out to 350,000 parsecs from the galaxies. The result provides a new empirical radial profile of neutral gas around typical galaxies at cosmic noon, which can be compared with galaxy formation simulations.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Mock-galaxy selection on the observed absorption features undermines the independence of the simulation comparison and the 19.3% bias correction.","rationale":"The paper presents a technically careful stacking analysis with multiple control tests, and the detection of Lyα absorption out to ~350 kpc is likely robust: the [O II]-galaxy control, the S/N dilution trend, and the sky-subtraction checks in §7.2 support a real signal. The main quantitative output, however, is the Wλ(Lyα) profile, and the two model-dependent inputs to that output—the 19.3% bias correction (§6.1.2) and the simulation validation (§6.1.3)—both rely on the same 840 ASTRID galaxies that were selected to match the observed central absorption troughs. This selection creates a circularity: the observed profile is corrected and then 'confirmed' using a mock population built to reproduce the observed features. The concern is not that the authors are deceptive; it is that the independence required for a genuine validation is absent. The reader correctly identifies this as the weakest assumption. A random or matched ASTRID sample would break the circularity and clarify whether the simulation genuinely reproduces the H I profile around typical LAEs. If the correction and agreement persist, the paper's conclusions are strengthened; if not, the absolute Wλ normalization and the ASTRID-comparison claim would need to be revised. Conditional acceptance with this additional analysis is the appropriate verdict, so I do not adjust the reader's verdict.","tokens_in":18173,"tokens_out":5856,"duration_ms":60835,"concrete_test":"Recompute the 19.3% bias correction and the mock radial Wλ profile of Figs. 7 and 8 using an ASTRID galaxy sample selected without reference to absorption features (e.g., a random subsample matched to the halo mass and star-formation rate distribution of HETDEX LAEs). If the bias correction shifts by more than the quoted 0.9% statistical error, or the mock profile no longer tracks the observed profile within the bootstrap error envelope, the circular selection is material. Also vary the box-fit selection threshold (e.g., retain the top 10% instead of the reported subset) and verify the correction is stable.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 6.1, the authors select 840 ASTRID galaxies by fitting box-shaped absorption profiles to their central line-of-sight spectra and retaining those that 'best match the observed absorption features.' This same sample is then used for two quantitative purposes: (i) to derive the 19.3 ± 0.9% continuum-normalization bias correction applied to all observed Wλ values (Section 6.1.2, Fig. 7) and (ii) to produce the mock radial Wλ profile that is claimed to agree with the observed profile (Section 6.1.3, Fig. 8). Because the selection directly enriches for galaxies with deep central absorption—presumably correlated with the CGM/IGM gas producing the annulus absorption—the 'agreement' in Fig. 8 is not an independent test of ASTRID's prediction; it is partially ensured by the sample construction. Furthermore, the bias correction is derived from a population preselected on absorption strength. If the true HETDEX LAE population spans a different range of continuum levels or absorption-continuum correlations, the 19.3% uniform correction may be inaccurate, shifting the absolute Wλ profile. The paper does not demonstrate that the correction is insensitive to the selection. This is the most load-bearing weakness because it affects both the absolute normalization of the main quantitative result and the strength of the simulation comparison that the abstract highlights.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper stacks millions of HETDEX fiber spectra in annuli around ~88,000 Lyman Alpha Emitters at 1.9<z<3.5 and reports detection of Ly-alpha absorption out to a projected ~350 kpc. The authors measure a rest-frame equivalent-width profile W_lambda(Ly-alpha) as a function of transverse distance, compare it with mock stacks from the ASTRID simulation, and compare with literature measurements. They derive a ~19.3% bias correction to account for the lack of per-spectrum continuum normalization before stacking, and they argue that the resulting profile is in qualitative agreement with ASTRID and modestly above previous LBG-based profiles, with selection and environmental effects as likely explanations.","tokens_in":18410,"tokens_out":3885,"duration_ms":44478,"significance":"If the central quantitative claims hold, this is a valuable new statistical probe of neutral hydrogen in the CGM around high-redshift LAEs, enabled by the unprecedented HETDEX sample size. The paper includes useful internal controls: the [O II] control stack, the low-S/N dilution test, bootstrap error estimation, and explicit forward-modeling of the HETDEX stacking steps. However, the absolute scale of the measured W_lambda profile and the claimed agreement with ASTRID both depend on a mock sample that is preselected on the very absorption features under study and on an assumed continuum-scatter model, so the numerical results are not as independent as the presentation suggests.","major_comments":[{"comment":"The 840 ASTRID galaxies are selected by fitting box-shaped absorption profiles to central line-of-sight spectra and retaining those that 'best match the observed absorption features' (Section 6.1). This same sample is then used both to derive the 19.3% continuum-normalization bias correction (Section 6.1.2, Fig. 7) and to produce the mock radial W_lambda profile compared with the data (Section 6.1.3, Fig. 8). Because the selection enriches for galaxies with strong absorption, the Fig. 8 agreement is not an independent validation of ASTRID, and the bias correction may not transfer to the full HETDEX LAE population, which is selected on Ly-alpha emission rather than on absorption strength. Please demonstrate the sensitivity of the 19.3% correction to the mock selection, for example by recomputing it using an absorption-unbiased sample matched to the LAE population by stellar mass, halo mass, or SFR, and by reporting the correction for several selection cuts.","section":"Section 6.1, 6.1.2, 6.1.3, Figs. 7-8"},{"comment":"The continuum-normalization procedure uses blue sidebands extending up to 1185-1195 Å, yet Fig. 6 shows a clear suppression in the data between roughly 1180 and 1200 Å that the authors attribute to metal lines (C III 1175, N III 1183, Si II 1190/1193) absent from the mocks. If part of this suppression falls inside the blue continuum window, the pseudo-continuum will be biased low, directly inflating the measured W_lambda values via Eq. (1). The paper should quantify this effect by recomputing W_lambda with the blue continuum restricted to wavelengths blueward of ~1175 Å and by comparing this against the fiducial measurement.","section":"Section 5 and Fig. 6"},{"comment":"The 19.3% bias correction is derived under an assumed model in which each sight-line spectrum is multiplied by a random factor drawn from a truncated N(1, 0.5) distribution. The magnitude of the correction is expected to depend on the assumed scatter amplitude and on the correlation between continuum level and absorption strength, but the paper provides no test of this dependence. Please show how the correction changes when the scatter is varied over a plausible range and when the mock continuum fluctuations are instead calibrated directly to the HETDEX background-light measurements cited from Weiss et al. 2025.","section":"Section 6.1.2"}],"minor_comments":[{"comment":"The notation 'nline' and 'ncont' is introduced in Eq. (2) but never defined; please define the number of line-region and continuum-region choices explicitly, since the total of 100 combinations follows from those definitions.","section":"Section 5"},{"comment":"The claim of 'good qualitative agreement' between the observed and mock radial profiles is not accompanied by a quantitative statistic. A chi-square or residual-based measure, even with a simple covariance model, would substantially strengthen the comparison.","section":"Section 6.1.3, Fig. 8"},{"comment":"The caption says 'The Lyα absorption is apparent in each spectrum until the largest radial bin at ~350 kpc,' but the text in Section 4 states the detection extends 'up to Dtran of ~350 kpc.' Please make the caption wording consistent with the actual outermost bin (e.g., 340-360 kpc) and with the significance of the detection in that bin.","section":"Figure 4 caption"},{"comment":"There is a grammatical error in 'Also LAEs whose their Lyα emitting region lies deeper into dusty ISM clumps will have lower intrinsic Lyα escape fractions'; please rephrase.","section":"Section 8"}],"recommendation":"major_revision","confidential_remarks":"The central detection of extended Ly-alpha absorption is well supported by internal controls and is likely robust. The main issue is that the quantitative normalization and the simulation comparison rely on a preselected mock sample and an assumed scatter model, which is a fixable but load-bearing methodological gap. I would encourage the editor to request the robustness tests outlined in the major comments rather than reject the paper, as the observational result is interesting and the required additional analyses appear feasible within the scope of a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know about this one. First, the core result—a radial Lyα absorption profile out to ~350 kpc around ~88,000 LAEs, built by stacking HETDEX fibers in annuli outside the galaxies—is a genuinely new measurement, and the detection itself looks believable. Second, the quantitative normalization and the headline comparison to ASTRID simulations both rest on a mock sample that was preselected to match the absorption features under study. That coupling is the main thing to probe if you referee it.\n\nWhat is actually new: prior work from this team stacked only central apertures; here they use annuli from 40–360 kpc and measure Wλ(Lyα) as a function of projected distance. That is a useful complement to quasar-sightline statistics and LBG stacking. The internal controls are decent: no absorption around [OII] emitters, the feature weakens at low S/N rather than growing (so it is not a false-positive artifact), and it survives sky-subtraction checks. The paper also explicitly acknowledges major caveats, including Lyα velocity-offset broadening and possible metal contamination in the continuum sidebands. That honesty earns credit.\n\nWhere it gets soft. The 19.3% bias correction is derived from 840 ASTRID galaxies selected because their box-shaped absorption profiles best match the observed ones. The same sample produces the mock radial profile that is then said to agree with the data. So the agreement is partly manufactured, and the absolute Wλ scale depends on how well that preselected sample represents the full LAE population. They do not show the correction is stable under a different selection or what a random ASTRID sample would give. That is load-bearing for the normalization, but not for the existence of the signal. The second soft spot is the 1180–1200 Å suppression they attribute to metal lines not in the mocks; those lines sit on the blue continuum reference for the Wλ measurement. They flag it but do not quantify the bias. It is probably a few percent, but it deserves a sensitivity check.\n\nBottom line: this paper deserves a serious referee. The detection is likely real, the method is new, and the caveats are mostly acknowledged. What is needed is a robustness demonstration that the bias correction and the simulation agreement are not artifacts of the preselected mock sample. I would send it to peer review with a request for those tests plus a release of the stacked spectra or at least the profile table. Not desk-rejectable; a revise-and-resubmit target.","headline":"A new and mostly sound stacked measurement of Lyα absorption around ~88,000 LAEs, but the mock-calibrated bias correction and the simulation comparison are coupled to the very absorption signal they aim to validate.","tokens_in":19059,"tokens_out":2630,"would_cite":true,"duration_ms":30364,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"By stacking millions of fiber spectra around ~88,000 Lyman-alpha emitters, this paper detects neutral-hydrogen absorption out to ~350 kpc and derives an empirical radial equivalent-width profile that a hydrodynamic simulation reproduces.","keywords":["circumgalactic medium","Lyman-alpha emitters","neutral hydrogen","Lyman-alpha absorption","spectral stacking","high-redshift galaxies","HETDEX","radial equivalent width profile"],"falsifier":"Recompute the 19.3% continuum-normalization correction using ASTRID galaxies selected on properties unrelated to absorption shape, such as halo mass or star-formation rate; if the new correction differs by more than the quoted uncertainty, the reported profile shifts. Alternatively, target quasar sightlines passing within 40-350 kpc of spectroscopically confirmed LAEs at $z\\sim2.5$; if the resulting individual $W_\\lambda(\\mathrm{Ly}\\alpha)$ values lie systematically below the stacked HETDEX profile by more than the error bars, the stacked measurement is inflated.","tokens_in":17951,"feed_emoji":"🔭","tokens_out":12801,"duration_ms":117638,"temperature":0.7,"pith_summary":"The paper aims to show that the diffuse neutral hydrogen around ordinary high-redshift star-forming galaxies can be mapped statistically without quasars: by stacking millions of fiber spectra in annuli around ~88,000 Lyman-$\\alpha$ emitters at $1.9<z<3.5$, it detects Lyman-$\\alpha$ absorption out to a projected ~350 kpc and converts it into an empirical radial profile of rest-frame equivalent width $W_\\lambda(\\mathrm{Ly}\\alpha)$. If the detection is real, it matters because it turns the circumgalactic medium of typical galaxies at cosmic noon into a measurable quantity for a sample thousands of times larger than quasar-sightline studies allow. The paper also argues that the resulting profile agrees qualitatively with the ASTRID hydrodynamic simulation's diffuse H I distribution, and that the absorption is strongest for low signal-to-noise LAEs, which it interprets as evidence that LAEs embedded in overdense H I regions have their Lyman-$\\alpha$ emission suppressed.","feed_headline":"Neutral hydrogen detected to 350 kpc around 88,000 early galaxies","feed_subtitle":"Stacking millions of spectra maps the gas halo around typical galaxies and gives simulations a profile to match.","key_machinery":"The load-bearing method is annulus spectral stacking: for each of ~88,000 LAEs, spectra from vetted fibers in 20 kpc annuli between 40 and 360 kpc are biweight-averaged, shifted to the LAE rest frame, and then stacked across the sample, so that the integrated light of background galaxies acts as a statistical continuum source. The measured quantity is a non-parametric rest-frame equivalent width, integrated over the Lyman-alpha trough with 100 combinations of continuum windows and integration ranges to carry systematic errors. The comparison tool is forward-modeled mock spectra from the ASTRID simulation, which are used both to reproduce the observed profiles and to calibrate the 19.3% correction for not continuum-normalizing individual spectra before stacking.","core_discovery":"On its own terms, the paper establishes that an average LAE at $z\\sim2.5$ is surrounded by neutral hydrogen that imprints a measurable Lyman-$\\alpha$ absorption trough on the integrated light of background galaxies, for projected separations from roughly 40 kpc to 350 kpc. After stacking spectra from about three million vetted fibers per radial bin and correcting for the ~19.3% bias introduced by normalizing the continuum only after stacking, the paper obtains a radial $W_\\lambda(\\mathrm{Ly}\\alpha)$ profile that declines with distance, matches the median-stacked LAE measurements from earlier work, lies modestly above continuum-selected Lyman-break-galaxy profiles, and is reproduced in depth and shape by mock spectra built from the ASTRID simulation. The paper further claims that the absorption-to-emission switch seen when binning by signal-to-noise reflects a physical hierarchy: LAEs sitting deeper in dense H I, noisier environments, or dusty interstellar clumps have suppressed Lyman-$\\alpha$ emission, so the lowest-signal LAEs are precisely the ones whose halos show the strongest absorption.","pith_inferences":["(Editorial inference) The 19.3% correction is calibrated on mock galaxies selected for matching the very absorption shapes under study, so the corrected equivalent widths and the claimed agreement with ASTRID are not fully independent; a correction computed from an ASTRID sample selected on halo mass or star-formation rate rather than absorption shape would test whether the calibration holds.","(Editorial inference) The unexplained continuum suppression between 1180 and 1200 Angstroms in the data but not the mocks hints that future, deeper stacks could detect metal-line absorption (C III, N III, Si II) in the same CGM gas, turning this method into a multi-species probe.","(Editorial inference) If the environmental picture is right, the profile should steepen when stacking only LAEs in overdense regions; the larger sample expected in the next data release could measure that directly, or disprove it if the profile stays flat.","(Editorial inference) The technique's dependence on a statistical continuum assumes background-galaxy light is uniform enough to normalize; quasars with foreground LAEs at the same redshifts could provide an independent, per-sightline check of the profile normalization."],"forward_implications":["A neutral-hydrogen census of galaxy halos at $z\\sim2.5$ no longer needs rare quasar sightlines; any sufficiently deep wide-field spectroscopic survey with dense fiber coverage can map diffuse CGM gas statistically.","Simulations of galaxy formation now have a direct, spatially resolved observable, the radial $W_\\lambda(\\mathrm{Ly}\\alpha)$ profile out to 350 kpc, that they must match rather than only reproduce as an average column density.","The systematic offset between LAE and Lyman-break-galaxy absorption profiles is partly a selection effect: continuum-bright LAEs show weaker absorption, so continuum-selected samples sit below the full LAE profile.","The signal-to-noise dependence of the stacked absorption implies that environment and dust geometry regulate the observed Lyman-alpha flux, and future samples can test this by splitting the stack by local density, redshift, and Lyman-alpha luminosity."],"supporting_citations":[{"why":"Establishes that central HETDEX LAE spectra show box-like Lyman-alpha absorption troughs, which motivates the galaxy selection and geometry used in the mock comparison.","marker":"Weiss et al. 2024"},{"why":"Shows that integrated background-galaxy light is detectable in HETDEX and supplies the ~50% UV-background continuum scatter used in the mock spectra.","marker":"Weiss et al. 2025"},{"why":"Provides the original spectral-stacking and pre-stacking correction methodology that this paper extends to annuli around LAEs.","marker":"Davis et al. 2023a"},{"why":"Presents the ASTRID cosmological simulation from which the 840 mock galaxies and their line-of-sight spectra are drawn.","marker":"Bird et al. 2022"},{"why":"Shows ASTRID reproduces the high-column-density absorber population at z~2-3, supporting the simulation's H I distribution as a benchmark.","marker":"Bird et al. 2023"},{"why":"Supplies the z~2.3 Lyman-break-galaxy radial Ly-alpha equivalent-width profile that the measured LAE profile is compared against.","marker":"Chen et al. 2020"},{"why":"Provides mean- and median-stacked z~3.3 LAE equivalent-width measurements, including the overdensity enhancement used to interpret the HETDEX profile.","marker":"Muzahid et al. 2021"},{"why":"Offers individual Lyman-break-galaxy equivalent-width data points at z~2.3 that anchor the low end of the comparison profile.","marker":"Steidel et al. 2010"},{"why":"Adds further Lyman-break-galaxy equivalent-width measurements used in the literature comparison.","marker":"Turner et al. 2014"}],"fun_headline_variants":["HETDEX maps neutral hydrogen to 350 kpc around 88,000 galaxies","Neutral hydrogen detected around 88,000 early galaxies out to 350 kpc","Stacking spectra reveals gas halo around 88,000 Lyman-alpha emitters","88,000 LAEs yield neutral hydrogen profile to 350 kpc"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central numbers rest on the assumption that the 840 simulated galaxies used to calibrate the correction and validate the profile are representative of real LAE halos; because those galaxies were chosen for having absorption features like the ones being measured, any mismatch between the mocks and reality propagates directly into the corrected equivalent widths and the agreement claim.","fun_headline_variants_meta":{"raw":{"variants":["HETDEX maps neutral hydrogen to 350 kpc around 88,000 galaxies","Neutral hydrogen detected around 88,000 early galaxies out to 350 kpc","Stacking spectra reveals gas halo around 88,000 Lyman-alpha emitters","88,000 LAEs yield neutral hydrogen profile to 350 kpc"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000588,"raw_usage":{"total_tokens":2807,"prompt_tokens":1037,"completion_tokens":1770,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":653,"completion_tokens_details":{"reasoning_tokens":1682}},"tokens_in":653,"tokens_out":1770,"duration_ms":13374,"temperature":1.0,"reasoning_tokens":1682,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:21:42.863383+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the 19.3% continuum-normalization correction using ASTRID galaxies selected on properties unrelated to absorption shape, such as halo mass or star-formation rate; if the new correction differs by more than the quoted uncertainty, the reported profile shifts. Alternatively, target quasar sightlines passing within 40-350 kpc of spectroscopically confirmed LAEs at $z\\sim2.5$; if the resulting individual $W_\\lambda(\\mathrm{Ly}\\alpha)$ values lie systematically below the stacked HETDEX profile by more than the error bars, the stacked measurement is inflated.","supporting_citations":[],"review_version":1}