{"id":"dabbd374-8a3d-449c-8b30-764d6582e982","arxiv_id":"2507.17693","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Using 107 localized FRBs with DESI BAO and supernova distances, the author estimates fIGM near 0.99 in the constant model but finds Bayesian evidence inconclusive between constant and evolving baryon fraction.","lead":"This paper combines 107 fast radio bursts with galaxy and supernova data to measure how much of the universe's normal matter sits between galaxies. It concludes that current data cannot yet tell whether that fraction changes with cosmic time.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The headline fIGM,0 ~ 0.99 sits at the physical prior boundary; the single-parameter host-DM model in Eq. 2.2 is the most load-bearing assumption and can bias this value.","rationale":"The reader's weakest-assumption statement identifies the same load-bearing issue: a single global host-DM parameter cannot capture the broad, asymmetric distribution of real host galaxy dispersion measures. The algebraic derivation in Eqs. 3.7 and 3.8 appears correct, and the paper's conservative statement that current data cannot conclusively distinguish constant from evolving fIGM is supported by the small Bayes factors. However, the quantitative claim fIGM,0 ~ 0.99 is made with the posterior mass pressing against the physical boundary fIGM <= 1, which heightens sensitivity to the host-DM model. A concrete hierarchical reanalysis would settle whether the headline value is an artifact of that assumption. Because this concern does not overturn the paper's main cautious conclusion, the existing CONDITIONAL verdict remains appropriate; no verdict change is needed, but the robustness check should be requested.","tokens_in":14609,"tokens_out":5159,"duration_ms":57687,"concrete_test":"Reanalyze the 107-FRB sample with a hierarchical host-DM model: draw per-FRB DM_host from a log-normal distribution with free mean and width, or use published host-galaxy DM posterior estimates, and recompute the fIGM,0 posterior for the constant case. If the posterior shifts from 0.999 by more than about 0.1, or if the upper bound no longer sits at 1, the central claim is not robust to host-DM modeling.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equations 2.2 and 3.1 model every host-galaxy contribution with a single source-frame mean DMhost,0 plus a known Gaussian scatter of 30 pc/cm3. In reality host DM is broad, asymmetric, and depends on galaxy type, inclination, and local plasma; the analysis excludes the extreme host FRB 20190520B but does not model the remaining scatter. Because the constant-case posterior peaks at fIGM,0 = 0.999, at the physical boundary fIGM <= 1, even a modest mis-modeling of the host contribution can push the estimate to the boundary and compress the upper error. The paper does not test this by marginalizing over a free host-scatter parameter or by using measured host-DM estimates. The central no-evolution conclusion is more conservative and is supported by the small Bayes factors, so it is less threatened, but the quantitative headline value fIGM,0 ~ 0.99 is not secure. Secondary issues, such as unseeded random draws for 41 missing sigma_obs values and omission of alpha terms in sigma_IGM for the time-dependent case, reinforce the need for a robustness pass.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper constrains the baryon fraction in the IGM, fIGM, by combining 107 well-localized fast radio bursts with distance information from DESI DR2 BAO transverse-mode measurements and from the Pantheon+ and DES Y5 supernova samples. Using the integration-by-parts method of the author's previous work, the dispersion measure of the IGM is re-expressed in terms of luminosity distance, and two models are fitted: a constant fIGM,0 and a time-dependent fIGM(z)=fIGM,0+alpha z/(1+z), with a shared host-galaxy DM parameter. For the constant case the paper reports fIGM,0 about 0.99 with host DM around 113 pc/cm3; for the time-dependent case it reports fIGM,0 around 0.94-0.97 and alpha around 0.6-0.7. A Bayesian model selection analysis gives |ln B| between 0.34 and 0.71, from which the paper concludes that a conclusive answer about fIGM evolution cannot be obtained from current FRB data. The central derivation in Section 3.2 appears correct, but several statistical modeling choices affect the quantitative headline values.","tokens_in":14729,"tokens_out":12310,"duration_ms":128328,"significance":"If the quantitative estimates survive robustness checks, the method provides a useful cosmology-independent cross-check of the baryon inventory and of IGM ionization at z<1, and the use of three independent distance datasets is a genuine strength. The paper is appropriately cautious in its main model-selection conclusion: the small Bayes factors do support the statement that current data are inconclusive about fIGM evolution. However, the headline constant-case value fIGM,0=0.999 sits at the physical boundary fIGM<=1 and rests on a single-parameter host-DM model, and the analysis also contains several reproducibility and error-propagation gaps. The qualitative conclusions are defensible, but the quantitative headline is not yet secure.","major_comments":[{"comment":"The host-galaxy contribution is modeled by a single global parameter DMhost,0 with a fixed 30 pc/cm3 Gaussian scatter. Host DM is known to vary with galaxy type, inclination, star formation, and local plasma; excluding FRB 20190520B removes one extreme object but does not model the remaining scatter. Since the constant-case posterior peaks at fIGM,0=0.999 (Table 1), the inferred fIGM,0 is sensitive to this host-DM model. Please add a robustness test that marginalizes over a free sigma_host,0 or uses measured host-DM estimates where available, and show how fIGM,0 and its uncertainty change.","section":"Section 2, Eq. (2.2); Section 3.1.1, Eq. (3.1)"},{"comment":"The prior ranges for fIGM,0, alpha, and DMhost,0 are not stated. The constant-case estimate fIGM,0=0.999 with upper error 0.007-0.010 sits at the edge of the physically allowed region fIGM<=1, so the quoted upper error and the agreement with fIGM=1 are sensitive to prior truncation. Please report the exact priors and quantify the boundary effect, for example by testing a prior extending above unity or by reporting the posterior mass within a small interval of fIGM,0=1.","section":"Section 3.2; Table 1"},{"comment":"For 41 FRBs, sigma_obs is randomly drawn from a Gaussian whose mean and standard deviation are computed from the other events, but no random seed or draw is documented. Because sigma_obs enters the likelihood through sigma_tot in Eq. (3.1), the quoted parameter intervals and Bayes factors are not reproducible and may depend on the particular realization. Please publish the seed and draws, or marginalize over the missing sigma_obs with a stated prior, and confirm that the central estimates are stable.","section":"Section 3.1.1; Table 3"},{"comment":"Equation (3.2) gives sigma_IGM = A fIGM,0 sqrt(sigma_DL^2/c^2 + sigma_I^2/c^2), which is the error propagation for the constant case. For the time-dependent model in Eq. (3.7), the coefficient of sigma_DL is A[fIGM,0 + alpha z/(1+z)] and the coefficient of sigma_I is A(fIGM,0 + alpha). Using A fIGM,0 for both terms underestimates the IGM variance in the time-dependent case and can bias alpha and the associated evidence. Please propagate the uncertainties directly from Eq. (3.7).","section":"Section 3.2, Eq. (3.2)"},{"comment":"The value delta = 230 sqrt(z) pc/cm3 is fixed, and Section 4 attributes the difference from the previous [18] results to exactly this choice. Since delta enters every sigma_tot in Eq. (3.1), the error bars and Bayes factors in Tables 1 and 2 are conditional on this single ad hoc noise model. The previous paper varied delta; the present analysis should either provide a sensitivity scan over delta or justify the fixed value, otherwise the reported uncertainties are not robust to a key modeling assumption.","section":"Section 3.1.1; Section 4"}],"minor_comments":[{"comment":"The model-selection language is not consistent with the Jeffreys' scale stated in Section 4: the reported values |ln B| = 0.34, 0.53, and 0.71 fall in the 'inconclusive' range (0-1), not 'weak evidence' as stated in the abstract and conclusions. Please revise the wording to say the evidence is inconclusive.","section":"Section 4; Table 2; Conclusions"},{"comment":"In the list of excluded FRBs, the third low-redshift event is printed as 'FRB [35]' with no event name; please insert the correct identifier.","section":"Section 3.1.1"},{"comment":"Equation (3.2) refers to 'Eq. 3.9' for sigma_I, but no Eq. (3.9) appears in the manuscript; either number the integral after Eq. (3.8) or refer to it directly.","section":"Section 3.1.1"},{"comment":"The text says that 41 FRBs are marked with the symbol dagger in the sigma_obs column, but the table as presented does not show these markers; please ensure they are visible in the published table.","section":"Table 3"},{"comment":"The conclusions refer to 'FRB + DES' instead of 'FRB + DES Y5'; please use consistent dataset labels throughout.","section":"Section 5"},{"comment":"Footnote 6 contains 'MBMB' instead of 'MB'; please fix the typo.","section":"Footnote 6"},{"comment":"The Gaussian Process reconstruction produces a joint predictive distribution for DL at the FRB redshifts, but the paper does not state whether correlations between reconstructed DL values, or between DL and the integral in Eqs. (3.7)-(3.8), are propagated into sigma_IGM. Please clarify this point or justify the diagonal approximation.","section":"Section 3.2"}],"recommendation":"major_revision","confidential_remarks":"This is a single-author follow-up to the author's previous work; the methodological novelty is limited to applying the existing method to a larger FRB sample and to two new distance datasets. That is acceptable for a journal like JCAP, but the robustness tests requested in the report (host-DM modeling, prior specification, reproducibility of the missing-sigma_obs draws, and error propagation) should be completed before publication. The paper is otherwise readable and the qualitative conclusion is appropriately conservative."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Thais Lemos has extended her earlier cosmology-independent FRB method to a 107-event sample combined with DESI DR2 BAO and two SN catalogs. The useful result is not the framework — that is from [18] — but the measurement: constant fIGM,0 ~ 0.99 across three datasets and a Bayes-factor comparison showing that current FRB data cannot distinguish constant from evolving fIGM. The integration by parts in Section 3.2 is correct, the sample is well documented, and the cautious headline conclusion is supported by the small Bayes factors. Credit where due: this is honest incremental work, not a breakthrough.\n\nThe soft spots are real but mostly secondary. The central constant-case estimate sits at the physical boundary fIGM ≤ 1, and the single-parameter host-DM model (Eq. 2.2) is doing a lot of work there. If true host DM is broad or asymmetric, the posterior can be dragged to the boundary and the upper error compressed. The paper excludes the most extreme host but doesn't test the host-scatter assumption, e.g., by marginalizing over σ_host or using measured host DMs. That makes the quantitative headline value less secure than the paper suggests. The no-evolution conclusion, being driven by small Bayes factors, is less threatened.\n\nSecondary issues: 41 missing σ_obs values are replaced with random draws from a Gaussian without reporting a seed or testing robustness; the error propagation for the time-dependent case omits α terms in σ_IGM; and the text calls ln B ~ 0.3–0.7 weak evidence while its own Jeffreys scale labels that range inconclusive. These are fixable in revision. The choice of fixed H0 and MB is at least declared, and the self-citation to [18] is transparent and appropriate.\n\nBottom line: for someone working on FRB cosmology or IGM baryon inventory, this is a paper worth engaging with. The method is not novel, but the expanded sample and three-dataset concordance are a legitimate step forward. It deserves peer review with a request for a robustness pass rather than desk rejection.","headline":"A transparent, incremental extension of the author's earlier FRB method to 107 events; the no-evolution conclusion is solid, but the headline fIGM ~ 0.99 sits at the physical boundary and depends on a single-parameter host-DM model that needs a robustness test.","tokens_in":15371,"tokens_out":1438,"would_cite":true,"duration_ms":15365,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["98.80.-k"],"model":"deepseek-v4-flash","headline":"Using 107 well-localized fast radio bursts and model-independent distances from DESI BAO and two supernova catalogs, this paper estimates the intergalactic-medium baryon fraction at fIGM,0 ≈ 0.99 and finds only weak evidence that it…","keywords":["fast radio bursts","baryon fraction","intergalactic medium","dispersion measure","DESI baryon acoustic oscillations","type Ia supernovae","cosmological model independence","missing baryons"],"falsifier":"Measure host-galaxy dispersion independently for a subset of the 107 FRBs, for instance from host-galaxy emission measures or from scattering screens in repeating bursts, and compare the resulting distribution to the single fitted value near 113 pc/cm$^3$; if the measured mean or median differs by more than the quoted 12 pc/cm$^3$ uncertainty, or if the host-DM distribution is strongly asymmetric, the constant-case $f_{\\mathrm{IGM},0} \\approx 0.99$ is not robust and the single-parameter host model must be replaced.","tokens_in":14284,"feed_emoji":"📡","tokens_out":10400,"duration_ms":91047,"temperature":0.7,"pith_summary":"This paper tries to pin down the fraction of cosmic baryons that live in the intergalactic medium, $f_{\\mathrm{IGM}}$, by combining the dispersion measures of 107 well-localized fast radio bursts with distance measurements from DESI BAO and two Type Ia supernova catalogs. Its central move is a relation that rewrites the IGM dispersion measure in terms of luminosity distance, so distances are taken from the data rather than from a chosen cosmological model. For a constant baryon fraction the authors find $f_{\\mathrm{IGM},0} \\sim 0.99^{+0.01}_{-0.07}$ at $1\\sigma$, with a mean host-galaxy dispersion contribution near $113$ pc/cm$^3$. A Bayesian model comparison between the constant case and an evolving parameterization yields only weak evidence, so the current FRB sample cannot decisively say whether $f_{\\mathrm{IGM}}$ changes with redshift. The result matters because it bears on the long-standing 'missing baryons' problem and shows that FRB dispersion measures can act as an independent, model-light census of cosmic baryons.","feed_headline":"FRB data put the IGM baryon fraction near 0.99","feed_subtitle":"Closes the low-redshift missing-baryon budget using distances taken from data, not a cosmological model.","key_machinery":"The load-bearing object is the dispersion-measure relation obtained by integrating Eq. (2.3) by parts and writing the Hubble parameter in terms of luminosity distance $D_L$, so that $\\mathrm{DM}_{\\mathrm{IGM}}$ depends on $D_L(z)$ and an integral of $D_L(z')/(1+z')$ rather than on an assumed cosmology. Distance–redshift curves are reconstructed non-parametrically from the BAO and supernova catalogs with Gaussian processes, and the free parameters — $f_{\\mathrm{IGM},0}$ and $\\mathrm{DM}_{\\mathrm{host},0}$, plus $\\alpha$ in the evolving case — are fit by MCMC. The host-galaxy dispersion enters through a single global parameter $\\mathrm{DM}_{\\mathrm{host},0}$ scaled by $(1+z)$, and the cosmic-web scatter in electron density is folded in as $\\delta = 230\\sqrt{z}$ pc/cm$^3$.","core_discovery":"The paper's claim is that the observed extragalactic dispersion measures of 107 well-localized FRBs, combined with luminosity distances reconstructed from DESI DR2, DES Y5, and Pantheon+, give $f_{\\mathrm{IGM},0} \\sim 0.99$ at $1\\sigma$ for the constant parameterization, with the host-galaxy contribution $\\mathrm{DM}_{\\mathrm{host},0}$ around $112{-}116$ pc/cm$^3$, and that these values are mutually consistent across the three independent distance datasets. For the time-dependent parameterization $f_{\\mathrm{IGM}}(z) = f_{\\mathrm{IGM},0} + \\alpha z/(1+z)$, the recovered $f_{\\mathrm{IGM},0}$ values are lower but statistically compatible at $2\\sigma$, and $\\alpha$ is consistent with zero within the quoted errors. The Bayesian evidence comparison gives log-Bayes factors of $-0.53$, $-0.71$, and $+0.34$ for the three datasets, all in the inconclusive-to-weak range, which the authors read as meaning that a conclusive answer about the time evolution of $f_{\\mathrm{IGM}}$ cannot be achieved from the current FRB data, with weak evidence favoring the constant case in two of the three combinations.","pith_inferences":["A natural extension is to replace the single host-DM parameter with a population distribution — e.g., a log-normal with scatter estimated from the observed spread in host-galaxy dispersion — to test whether $f_{\\mathrm{IGM},0} \\approx 0.99$ survives when host variation is modeled rather than averaged away.","The sign flip of the Bayes factor between Pantheon+ and the other two datasets suggests that the distance reconstruction, such as which supernova sample or Gaussian-process kernel is used, may be influencing model preference; checking this would clarify whether evolution claims are data-driven or method-driven.","If a thousand-scale localized FRB sample pushed $f_{\\mathrm{IGM}}$ toward unity at all redshifts, the baryon census problem would shift from 'missing baryons in the IGM' to locating the remaining baryons in galaxies, halos, and the warm-hot phase, a map FRB statistics could provide directly.","The same $D_L$-based rewriting could be applied to the transverse comoving distance from future BAO or gravitational-wave standard siren measurements, making the probe applicable to datasets where supernova distances are not available."],"forward_implications":["If $f_{\\mathrm{IGM},0} \\approx 0.99$ is correct, the intergalactic medium at low redshift holds essentially all baryons predicted by big-bang nucleosynthesis, effectively closing the low-redshift 'missing baryons' budget.","The estimated host-galaxy dispersion, $\\mathrm{DM}_{\\mathrm{host},0} \\sim 112{-}116$ pc/cm$^3$, gives a quantitative prior for modeling FRB host environments and for separating host and IGM contributions in future samples.","Distinguishing a constant from an evolving $f_{\\mathrm{IGM}}$ will require many more localized FRBs rather than more precise distance catalogs, because the current Bayes factors are all in the weak-evidence regime.","Since DESI, DES Y5, and Pantheon+ yield consistent $f_{\\mathrm{IGM}}$ values despite disagreeing on dark-energy parameters, the probe is insensitive to the very cosmological tensions motivating it."],"supporting_citations":[{"why":"Supplies the original model-independent DM–$D_L$ method and the 17-FRB analysis this paper extends.","marker":"[18]"},{"why":"Sets the Milky Way halo dispersion contribution ($\\mathrm{DM}_{\\mathrm{halo}}=50$ pc/cm$^3$) and frames the localized-FRB baryon census.","marker":"[28]"},{"why":"Provides the statistical model for cosmic electron-density fluctuations, used here as $\\delta = 230\\sqrt{z}$ pc/cm$^3$.","marker":"[20]"},{"why":"Supplies the DESI DR2 BAO transverse distances converted to luminosity distances via $D_L=(1+z)D_M$.","marker":"[5]"},{"why":"Provides one of the Type Ia supernova distance catalogs whose distance moduli are reconstructed into $D_L$.","marker":"[71]"},{"why":"Provides the second Type Ia supernova catalog used to reconstruct luminosity distances.","marker":"[72]"},{"why":"Fixes the absolute magnitude $M_B$ and Hubble constant used to convert distance moduli and to evaluate the theoretical dispersion measure.","marker":"[73]"},{"why":"Supplies the Gaussian-process reconstruction used to estimate $D_L$ and its uncertainty at each FRB redshift.","marker":"[75]"},{"why":"Provides the MCMC sampler used to fit $f_{\\mathrm{IGM},0}$, $\\alpha$, and $\\mathrm{DM}_{\\mathrm{host},0}$.","marker":"[76]"},{"why":"Provides the nested-sampling algorithm used to compute Bayesian evidence and the Bayes factors that drive the model-comparison conclusion.","marker":"[84]"}],"fun_headline_variants":["FRBs and DESI pin IGM baryon fraction at ~0.99","107 FRBs yield IGM baryon fraction ~0.99","FRB distances close missing baryon budget, f_IGM≈0.99","IGM baryon fraction from FRBs: ~0.99 with DESI data","Weak evidence for constant IGM baryon fraction from FRBs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The estimate rests on treating host-galaxy dispersion as a single global parameter, $\\mathrm{DM}_{\\mathrm{host},0}$ times $(1+z)$; in reality host galaxies vary widely in gas content, orientation, and local plasma, and if that spread is broad or skewed the inferred IGM baryon fraction could shift.","fun_headline_variants_meta":{"raw":{"variants":["FRBs and DESI pin IGM baryon fraction at ~0.99","107 FRBs yield IGM baryon fraction ~0.99","FRB distances close missing baryon budget, f_IGM≈0.99","IGM baryon fraction from FRBs: ~0.99 with DESI data","Weak evidence for constant IGM baryon fraction from FRBs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000295,"raw_usage":{"total_tokens":1785,"prompt_tokens":1086,"completion_tokens":699,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":702,"completion_tokens_details":{"reasoning_tokens":598}},"tokens_in":702,"tokens_out":699,"duration_ms":6538,"temperature":1.0,"reasoning_tokens":598,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:19:13.249814+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure host-galaxy dispersion independently for a subset of the 107 FRBs, for instance from host-galaxy emission measures or from scattering screens in repeating bursts, and compare the resulting distribution to the single fitted value near 113 pc/cm$^3$; if the measured mean or median differs by more than the quoted 12 pc/cm$^3$ uncertainty, or if the host-DM distribution is strongly asymmetric, the constant-case $f_{\\mathrm{IGM},0} \\approx 0.99$ is not robust and the single-parameter host model must be replaced.","supporting_citations":[],"review_version":1}