{"id":"a3386d8c-b706-498b-8cbd-765e634ec8b0","arxiv_id":"2504.20299","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A DESI DR2 catalog of 101,487 CIV absorbers shows the absorber path density rising 2 to 5 fold and the CIV mass density rising 3.8 fold from z about 4.5 to z about 1.4.","lead":"Using 300,637 DESI quasar spectra, the authors built the largest catalog of triply ionized carbon (CIV) absorbers to date, with 101,487 systems in the redshift range 1.4 to 4.5. They find the cosmic density of CIV gas grows about 3.8 times from early to late cosmic time, tracking the rise of star formation.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Completeness from idealized double-Gaussian mocks is not validated against real absorber profiles; a profile-dependent mismatch could bias the 1/C_i corrections and the 3.8x Ω_CIV / 2-5x dN/dX evolution.","rationale":"I read the paper as a careful, large-scale observational study whose central claim is the measured evolution of CIV absorber incidence and mass density between z~4.5 and z~1.4. The quantitative results depend on the completeness correction in a direct way: every statistic in Sec 5 uses per-system weights wi = 1/Ci(EW,z), and C(EW,z) is built from Monte Carlo injections of idealized double-Gaussian profiles. This is the least externally validated ingredient in the analysis. The internal checks provided (flat stacked residual, ~95% purity estimate, consistency with Cooksey et al. re-measured values) are valuable but none of them measures the detection efficiency for real, complex absorber profiles. A systematic difference between mock and real profiles, especially if it varies with redshift, could bias the 3.8x Ω_CIV and 2-5x dN/dX factors beyond the quoted statistical errors. I therefore agree with the reader's weakest_assumption. The concrete high-resolution degradation test proposed above would settle whether the concern lands: if real-profile completeness matches C(EW,z), the headline numbers are secure; if not, the corrected statistics and evolution factors need revision. Secondary issues raised by the reader (abstract wording, DR2 catalog availability, saturation lower limits) are real but either acknowledged in the text or do not change the conditional status of the paper. The verdict should remain CONDITIONAL: the paper is scientifically sound and likely correct in its qualitative conclusions, but the completeness validation and catalog release should be addressed before full acceptance.","tokens_in":29975,"tokens_out":9081,"duration_ms":103637,"concrete_test":"Take ~100-200 CIV absorbers at 1.4 < z < 4.5 from high-resolution spectra (VLT/UVES, Keck/HIRES, or X-shooter) with measured velocity profiles. Normalize each spectrum, degrade it to the DESI line-spread function (R~2000, 0.8 Å pixels), rescale to a range of S/N representative of the DESI sample (e.g., ⟨S/N⟩ ≈ 3, 6, 10), inject the degraded profiles into the same real DESI residual spectra used in Sec 4.4, and run qsoabsfind with identical settings. Recover the detection fraction as a function of EW1548 and z and compare it bin-by-bin with the mock-based C(EW,z) of Fig 7. If the two agree within ~10% for EW > 0.4 Å and show no redshift-dependent offset, the completeness assumption is supported; if not, the 1/Ci weights in Eqs 7, 9, 10 need revision or an additional profile-dependent completeness term.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline evolution (dN/dX 2-5x, Sec 5.2/Fig 9; Ω_CIV 3.8x, Sec 5.3/Eq 10/Fig 10) is obtained by weighting every detected system by wi = 1/Ci(EW,z) in Eqs 7, 9, and 10. C(EW,z) is derived in Sec 4.4 from Monte Carlo injections of synthetic doublets into real DESI residual spectra. The mocks are single double-Gaussian profiles: EW drawn uniformly from 0.1-3 Å, doublet ratio from U[1,2], and amplitude from U(0,1), with the same line width for both components; the resulting width is not drawn from an observed distribution and can be far narrower or broader than real CIV kinematic structure. Real CIV absorbers at R~2000 are often blends of multiple velocity components, with non-Gaussian profiles, unresolved saturation, and occasional contamination by other transitions; at a fixed EW their peak optical depth and apparent width can differ substantially from a single Gaussian. Because the detection algorithm uses a matched double-Gaussian kernel plus adaptive SNR thresholds and doublet-ratio/EW cuts, detectability at fixed EW can depend on profile shape. C(EW,z) is binned only in EW and z, so the weight 1/Ci implicitly assumes the mock shape distribution is representative. The paper validates purity (~95%) and the composite spectrum, but neither test measures completeness for real profiles; a missed population of broad, multi-component systems at EW~0.4-0.6 Å would bias the corrected f(Wr), dN/dX, and Ω_CIV. If the shape/SNR mismatch evolves with redshift, the quoted 3.8x and 2-5x factors could be systematically in error beyond the small statistical errors.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an automated matched-kernel search for C IV doublets in 300,637 DESI DR2 quasar spectra, yielding 101,487 C IV absorber systems over 1.4 < z < 4.5. It derives a two-dimensional completeness function C(EW, z) from roughly 3 million injected mock doublets, estimates catalog purity near 95% from a median stacked spectrum, and constructs completeness-corrected statistics: f(Wr) is exponential with weak redshift evolution, dN/dX increases by a factor of about 2-5 from z ~ 4.5 to z ~ 1.4, and Omega_CIV increases from (0.82 +/- 0.05) x 10^-8 at z ~ 4.5 to (3.16 +/- 0.2) x 10^-8 at z ~ 1.4, a factor of about 3.8. The Omega_CIV values are presented as lower limits because the apparent optical depth method underestimates column densities for saturated systems. The paper validates its path-density and mass-density estimator by re-measuring Omega_CIV from the Cooksey et al. (2013) catalog under Planck and WMAP cosmologies in Appendix B, and interprets the observed trends as tracing cosmic star formation, He II photoheating, and lower limits on IGM metallicity.","tokens_in":30307,"tokens_out":5678,"duration_ms":60510,"significance":"If the central claims hold, this is the largest C IV absorber catalog to date and the most precise measurement of the redshift evolution of C IV path density and mass density over 1.4 < z < 4.5, with a public catalog and code as community resources. The paper has genuine strengths: Monte Carlo completeness based on roughly 3 million mock injections, a stacking-based purity estimate, an external validation that reproduces Cooksey et al. (2013) under two cosmologies, repeated statements that Omega_CIV values are lower limits, and reproducible code and data links. The 3.8x Omega_CIV and 2-5x dN/dX evolution are plausible and broadly consistent with prior work, but the quantitative reliability of the headline evolution depends on two assumptions that need direct testing: that idealized double-Gaussian mocks capture the completeness of real multi-component C IV absorbers, and that the lower-limit nature of AODM column densities does not introduce a redshift-dependent bias into the Omega_CIV evolution.","major_comments":[{"comment":"The completeness function C(EW, z) is calibrated solely on synthetic double-Gaussian profiles injected into real residual spectra, with EW drawn from a uniform distribution, doublet ratio from [1, 2], and a single common line width for both components. Real C IV absorbers at the DESI resolution are frequently blends of multiple velocity components with non-Gaussian profiles and possible unresolved saturation, so detectability at fixed EW and z can differ from the mock population. Because every corrected statistic uses weights w_i = 1/C_i in Eqs. (7), (9), and (10), a shape-dependent mismatch would propagate directly into f(Wr), dN/dX, and Omega_CIV. I request a direct validation: inject realistic multi-component or high-resolution observed C IV profiles convolved to the DESI resolution into the same residual spectra, recompute C(EW, z), and show that the corrected dN/dX and Omega_CIV evolution change by less than the quoted statistical errors; alternatively, compare completeness-corrected statistics against an independent high-resolution subsample.","section":"Section 4.4 and Eqs. (7), (9), (10)"},{"comment":"The Omega_CIV values are lower limits obtained with the apparent optical depth method, and the text states that about 88.7% of systems are partially saturated and that strong absorbers have doublet ratios near 1. If the partially saturated fraction or the effectiveness of the Savage-Sembach correction evolves with redshift, the factor of about 3.8 rise in the lower-limit Omega_CIV could be a lower-limit artifact rather than an increase in the true C IV mass density. The paper should present the partially saturated fraction and the logN(1550)/logN(1548) distribution as functions of redshift, and quantify how much a redshift-independent versus redshift-dependent saturation correction changes the 3.8x factor. The authors' repeated caution that these are lower limits is appropriate, but it does not by itself establish that the evolution factor is robust.","section":"Section 5.3 and Eq. (10)"},{"comment":"C(EW, z) is binned only in EW and z, even though the mock absorbers are inserted into residual spectra with a wide range of signal-to-noise ratios and the DESI quasar SNR distribution has strong redshift structure (Figure 1). If detection efficiency at fixed EW depends on quasar SNR, the completeness averaged over all quasars may impart a redshift-dependent bias to dN/dX and Omega_CIV. Please test whether C(EW, z) is flat in SNR or include SNR as an additional completeness dimension, and confirm that the dN/dX evolution is unchanged when the analysis is restricted to a narrow SNR slice.","section":"Sections 4.4 and 5.2"}],"minor_comments":[{"comment":"The text refers to the 'Dark Energy Survey Instrument (DESI)', but the survey is the Dark Energy Spectroscopic Instrument; this should be corrected.","section":"Section 2"},{"comment":"The notation 'CIV' appears without a space in several places, including Section 5.3 and Figure 10, while 'C IV' is used elsewhere; the notation should be standardized.","section":"Throughout the manuscript"},{"comment":"The statement that g(EW_r, z) assumes 'absorbers with any strength can be detected' is a strong approximation that is not literally compatible with the completeness function C(EW, z); this assumption should be stated more carefully because it enters the Delta X calculation used for all path-density estimates.","section":"Section 4.5, Eq. (4)"},{"comment":"The sixteen panels make the EW-axis labels and the fitted parameters difficult to read; a table of N0 and alpha values or a more compact summary figure would improve usability.","section":"Figure 8"},{"comment":"The metallicity lower limit depends on the adopted values A_C = 0.178 and f_CIV <= 0.35 from the literature; a brief sensitivity test of log(Z_IGM/Z_sun) to these choices would help readers judge the robustness of the metallicity evolution claim.","section":"Section 5.3, Eq. (11)"}],"recommendation":"major_revision","confidential_remarks":"To the editor: This is a strong manuscript with an important public catalog and careful validation of the measurement pipeline. The main correctness risk is the profile-dependence of the completeness function, since the headline Omega_CIV and dN/dX evolution are corrected with 1/C_i weights derived from idealized double-Gaussian mocks. A second risk is that the Omega_CIV evolution is based on lower limits for a sample that is mostly partially saturated, and the redshift dependence of the saturation correction is not quantified. Both issues are addressable with targeted tests, so I recommend major revision rather than rejection. The Appendix B external validation and the public release of code and data are substantial positives."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is the catalog paper the CIV absorption field has been waiting for, and the quantitative results look solid enough to ship after minor revision. The genuinely new thing is the sample: 101,487 CIV systems from 300,637 DESI DR2 quasars, 3-6x larger than SDSS DR7, with fine EW-redshift binning. The detection code is an existing matched-kernel doublet finder and the continuum code is picca, but the validation is real: ~3 million mock injections, a ~95% purity estimate from stacking, and an appendix that re-derives Omega_CIV from Cooksey+13 under two cosmologies and gets their trends back. That external check matters, and it makes me more confident than the reader was.\n\nThe completeness function is the main worry, and the stress-tester names it correctly: mocks are single double Gaussians with uniform EW, DR, and amplitude, while real CIV profiles are multi-component. If detectability at fixed EW depends on profile shape in a way that evolves with redshift, the 1/C_i weights would bias the 2-5x dN/dX and 3.8x Omega_CIV factors. But I want to keep this in proportion. The paper uses real DESI noise, injects ~3M systems, and validates the overall pipeline against an independent catalog; a profile mismatch would have to be large and strongly redshift-dependent to erase the headline trends, which agree qualitatively with earlier work. A direct high-res cross-check would settle it, and the paper does not have one. That is a legitimate revision request, not a rejection.\n\nSecond soft spot: the abstract states Omega_CIV values without the lower-limit qualifier that the body and figure caption clearly include. Since ~89% of systems are flagged partially saturated under AODM, this is more than a wording nitpick; the numbers are lower limits and should be labeled as such where the headline lives.\n\nThird: reproducibility. The DR1 catalog is public as a DESI VAC, and code and plot data are on Zenodo, but the DR2 catalog driving the paper's numbers is not out yet. The data availability section says it is coming; the paper should either release it with the arXiv posting or state the VAC timeline explicitly. Not a blocker, but it is the difference between a catalog paper people can use now and one they can only read about.\n\nWho this is for: IGM/CGM observers and simulators who need large absorber samples at 1.41.2$ Å). Using column densities from the apparent optical depth method, the authors find that the cosmic mass density of C IV, $\\Omega_{\\mathrm{CIV}}$, rises by a factor of about 3.8, from $(0.82\\pm0.05)\\times10^{-8}$ at $z\\approx4.5$ to $(3.16\\pm0.2)\\times10^{-8}$ at $z\\approx1.4$, which they state are lower limits because many systems are saturated. The differential equivalent-width frequency distribution declines exponentially and shows weak redshift evolution. From $\\Omega_{\\mathrm{CIV}}$ they derive a lower limit on the IGM metallicity, $\\log(Z_{\\mathrm{IGM}}/Z_\\odot)\\gtrsim -3.25$ at $z\\sim2.3$, and they argue that the overall trends trace the cosmic star formation history and He II photoheating rate, connecting carbon enrichment to the UV background over about 3 Gyr.","pith_inferences":["If the completeness function is biased toward Gaussian, single-component absorbers, the true evolution of $\\Omega_{\\mathrm{CIV}}$ could be shallower than the reported 3.8-fold factor; this could be tested by comparing against a high-resolution subsample, as described in the falsifier.","Combining this $\\Omega_{\\mathrm{CIV}}$ lower limit with C II/C III absorber measurements from the same spectra could yield an ionization-corrected carbon abundance, potentially raising the IGM metallicity estimate above the stated lower limit.","The catalog's size makes it possible to measure the C IV column-density distribution function per redshift bin, which would directly test whether the growth in $\\Omega_{\\mathrm{CIV}}$ is driven by more massive absorbers or more numerous ones.","Cross-correlating these absorbers with DESI galaxies would map the circumgalactic metal distribution at cosmic noon and could distinguish outflow-driven enrichment from accretion flows, something the current paper discusses but does not execute."],"forward_implications":["A factor-of-3.8 growth in $\\Omega_{\\mathrm{CIV}}$ from $z\\approx4.5$ to $z\\approx1.4$ provides a precise target for galaxy-formation simulations that model metal ejection and the multiphase circumgalactic medium.","The 2–5-fold increase in $dN/dX$ implies that either the absorber number density, the absorber cross-section, or both grew over this interval; the paper connects this to a roughly factor-of-2 growth in absorber radius and hence a factor-of-4 growth in cross-section.","Because the equivalent-width frequency distribution evolves weakly while $\\Omega_{\\mathrm{CIV}}$ grows strongly, the added mass is carried mainly by stronger, denser systems rather than by a change in the shape of the population.","The lower limits on IGM metallicity, $\\log(Z/Z_\\odot) \\gtrsim -3.25$ at $z\\sim2.3$, constrain the enrichment epoch of diffuse gas and are consistent with carbon being synthesized and expelled in step with the peak of cosmic star formation.","The public catalog of 101,487 systems, 50% complete at $EW_{1548}\\geq0.4$ Å and roughly 95% pure, can be used for absorber–galaxy cross-correlation studies near cosmic noon."],"supporting_citations":[{"why":"Supplies the matched double-Gaussian-kernel doublet finder (qsoabsfind) and the Monte Carlo completeness methodology used to detect and correct C IV absorbers.","marker":"Anand et al. 2021"},{"why":"Provides the continuum model (mean transmission times average quasar spectrum with linear color corrections) used to normalize DESI quasar spectra.","marker":"du Mas des Bourboux et al. 2020"},{"why":"Provides the apparent optical depth method and the empirical saturation corrections used to derive column densities and their uncertainties.","marker":"Savage & Sembach 1991"},{"why":"Establishes the completeness-weighted path-density definitions and the AODM approach for C IV that this paper extends.","marker":"Cooksey et al. 2010"},{"why":"Is the SDSS DR7 C IV catalog used as the main comparison baseline and as the validation target for the measured path densities and mass densities.","marker":"Cooksey et al. 2013"},{"why":"Provides the absorber-radius growth model (factor of about 2 from z ≈ 4 to z ≈ 1.5) used to interpret the dN/dX evolution as a cross-section effect.","marker":"Hasan et al. 2022"},{"why":"Supplies the linear relation between the C IV cosmic mass density and IGM metallicity used to convert mass densities into metallicity lower limits.","marker":"Ryan-Weber et al. 2009"},{"why":"Provides the adopted fraction of carbon in C IV (no more than 0.35) used in the metallicity estimate.","marker":"Oppenheimer & Davé 2006"},{"why":"Supplies the cosmology (Omega_m = 0.307, H0 = 67.7) used for comoving path lengths and the critical density.","marker":"Planck Collaboration et al. 2016"}],"fun_headline_variants":["101,487 CIV absorbers: cosmic carbon mass jumps 3.8x by z~1.4","Cosmic carbon density rises 3.8x from z=4.5 to z=1.4, per 101k CIV systems","DESI's 101k CIV absorbers: carbon mass grows 3.8x from z=4.5 to z=1.4","Carbon census: 101,487 CIV systems show 3.8x density rise to z=1.4","101k CIV absorbers chart 3.8x carbon rise over 3 Gyr"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The completeness function $C(EW,z)$ derived from idealized double-Gaussian mock absorbers is assumed to measure the true detection efficiency for real C IV systems; because real absorbers have complex velocity structure, blending, and non-Gaussian profiles, a mismatch would bias the completeness-corrected counts and the reported 3.8-fold evolution.","fun_headline_variants_meta":{"raw":{"variants":["101,487 CIV absorbers: cosmic carbon mass jumps 3.8x by z~1.4","Cosmic carbon density rises 3.8x from z=4.5 to z=1.4, per 101k CIV systems","DESI's 101k CIV absorbers: carbon mass grows 3.8x from z=4.5 to z=1.4","Carbon census: 101,487 CIV systems show 3.8x density rise to z=1.4","101k CIV absorbers chart 3.8x carbon rise over 3 Gyr"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001193,"raw_usage":{"total_tokens":5085,"prompt_tokens":1270,"completion_tokens":3815,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":886,"completion_tokens_details":{"reasoning_tokens":3662}},"tokens_in":886,"tokens_out":3815,"duration_ms":25186,"temperature":1.0,"reasoning_tokens":3662,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T05:33:04.269314+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Independent validation would settle the claim: take a few thousand DESI quasar sightlines with existing high-resolution follow-up spectra, identify C IV systems without any completeness correction, and measure $dN/dX$ and $\\Omega_{\\mathrm{CIV}}$ in matched redshift bins; if the high-resolution values do not rise by a factor of about 3.8 from $z\\approx4.5$ to $z\\approx1.4$, the Monte Carlo completeness function is the cause. A cheaper test is to inject realistic, multi-component non-Gaussian C IV profiles into the residual spectra and check whether $C(EW,z)$ changes by more than the quoted uncertainties.","supporting_citations":[{"cited_title":"W., Stemock , B., et al","cited_arxiv_id":null,"evidence_quote":"Provides the absorber-radius growth model (factor of about 2 from z ≈ 4 to z ≈ 1.5) used to interpret the dN/dX evolution as a cross-section effect."}],"review_version":1}