{"id":"3d229f1e-7bd6-4d28-a48b-8d0ae0eccfa9","arxiv_id":"2509.18271","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":15,"one_line_summary":"The XQ100 Lyman-alpha forest power spectrum, modeled with the PRIYA simulations, gives cosmological constraints consistent with eBOSS and Planck and constrains the IGM thermal history without external temperature data, while KODIAQ-SQUAD yields a biased amplitude due to Lyman-limit system contaminat","lead":"This paper uses two high-resolution quasar catalogs to test how the Lyman-alpha forest, the absorption imprint of gas between quasars and us, constrains cosmology and gas temperature. The smaller catalog agrees with larger surveys and can measure thermal history alone; the larger catalog is shown to be biased by dense hydrogen absorbers.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Resolution-correction ratio is assumed cosmology-independent in §2.1; if it varies across PRIYA's thermal/cosmological parameter space, XQ100's high-k thermal and (A_P, n_P) posteriors shift.","rationale":"The reader's weakest_assumption correctly identifies the small-scale resolution correction as the most load-bearing point. The paper's novelty and primary consistency claim depend on the XQ100 P1D at k > 0.02 s/km, where the cosmology-independent correction is applied. This is also where thermal history information is concentrated (§5.1), so an unvalidated transfer of the resolution ratio from one thermal/cosmological state to the entire PRIYA parameter space could bias the very parameters the paper claims to constrain. The paper presents useful supporting evidence: LOO-CV validation (Figure 1), simulated-data recovery (Figure 2), and a CDDF comparison (Figure 13). These checks strengthen the analysis but do not directly test the parameter dependence of the resolution ratio. The concrete test proposed would settle the transfer assumption. If the ratio is stable across parameter extremes, the concern is resolved and the central XQ100 claim stands; if not, the quoted posteriors are overconfident and the CONDITIONAL status is warranted. Since the reader's verdict is already CONDITIONAL and this concern is consistent with that assessment, no change in verdict is recommended.","tokens_in":31863,"tokens_out":7732,"duration_ms":71885,"concrete_test":"Run the 15 Mpc/h resolution-test boxes at 2x512^3 and 2x768^3 for at least two extreme PRIYA parameter corners within Table 2 (e.g., low/high α_q and zHeII_i/f, and low/high A_P). Compute the ratio P1D_768/P1D_512 at k = 0.02-0.06 s/km at each corner and compare across corners. If the ratio varies by more than the quoted ~1-3% convergence error, or by a non-negligible fraction of the XQ100/KODIAQ-SQUAD error bars at those bins, the cosmology-independent correction assumption fails and the analysis should be rerun with a parameter-dependent correction or with the impacted high-k bins removed/downweighted. A complementary check is to refit XQ100 with the resolution correction disabled and with its uncertainty added to the covariance; if (A_P, n_P, T0) posteriors move by <0.5σ, the concern is not currently load-bearing.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central XQ100 result--that the high-resolution P1D alone constrains IGM thermal history and yields (A_P, n_P) consistent with eBOSS/Planck--rests on the small-scale bins k = 0.02-0.06 s/km. In §2.1, the emulator applies a fixed, cosmology-independent resolution correction: the P1D in this k-range is multiplied by the ratio of the two highest-resolution simulations from Ref. [63] (2x768^3 vs 2x512^3 particles in a 15 Mpc/h box). This ratio is measured at effectively one thermodynamic/cosmological state, then applied to every parameter combination in the PRIYA emulator. Priors in Table 2 include large variations in HeII reionization parameters (α_q, zHeII_i, zHeII_f) that change the gas temperature and small-scale smoothing, and A_P variations that change small-scale clustering; each can plausibly alter the fractional resolution error at high k. If the correction ratio is not universal, the emulator's high-k P1D is biased in a parameter-dependent way, and because A_P is degenerate with α_LLS and with thermal parameters, the XQ100 posteriors could be systematically shifted. The correction uncertainty is also not propagated into the covariance matrix (Eq. 3.1) or the likelihood. This is not an accusation that the correction is wrong; it is an unvalidated transfer assumption at exactly the scales that carry XQ100's thermal information.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a cosmological analysis of high-resolution Lyman-alpha forest 1D flux power spectra (P1D) from XQ100 and KODIAQ-SQUAD, using the PRIYA multi-fidelity simulation emulator extended to k = 0.003–0.06 s/km. The inference includes cosmological parameters (A_P, n_P, h, Omega_M h^2), hydrogen and helium reionization parameters, mean optical depth parameters, and a four-parameter HCD/LLS contamination template. The authors report that the XQ100 P1D alone, without external IGM temperature data, yields (A_P, n_P) constraints consistent with eBOSS DR14 and Planck, and provides thermal-history information comparable to the eBOSS+T0 baseline. In contrast, KODIAQ-SQUAD favors a high A_P and a high LLS abundance, which the authors attribute to selection bias toward dense absorbers; restricting the analysis to z = 3.4–4.2 and k < 0.045 s/km brings it into agreement with XQ100/eBOSS. They conclude that thermal and cosmological information in the P1D are largely scale-separated and that high-resolution P1D data can serve as thermal nuisance constraints in future DESI joint fits.","tokens_in":32278,"tokens_out":10199,"duration_ms":85597,"significance":"If the central claims hold, this is a valuable demonstration that high-resolution P1D can simultaneously constrain the IGM thermal history and remain consistent with large-volume survey cosmology. The paper's strengths are its forward-modeling approach, the explicit validation of the likelihood on simulated data, the inclusion of a detailed HCD template, and multiple robustness checks (restricted k/z ranges, per-redshift tau_0, per-redshift alpha_LLS, MAP diagnostics). The paper is also appropriately cautious about the KODIAQ-SQUAD selection bias and about the interpretation of tau_0 as an effective nuisance parameter. The main technical risk is the unvalidated cosmology-independent resolution correction applied at exactly the scales that carry XQ100's thermal information; this is a load-bearing assumption that needs further support or sensitivity testing. Overall the manuscript is a solid contribution to the Lyman-alpha cosmology literature, conditional on resolving that point.","major_comments":[{"comment":"The resolution correction is assumed to be cosmology-independent. The emulator multiplies P1D for k = 0.02–0.06 s/km by the ratio of the two highest-resolution simulations from Ref. [63] (2×768^3 vs 2×512^3 in a 15 Mpc/h box), a ratio measured at effectively one thermodynamic/cosmological state. It is then applied to the entire prior range of Table 2, which includes large variations in HeII reionization parameters (alpha_q, z_HeII^i, z_HeII^f) and A_P. No test is shown that this ratio is universal, and its uncertainty is not propagated into the covariance of Eq. (3.1). Since the XQ100 thermal-history claim relies on k > 0.045 s/km (§5.1, Figs. 5 and 7), a parameter-dependent correction would bias the high-k bins that carry the thermal signal. I recommend either recomputing the resolution ratio at several corner points of the prior, or rerunning the inference with a deliberately varied co","section":"§2.1"},{"comment":"There is an inconsistency in the stated k-range of the emulator vs the data cuts. Section 2.1 says the emulator predicts P1D for k = 0.003–0.06 s/km, but Section 3.5 and Table 3 use XQ100 data out to k = 0.064–0.065 s/km and KODIAQ-SQUAD out to k = 0.065 s/km. If the emulator is trained and LOO-CV-validated only up to k = 0.06, then the bins at 0.06–0.065 are outside the surrogate's domain. Please clarify whether the emulator actually covers k = 0.065, or truncate the data at k = 0.06 for consistency.","section":"§2.1 vs §3.5"},{"comment":"The covariance matrix in Eq. (3.1) includes the data covariance and sample variance sigma_CV, but does not include the emulator interpolation error quoted in §2.1 (LOO-CV ~1–2%) or the resolution-correction uncertainty. While the emulator error is smaller than the diagonal statistical errors of XQ100 (~20%) and KODIAQ-SQUAD (~10%), a 1–2% systematic can bias the mean of the model when summed over many bins, and the resolution correction is an unmodeled systematic at high k. I recommend adding an emulator-error term to Eq. (3.1) or demonstrating via a sensitivity test that the posteriors are unaffected by this omission.","section":"§3.1–3.2"}],"minor_comments":[{"comment":"The equation for tau_eff appears to have a redundant or misplaced tau_Kim(z) factor: the second line contains both the ratio term and a trailing tau_Kim(z), making the expression dimensionally confusing. Please rewrite for clarity.","section":"§2.2, Eq. (2.2)"},{"comment":"The text in §4.1 states that LLSs have minimal damping wings, while the caption of Figure 5 attributes the large-scale boost from LLSs to 'damping wings of their absorption profiles.' These statements should be reconciled.","section":"§4.1 and Fig. 5 caption"},{"comment":"The abstract says the data probe 'down to k ~ 6 h Mpc^-1', but the analysis is primarily in s/km units. Consider stating the k-range in s/km consistently to avoid confusion.","section":"Abstract and §3.5"},{"comment":"Several entries show double uncertainties (e.g., tau_0, n_P, A_P) without a clear definition of the two error levels. Please specify what the inner and outer errors represent (e.g., 68% and 95% credible intervals).","section":"Table 3"},{"comment":"The paper notes that XQ100 cannot robustly infer the peak IGM temperature at z ~ 3 because its redshift range starts at z = 3.4. This is an important limitation and should be stated earlier in the paper, ideally in the abstract or introduction.","section":"§5.4"}],"recommendation":"major_revision","confidential_remarks":"The main technical risk is the unvalidated resolution-correction transfer in §2.1; the authors should either validate it or show insensitivity. Much of the modeling infrastructure (PRIYA suite, HCD template) comes from the same group; an external validation of the emulator at these high-k bins would increase confidence. The KODIAQ-SQUAD selection-bias interpretation is reasonable but inferred indirectly through the HCD template; the paper is appropriately cautious, which is a strength. If the resolution-correction issue is convincingly resolved, the paper would be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main thing: this is a solid, careful paper, and the XQ100 result looks right. The more interesting claim is that KODIAQ-SQUAD's high-A_P posterior comes from a redshift-dependent LLS selection bias, not new cosmology. I buy that, largely because the bias grows with redshift, disappears when you restrict to k<0.045 s/km and z=3.4-4.2, and the CDDF comparison argues PRIYA's LLS population is not off by a factor of three.\n\nWhat's actually new: first PRIYA-based cosmological and thermal inference from the KODIAQ-SQUAD P1D; an emulator extension to k=0.06 s/km; and the demonstration that XQ100 alone constrains the IGM thermal history without external T0 while giving (A_P, n_P) consistent with eBOSS and Planck. The pipeline is well checked: simulated-data validation, a sample-variance term in the covariance, per-redshift tau0 and alpha_LLS runs, and restricted k/z cuts. Those are the right robustness checks.\n\nThe soft spot is the resolution correction in §2.1. They multiply the emulated P1D at k=0.02-0.06 s/km by a fixed ratio measured in 15 Mpc/h boxes (2x512^3 vs 2x768^3) and assume it is cosmology-independent. That is plausible but not tested. A_P and the HeII reionization parameters change small-scale clustering and gas temperature, so a parameter-dependent resolution error would shift exactly the high-k bins that carry XQ100's thermal information. They also do not propagate any uncertainty on the correction. The stated resolution convergence error (1-3% at these k) is smaller than the 10-20% statistical errors, so this is not fatal, but it should be checked. A simple sensitivity test—rerun a few emulator points with and without the correction, or compute the ratio across a few thermal/cosmological states—would settle it.\n\nTwo smaller gaps: no code, chains, or analysis data are released, which limits reproducibility; and the HCD template from Illustris has unquantified uncertainty, although the CDDF comparison and per-z alpha_LLS partially address that. I don't think the stress-test concern invalidates the central XQ100-eBOSS consistency; the restricted-range tests already show that result is robust. But it is the right question to ask.\n\nThis paper is for Ly-a forest cosmologists, DESI/eBOSS analysts, and people building P1D emulators. It deserves a serious referee. Send it to peer review and ask for a resolution-correction robustness test and release of the chains/data.","headline":"Solid, useful Lyman-alpha P1D paper: the XQ100-eBOSS consistency holds and the KODIAQ-SQUAD LLS-selection-bias interpretation is the real result, but the cosmology-independent resolution correction is a genuine unquantified systematic that needs a sensitivity test.","tokens_in":32811,"tokens_out":5891,"would_cite":true,"duration_ms":49747,"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 XQ100 quasar spectra, modeled with the PRIYA emulator, constrain the primordial power spectrum amplitude and tilt consistently with eBOSS and Planck—without any external IGM temperature data.","keywords":["Lyman-alpha forest","flux power spectrum","primordial power spectrum","intergalactic medium","helium reionization","Lyman-limit systems","hydrodynamical simulations","PRIYA emulator"],"falsifier":"Run the two highest-resolution small-box simulations (2x512³ and 2x768³ in 15 Mpc/h) at the extremes of the emulator's A_P and thermal parameter ranges and recompute the ratio used to correct k=0.02-0.06 s/km; if the ratio changes by more than the stated ~1% at any varied parameter, the cosmology-independent assumption fails and the XQ100 posteriors would need revision.","tokens_in":31750,"feed_emoji":"🔭","tokens_out":4435,"duration_ms":38383,"temperature":0.7,"pith_summary":"This paper aims to show that high-resolution Lyman-alpha forest spectra from the XQ100 survey, interpreted through the PRIYA simulation emulator, can simultaneously constrain the primordial power spectrum of density fluctuations and the thermal history of the intergalactic medium. The inferred amplitude and tilt of the primordial spectrum at the Lyman-alpha pivot scale agree with the eBOSS survey and Planck CMB measurements, though with wider uncertainties. Crucially, this agreement is reached without using external measurements of the IGM temperature, implying that the small-scale flux power spectrum carries its own thermal information. The companion KODIAQ-SQUAD dataset, by contrast, yields a biased high amplitude that the paper traces to a selection preference for high-column-density absorbers (Lyman-limit systems), which mimic the spectral effect of a larger primordial amplitude. The paper therefore identifies Lyman-limit system contamination as the main obstacle to cosmological inference on small scales.","feed_headline":"XQ100 quasar spectra alone match Planck's primordial power spectrum","feed_subtitle":"A 100-sightline sample constrains cosmology and IGM heating in one fit; a larger archive is skewed by absorber bias.","key_machinery":"The PRIYA multi-fidelity emulator—a machine-learning model interpolating between low- and high-resolution hydrodynamical Lyman-alpha simulations to predict the 1D flux power spectrum at few-percent accuracy from k=0.003 to 0.06 s/km—carries the argument, together with a four-parameter template for high-column-density absorbers (damped Lyman-alpha systems, sub-DLAs, and Lyman-limit systems). The resolution correction that extends the emulator to small scales is a cosmology-independent ratio applied to k=0.02-0.06 s/km bins.","core_discovery":"The central claim is that the one-dimensional flux power spectrum (P1D) of the Lyman-alpha forest measured from high-resolution quasar spectra, when modeled with an emulator built on the PRIYA simulation suite, contains enough information to jointly constrain the primordial power spectrum parameters (A_P, n_P) and the IGM thermal history set by inhomogeneous helium reionization. Applying the emulator to the XQ100 P1D yields (A_P, n_P) posteriors consistent with the eBOSS DR14 analysis and Planck, without any external temperature prior. The same analysis applied to the KODIAQ-SQUAD P1D returns a high A_P that the paper attributes to a redshift-dependent excess of Lyman-limit systems in that d","pith_inferences":["If the scale separation holds, a dedicated high-resolution survey at z<3.2 with HI-blind selection could sharpen thermal constraints without requiring exotic cosmology—a testable prediction of the paper's picture.","The observed LLS excess in KODIAQ-SQUAD might be verified by stacking quasar spectra or cross-correlating with LLS catalogs to measure the selection function directly, something the paper leaves for future work.","The cosmology-independent resolution correction could be validated by running the highest-resolution small-box simulations at varied (A_P, n_P) or thermal parameters; if the ratio changes, the reported small-scale constraints would shift.","The paper's suggestion that τ0 absorbs LLS saturation implies that mean-flux measurements from high-resolution surveys should be interpreted with care when used to calibrate the UV background."],"forward_implications":["XQ100-type high-resolution P1D can supply thermal-history constraints in joint fits with eBOSS or DESI, removing the need for external IGM temperature measurements.","Cosmological parameters from the small-scale forest are consistent with Planck and large-volume surveys, supporting the standard cosmological model at scales down to k~6 h/Mpc⁻¹.","Any future small-scale Lyman-alpha dark matter analysis must marginalize over Lyman-limit system contamination, since LLSs mimic a suppression of the linear power amplitude.","KODIAQ-SQUAD's biased selection function means its small-scale P1D should not be used alone for cosmology above k=0.045 s/km unless the selection bias is corrected.","The scale separation (thermal at high k, cosmology at low k) means joint fits can be designed to isolate the two information channels.","The paper's finding that imposing (A_P, n_P) priors barely changes the reduced chi-squared supports the claim that cosmological and thermal parameters are largely scale-separate."],"fun_headline_variants":["XQ100 alone matches Planck without external temperature priors","Small-scale Lyman-alpha forest from XQ100 rivals eBOSS cosmology","Quasar sample XQ100 forces Planck-consistent cosmology alone","100 sightlines beat eBOSS thermal history constraints unaided","Lyman-alpha P1D from XQ100 yields Planck-like parameters"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The small-scale emulator assumes a single, cosmology-independent resolution correction—measured in a 15 Mpc/h box—applies to the full 120 Mpc/h simulation suite at every parameter combination; if that ratio depends on cosmology or volume, the inferred amplitude, tilt, and thermal parameters are shifted.","fun_headline_variants_meta":{"raw":{"variants":["XQ100 alone matches Planck without external temperature priors","Small-scale Lyman-alpha forest from XQ100 rivals eBOSS cosmology","Quasar sample XQ100 forces Planck-consistent cosmology alone","100 sightlines beat eBOSS thermal history constraints unaided","Lyman-alpha P1D from XQ100 yields Planck-like parameters"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000679,"raw_usage":{"total_tokens":3011,"prompt_tokens":918,"completion_tokens":2093,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":662,"completion_tokens_details":{"reasoning_tokens":2005}},"tokens_in":662,"tokens_out":2093,"duration_ms":15599,"temperature":1.0,"reasoning_tokens":2005,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T15:44:55.114738+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the two highest-resolution small-box simulations (2x512³ and 2x768³ in 15 Mpc/h) at the extremes of the emulator's A_P and thermal parameter ranges and recompute the ratio used to correct k=0.02-0.06 s/km; if the ratio changes by more than the stated ~1% at any varied parameter, the cosmology-independent assumption fails and the XQ100 posteriors would need revision.","supporting_citations":[],"review_version":1}