{"id":"6a19d2a1-32e5-4767-b28c-f7ad83023aec","arxiv_id":"1909.02513","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Strong coherent Lyman-alpha absorption is not a unique or complete tracer of protoclusters at z about 2.4, according to Sherwood, EAGLE and Illustris simulations.","lead":"This paper uses three hydrodynamical simulations to test whether strong Lyman-alpha absorption can identify protoclusters at z about 2.4. It finds the method is highly incomplete and often contaminated, so such absorption alone cannot select protoclusters.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The headline fractions depend on the simulated LLS/SLLS population, whose line-of-sight clustering is never validated; the CDDF check in Fig. 1 has no displayed data in the 10^17-10^20 cm^-2 range that drives CoSLAs after DLA removal.","rationale":"The reader's conditional verdict identifies the simulation of high-column-density absorbers as the weakest assumption, and I agree. The paper is a careful simulation study: it uses three independent codes, explicitly tests resolution and box size in Section 5.3, compares against the QLy-alpha run to isolate the role of dense gas, and flags its own main limitations, including the absence of 10^15 M_sun progenitors and the possibly fortuitous high-NHI CDDF agreement. The central negative result, that CoSLAs are not unique tracers of M~10^14 M_sun protoclusters, is supported by multiple independent lines: CoSLAs span underdense volumes (Figure 7), many are not associated with protoclusters (Figure 15), and 84 per cent of protocluster sightlines are indistinguishable from the field (Figure 13). The 0.1 per cent completeness is so small that even a factor-of-two systematic in absorber modelling would not reverse the qualitative incompleteness conclusion. However, the exact purity figure, about half or 55 per cent, is a quantitative claim central to the paper's message, and it rests on only 22 Sherwood CoSLAs and on a simulated LLS/SLLS population whose line-of-sight clustering is not validated. The displayed CDDF validation in Figure 1 has no points in the 10^17-10^20 cm^-2 range that dominates post-DLA CoSLAs. Thus the verdict remains conditional: the qualitative conclusion is secure, but the headline fractions should be presented with uncertainties and, ideally, with a clustering check. My proposed check targets exactly that gap.","tokens_in":26214,"tokens_out":9827,"duration_ms":112462,"concrete_test":"Measure the 1D line-of-sight two-point correlation function (or close-pair frequency) of LLSs and SLLSs (10^17.2 <= NHI/cm^-2 < 10^20.3) in the Sherwood, EAGLE and Illustris mock sightlines at z~2.4, and compare with observed absorber clustering from quasar close pairs and absorber-galaxy cross-correlations at z~2-3. Then reweight or resample the simulated LLS/SLLS populations to match the observed clustering while holding the CDDF fixed, rerun the CoSLA selection of Section 5.2, and recompute the Sherwood association fraction (Figure 15) and the contamination/completeness curves (Figure 16). If the association fraction moves outside the binomial range expected from 22 CoSLAs (roughly 33-75 per cent), the central 'about half' claim needs revision; if it stays inside, the concern is retired.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After damped systems are removed, CoSLAs are produced by LLSs and high-column Ly-alpha forest systems (Section 5.2, Figures 15 and 16). The central non-uniqueness claim, e.g. 55 per cent of CoSLAs trace protoclusters in Sherwood, therefore rests on these absorbers being realistic in both abundance and, crucially, in their alignment along 15 h^-1 cMpc sightlines. The validation in Section 3 and Figure 1 checks only the one-point CDDF, with plotted observational points at NHI < 10^17 and NHI > 10^20 cm^-2 but not in the 10^17-10^20 LLS/SLLS interval that dominates the post-DLA CoSLA population; the authors themselves call the high-NHI agreement 'possibly fortuitous.' Nothing in Section 3 or later tests the line-of-sight clustering of LLSs and SLLSs, which is what makes a 15 h^-1 cMpc segment coherently strong. The CoSLA count varies by a factor ~1.6 across simulations (22, 32, 36), and Section 5.3 shows the incidence is not converged with resolution, so the 55/45 split in Figure 15 is not pinned down. If real LLS/SLLS clustering is weaker than simulated, field CoSLAs would be rarer and purity could rise; if stronger, purity could fall. The 0.1 per cent completeness is more robust because the denominator is thousands of protocluster sightlines, but the quantitative 'about half' statement is not secure without a clustering check.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses three hydrodynamical simulations (Sherwood, EAGLE, Illustris) at z ~ 2.4 to test whether coherently strong Ly-alpha absorption on 15 h^-1 cMpc scales (CoSLAs, defined by delta_tau_eff > 3.5 after removing damped systems) traces z = 0 M >= 10^14 Msun protoclusters. It finds a weak delta_m - delta_tau_eff correlation with large scatter and shows that most sightlines through protoclusters are indistinguishable from the low-column-density Ly-alpha forest. In Sherwood, 12 of 22 post-DLA CoSLAs are associated with protoclusters; a threshold delta_tau_eff > 4.1 produces zero contamination but only 17% completeness, and fewer than 0.1% of sightlines through protoclusters meet the CoSLA threshold. The paper concludes that CoSLAs are incomplete and non-unique protocluster tracers and may instead trace filamentary structure.","tokens_in":26589,"tokens_out":7982,"duration_ms":85444,"significance":"If the quantitative results hold, this is an important correction to the Cai et al. (2016) picture and is directly relevant to upcoming DESI and WEAVE-QSO absorption surveys. Strengths of the paper include the use of three independent simulation codes with different feedback schemes, explicit tracking of z = 0 haloes back to z ~ 2.4, a CDDF validation, and resolution/box-size convergence tests. The paper is also careful to caveat its small box sizes and the absence of 10^15 Msun progenitors. The main quantitative claims, however, rest on small samples (22, 32 and 36 CoSLAs) and on the simulated line-of-sight clustering of LLSs/SLLSs, which is not validated; the central inference about non-uniqueness is therefore plausible but not yet fully pinned down.","major_comments":[{"comment":"The central non-uniqueness result relies on the simulated population of LLSs and high-column Ly-alpha forest absorbers that remain after damped systems are removed, because these systems produce the CoSLA tail after DLA/SLLS removal. The only absorber validation, in Section 3 and Figure 1, is a one-point CDDF comparison with observational points displayed at N_HI < 10^17 cm^-2 and N_HI > 10^20 cm^-2, but not in the 10^17-10^20 cm^-2 interval that dominates the post-DLA CoSLA population; the authors themselves call the high-NHI agreement 'possibly fortuitous.' Because a 15 h^-1 cMpc segment is coherently strong only if these absorbers are aligned along the sightline, and no two-point or pair-sightline clustering test is provided, the 55/45 protocluster/field split in Figure 15 is not secure. I request either a quantitative clustering comparison (e.g., LLS/SLLS pair separations or quasar-pair statistics) or an explicit sensitivity test that varies LLS abundance and clustering while preserving the CDDF.","section":"Section 3; Section 5.2, Figures 7 and 15"},{"comment":"The resolution convergence test shows that the incidence of CoSLAs is not converged: N_CoSLA/V decreases from 4.9 x 10^-5 (80-512) to 3.1 x 10^-5 (80-1024) to 1.4 x 10^-5 (80-2048), and these tests use the QLy-alpha variant rather than the full-physics fiducial runs. Because the headline purity statement is based on only 22 CoSLAs in Sherwood, and the three full-physics simulations yield 22, 32 and 36 CoSLAs, the bootstrap uncertainties quoted in Section 5.2 understate the systematic uncertainty. The <0.1% completeness claim is more robust because it uses thousands of protocluster sightlines, but the quantitative 'approximately half' purity claim is not pinned down by the current tests.","section":"Section 5.3, Table 2, Figures 10-11"},{"comment":"The zero-contamination threshold delta_tau_eff > 4.1 is selected post hoc from the same Sherwood sightlines used to evaluate it, and the evaluation sample contains only 22 CoSLAs. With such small numbers, the drop to zero contamination at delta_tau_eff = 4.1 may be a small-sample fluctuation rather than a stable prediction, and the threshold is explicitly model-dependent. I recommend presenting this as a proof-of-concept that a clean threshold exists in the simulations, with bootstrap or split-sample uncertainties, rather than as a predicted observable threshold.","section":"Section 7, Figure 16"}],"minor_comments":[{"comment":"The conclusions contain duplicated 'per cent per cent' and describe the CoSLA threshold as '4.5 times the average' while the body defines it as delta_tau_eff > 3.5; both should be cleaned up for consistency.","section":"Section 8"},{"comment":"The association criterion 'at least one third of the segment passes within R95' is plausible but arbitrary; a brief sensitivity test (e.g., varying the required fraction between 1/4 and 1/2) would make the protocluster/field classification less binary.","section":"Section 4.1, footnote 8"},{"comment":"Please specify the numerical value(s) of the optical-depth rescaling constant applied to each simulation, since the delta_tau_eff thresholds depend sensitively on the calibration to the Becker et al. (2013) tau_eff measurement.","section":"Section 2.4"},{"comment":"The caption should state explicitly that no observational CDDF data are displayed in the LLS/SLLS range 10^17-10^20 cm^-2; this is currently only clear from the body text.","section":"Figure 1"}],"recommendation":"major_revision","confidential_remarks":"The paper is within MNRAS scope and I have no concerns about prior publication or simulation provenance. The requested revisions concern strengthening or qualifying the quantitative claims about CoSLA purity and the clean threshold, not redoing the entire analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper overturns the current picture of CoSLAs as protocluster tracers. Using three hydro simulations, the authors show that strong coherent Ly-alpha absorption on 15 cMpc scales arises mostly from line-of-sight alignments of LLS/SLLS systems, not from overdense protocluster volumes. Previous work by Cai et al. (2016) with a collisionless simulation concluded almost all CoSLAs trace overdense regions; this paper shows that conclusion depended on missing LLSs and on unresolved underdense gas.\n\nThe genuine strengths: the CDDF validation, the comparison across Sherwood, EAGLE, and Illustris with different codes and feedback schemes, the resolution and box-size tests, and a clear statement of the key limitations. The authors honestly note that high-column-density agreement may be fortuitous and that the completeness estimate is likely a lower limit. The central qualitative claim — that CoSLAs do not uniquely trace protoclusters and that CoSLA-selected samples are highly incomplete — is robust. The <0.1% completeness number is the most secure, since it comes from a large denominator of protocluster sightlines.\n\nThe soft spots are real but not fatal. The CDDF validation in Figure 1 has no observational data in the 10^17–10^20 cm^-2 range, which is exactly the LLS/SLLS population that dominates CoSLAs after DLA removal. More importantly, the one-point CDDF says nothing about the line-of-sight clustering of LLSs/SLLSs that makes a 15 cMpc segment coherently strong. The 55% purity figure rests on that unvalidated clustering and on only 22 CoSLAs in Sherwood. The CoSLA count varies by a factor ~1.6 across simulations, and the resolution tests show the incidence is not converged. So the \"about half\" number is illustrative, not pinned down. The delta_tau_eff > 4.1 clean-sample threshold is post hoc from the same Sherwood data; it should be labelled as such.\n\nThe citation pattern is fair: the disagreement with C16 is explicit and tested, not a strawman. The paper deserves a serious referee and, after minor revision, publication. I would ask the authors to add uncertainty estimates on the headline fractions, explicitly frame the clean threshold as an illustrative fit, and address the missing LLS/SLLS clustering validation.\n\nWho should read it: anyone interpreting CoSLA searches in BOSS, eBOSS, DESI, or WEAVE-QSO, and anyone modelling the IGM-galaxy connection. I'd bring it to reading group.\n\nRecommendation: engage with it; accept for review.","headline":"A careful hydro simulation study that credibly overturns the prior claim that CoSLAs are good protocluster tracers, though the quantitative fractions need more validation.","tokens_in":27105,"tokens_out":3543,"would_cite":true,"duration_ms":38317,"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":"Coherent strong Lyman-alpha absorption is not a reliable protocluster tracer: even with perfect cleaning, roughly half of such sightlines miss the clusters, and fewer than 0.1 per cent of protocluster sightlines pass the standard threshold.","keywords":["protoclusters","Lyman-alpha forest","intergalactic medium","quasar absorption lines","hydrodynamic simulations","damped Lyman-alpha absorbers","coherently strong Lyman-alpha absorption","galaxy cluster formation"],"falsifier":"If observed CoSLA-selected fields at $z\\sim2.3$ are followed up and the fraction that turns out to be associated with massive $z=0$ cluster descendants differs substantially from the roughly 55 per cent predicted here, the simulated clustering of high-column-density absorbers along sightlines would be ruled out. A more direct test is to measure the line-of-sight clustering of LLSs, SLLSs, and DLAs at $z\\simeq2.4$ toward quasar pairs and compare it with the simulations' predictions.","tokens_in":26003,"feed_emoji":"🔭","tokens_out":8965,"duration_ms":85507,"temperature":0.7,"pith_summary":"This paper tests a proposed shortcut for finding protoclusters at $z\\sim2.4$: look for coherent stretches of strong Lyman-$\\alpha$ absorption in background quasar spectra, dubbed CoSLAs. Using three hydrodynamic simulations that follow gas and galaxies down to $z=0$, the authors check whether these absorption segments actually pass through the regions that will become $M_{\\rm z=0}\\gtrsim10^{14}\\,M_\\odot$ clusters. They find only a weak, highly scattered correlation between mass overdensity $\\delta_{\\rm m}$ and effective optical depth contrast $\\delta_{\\tau_{\\rm eff}}$ on $15\\,h^{-1}\\,\\rm cMpc$ scales. Even assuming perfect removal of damped Ly-$\\alpha$ systems, only about half of CoSLA sightlines trace protoclusters, and less than $0.1$ per cent of sightlines through protoclusters meet the standard $\\delta_{\\tau_{\\rm eff}}>3.5$ threshold. The same absorption signal, however, may be a useful geometric tracer of gas filaments aligned along the line of sight.","feed_headline":"Strong Lyman-alpha absorption misses most protoclusters","feed_subtitle":"Hydro simulations show the absorption signal finds clusters half the time and overlooks over 99.9 percent.","key_machinery":"The machinery is a set of mock Lyman-$\\alpha$ spectra generated from three independently calibrated cosmological hydrodynamic simulations, with protoclusters defined by tracing $z=0$ friends-of-friends haloes of mass $M_{\\rm z=0}\\ge10^{14}\\,M_\\odot$ back to $z\\simeq2.4$. Each sightline is divided into $15\\,h^{-1}\\,\\rm cMpc$ segments; the mass overdensity $\\delta_{\\rm m}$ of the surrounding cubic volume is then compared with the effective optical depth contrast $\\delta_{\\tau_{\\rm eff}}$ of the spectrum. The CoSLA criterion ($\\delta_{\\tau_{\\rm eff}}>3.5$ after removing segments containing SLLSs or DLAs) is the selection rule under test, and completeness and contamination are computed as functions of the $\\delta_{\\tau_{\\rm eff}}$ threshold.","core_discovery":"The central claim is that coherently strong intergalactic Lyman-$\\alpha$ absorption (CoSLAs, defined as $15\\,h^{-1}\\,\\rm cMpc$ spectral segments with $\\delta_{\\tau_{\\rm eff}}>3.5$ after removing damped systems) is neither a unique nor a complete signpost for $M_{\\rm z=0}\\ge10^{14}\\,M_\\odot$ protoclusters at $z\\simeq2.4$. In the primary simulation, CoSLAs are rare, cover a broad $\\delta_{\\rm m}$ range including underdense volumes, and $55$ per cent are associated with a protocluster. Conversely, $84$ per cent of sightlines through protoclusters contain only low-column-density Ly-$\\alpha$ forest absorption with $\\delta_{\\tau_{\\rm eff}}<1$, and fewer than $0.1$ per cent of protocluster sightlines reach the CoSLA threshold once damped systems are excluded. Raising the threshold to $\\delta_{\\tau_{\\rm eff}}>4.1$ yields a completely clean protocluster sample, but completeness drops to about $17$ per cent; the authors note this is likely a lower limit because their volumes contain no $10^{15}\\,M_\\odot$ progenitors.","pith_inferences":["If CoSLAs preferentially select line-of-sight-aligned filamentary structures, their galaxy associations should differ from randomly chosen overdense regions: one would expect coherent absorption to correlate with elongated galaxy distributions and velocity caustics, a testable prediction for tomographic surveys.","Real surveys cannot achieve perfect damped-system removal, so in practice contamination will be higher than the paper's roughly 45 per cent; the technique likely needs ancillary galaxy or metal-line information to be usable.","The strong resolution dependence of CoSLA incidence implies that forecasts from older low-resolution dark-matter-only simulations need revisiting; observed CoSLA counts could be used to calibrate small-scale intergalactic-medium opacity models.","The clean threshold at $\\delta_{\\tau_{\\rm eff}}>4.1$ is model-dependent; a robust estimator should be calibrated in each observed data set using the distribution of $\\delta_{\\tau_{\\rm eff}}$ itself."],"forward_implications":["Surveys that use CoSLAs to find protoclusters at $z\\sim2$--$3$ will be heavily incomplete even in the ideal case of perfect damped-system removal; their selection functions must be calibrated before statistical use.","Raising the threshold to $\\delta_{\\tau_{\\rm eff}}\\sim4.1$ after cleaning damped systems gives a pure but low-completeness protocluster sample (roughly 17 per cent of clusters), useful for studying individual extreme systems rather than populations.","Coherent strong absorption is a better geometric tracer of matter aligned along the line of sight than of bound overdensity; it may be used to map intergalactic filaments at $z>2$.","Correctly modelling both high-column-density absorbers and the opacity of underdense gas is necessary to predict the incidence of coherent absorption; low-resolution simulations overpredict CoSLA counts.","Rare, very massive protoclusters ($M_{\\rm z=0}>10^{15}\\,M_\\odot$) are absent from these volumes, so the relation could be more favourable for the most extreme structures; the quoted completeness is likely a lower limit."],"supporting_citations":[{"why":"Defines the CoSLA concept and the $\\delta_{\\tau_{\\rm eff}}>3.5$ threshold that this paper tests against hydrodynamical simulations.","marker":"Cai et al. 2016"},{"why":"Provides the probability that a $15\\,\\rm cMpc$ volume with $\\delta_{\\rm m}\\ge0.8$ collapses to a $z=0$ cluster, linking overdensity to protoclusters.","marker":"Chiang et al. 2013"},{"why":"Supplies the protocluster size scale and descendant-tracing method used to identify $M_{\\rm z=0}\\ge10^{14}\\,M_\\odot$ progenitors.","marker":"Muldrew et al. 2015"},{"why":"Provides the observed mean effective optical depth used to rescale all mock spectra.","marker":"Becker et al. 2013"},{"why":"Supplies the low-column-density Lyman-$\\alpha$ forest column density distribution used to validate the simulations.","marker":"Kim et al. 2013"},{"why":"Supplies the observed incidence of high-column damped absorbers at $z\\simeq2.5$ used to validate the simulations.","marker":"Noterdaeme et al. 2012"},{"why":"Describes the primary hydrodynamic simulation and its resolution suite used for the main analysis.","marker":"Bolton et al. 2017"},{"why":"Describes an independent hydrodynamic simulation used as a cross-check.","marker":"Schaye et al. 2015"},{"why":"Describes another independent hydrodynamic simulation used as a cross-check.","marker":"Vogelsberger et al. 2014"},{"why":"Provides the self-shielding prescription applied to two of the simulations to capture Lyman-limit systems and super Lyman-limit systems.","marker":"Rahmati et al. 2013"}],"fun_headline_variants":["Lyman-alpha absorption rarely flags protoclusters","Protoclusters hide from strong Lyman-alpha absorption","Most protocluster sightlines show weak Lyman-alpha absorption","Strong Lyman-alpha absorption misses 99.9% of protoclusters","Coherent Lyman-alpha absorption is a poor protocluster tracer"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The results assume the simulated incidence and line-of-sight clustering of high-column-density H I absorbers (LLSs, SLLSs, DLAs) at $z\\sim2.4$ matches reality, since these systems dominate the high-$\\delta_{\\tau_{\\rm eff}}$ tail; only the one-point column-density distribution is checked against observations, and the agreement at the highest columns is admitted to be possibly fortuitous.","fun_headline_variants_meta":{"raw":{"variants":["Lyman-alpha absorption rarely flags protoclusters","Protoclusters hide from strong Lyman-alpha absorption","Most protocluster sightlines show weak Lyman-alpha absorption","Strong Lyman-alpha absorption misses 99.9% of protoclusters","Coherent Lyman-alpha absorption is a poor protocluster tracer"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000274,"raw_usage":{"total_tokens":1777,"prompt_tokens":1223,"completion_tokens":554,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":839,"completion_tokens_details":{"reasoning_tokens":469}},"tokens_in":839,"tokens_out":554,"duration_ms":4996,"temperature":1.0,"reasoning_tokens":469,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:47:42.219908+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"If observed CoSLA-selected fields at $z\\sim2.3$ are followed up and the fraction that turns out to be associated with massive $z=0$ cluster descendants differs substantially from the roughly 55 per cent predicted here, the simulated clustering of high-column-density absorbers along sightlines would be ruled out. A more direct test is to measure the line-of-sight clustering of LLSs, SLLSs, and DLAs at $z\\simeq2.4$ toward quasar pairs and compare it with the simulations' predictions.","supporting_citations":[],"review_version":1}