{"id":"4f451f7f-d20c-49be-a38f-009261e17a35","arxiv_id":"2412.15849","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Using a wrong reionization model can bias LiteBIRD's constraints on the shape of the primordial gravitational-wave spectrum, but large-scale E-mode data would expose and remove such models.","lead":"This forecast study simulates how assuming the wrong reionization history could bias future measurements of primordial gravitational waves with the LiteBIRD-like satellite. It finds that large-scale CMB polarization (E-modes) can identify and discard the problematic reionization scenarios, preserving reliable gravitational-wave constraints.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The abstract's robustness claim relies on an E-mode veto that is never applied or quantified; the biased posterior in Fig. 4 is obtained with E-modes already in the likelihood, so 'easily excluded' is an assertion.","rationale":"The reader's CONDITIONAL verdict is appropriate. My main concern overlaps with the reader's weakest assumption but is more immediate: even before introducing foregrounds, 1/f noise, or cut-sky effects, the paper never performs the statistical test that would justify the E-mode veto. The biased posterior in Fig. 4 is produced by an MCMC that already includes E-mode data, which makes the subsequent claim that E-modes would exclude the scenario an unsupported assertion. The visual mismatch in Fig. 6 may look large, but a robust forecast needs a quantitative criterion and a demonstration that applying it removes the bias. The paper does provide a credible and transparent simulation: the exponential-model null result is a useful check, the code modifications to CLASS and emcee are standard, and the binning of the tensor spectrum is a reasonable extension of earlier work. However, the robustness claim is the main advertised conclusion, and it is not backed by a computed test. The reader's secondary worry about E-mode systematics is real, but it is a further layer on top of the missing statistical veto. I therefore keep the verdict at CONDITIONAL rather than moving it, with the condition being that the authors should add a quantitative E-mode exclusion test, e.g., a model-comparison statistic, and ideally repeat the forecast after applying that veto or with a more flexible reionization template.","tokens_in":12819,"tokens_out":5123,"duration_ms":53019,"concrete_test":"Compute the full E+B likelihood ratio between the true exotic model and the best-fit tanh-template model using Eq. (10) with the same mock covariance and lmin=2. If the resulting delta-chi^2 is not large enough to reject the exotic model at high significance, the proposed E-mode veto does not exclude this scenario even under ideal white noise, and the 'robust' claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim has two parts: (1) an exotic reionization history can bias r, deltaP6, and deltaP7 by more than 1 sigma, and (2) large-scale E-modes would 'easily exclude' such a scenario, making the PGW constraints robust. The first part is demonstrated in Sec. IV. The second part is not. The MCMC likelihood in Eq. (10) already includes E-mode data with lmin=2 and lmax=1300, yet the exotic-model run in Fig. 4 still produces a badly biased posterior (tau = 0.0954 +/- 0.0006 versus fiducial 0.08, a >20 sigma bias). The claimed exclusion is based only on a visual mismatch between the best-fit and fiducial E-mode spectra in Fig. 6. No goodness-of-fit statistic, delta-chi^2, or Bayesian evidence is reported, and no procedure is specified for deciding when to discard a fit. Thus the paper shows that a wrong tanh template biases PTPS constraints, but it does not show that any implementable analysis step would catch this before reporting r. This gap is load-bearing because the 'robust' conclusion is exactly the claim that the bias would not survive in a real analysis. The paper's own Sec. V caveats (Galactic foregrounds, 1/f noise, cut-sky likelihood approximation) further weaken the unquantified veto, and the single hand-picked exotic model from Sec. II B 2 means the frequency of dangerous reionization histories that evade the veto is unknown.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses a forward-simulation MCMC forecast to study how an incorrect assumption about the reionization history affects constraints on a binned primordial tensor power spectrum (PTPS) for a LiteBIRD-like all-sky CMB experiment. The authors generate mock E- and B-mode polarization data with two alternative 'true' reionization histories (an exponential model and a hand-picked 'exotic' model biased toward double-reionization behavior) and fit them with a tanh reionization template, simultaneously varying r, eight PTPS bin amplitudes δP_i, and τ. They find that the exponential model produces no significant bias, while the exotic model shifts r, δP6, and δP7 by more than 1σ. The abstract concludes that large-scale E-mode measurements would 'easily exclude' the exotic scenario, making the PGW constraints robust.","tokens_in":13207,"tokens_out":3261,"duration_ms":31354,"significance":"If fully substantiated, the paper would provide a useful cautionary result for LiteBIRD-era analyses: the tanh template is adequate for smooth reionization histories, but some non-standard histories can bias nontrivial PTPS parameters even when current external constraints are satisfied. The study extends Mortonson & Hu (2007) and Hiramatsu et al. (2018) to a more realistic LiteBIRD noise level and to a generic PTPS, and the exponential-model result is a clean, well-posed demonstration. The main weakness is that the central robustness claim—that an E-mode veto protects the analysis—is asserted from a visual discrepancy rather than demonstrated with a quantitative model-selection or goodness-of-fit statistic. The paper also relies on a single adversarially selected exotic model and an idealized noise model, so the practical relevance of the veto, and the frequency of dangerous reionization histories, remain unknown. The manuscript is transparent about these simplifications, which is a strength, but the abstract's 'robust' conclusion is not yet supported by the presented evidence.","major_comments":[{"comment":"The claim that large-scale E-mode power would 'easily exclude' the exotic scenario is not demonstrated. The likelihood in Eq. (10) already includes E-mode data with lmin=2 and lmax=1300, yet the MCMC run for the exotic model in Fig. 4 still converges to a heavily biased posterior (τ = 0.0954 ± 0.0006 versus the fiducial 0.08, a >20σ shift). The exclusion is based on a visual mismatch between the best-fit and fiducial E-mode spectra in Fig. 6, with no χ² difference, p-value, or Bayesian evidence reported. To make the abstract's robustness claim load-bearing, the authors should specify and apply an explicit analysis step: for example, compute the Δχ² of the best-fit tanh model against the mock E-mode data, show that it would be rejected at high significance for the exotic model while accepted for the exponential model, and discuss the threshold at which a fit would be discarded before reporting r.","section":"Abstract and Sec. IV, Fig. 6"},{"comment":"The 'exotic' model is a single realization selected after generating many random models to find one that maximizes the bias, but the paper does not report how many models were generated, how the selection was performed, or whether other models with similar bias exist. This makes it impossible to assess whether the demonstrated bias represents a realistic, non-negligible risk or an extreme adversarial corner. The paper should quantify the frequency of dangerous reionization histories: for instance, report the fraction of generated models that pass the Planck E-mode χ² selection (for τ0=0.054) or the τ0=0.08 prior and that produce >1σ bias in any PTPS parameter. Without this, the statement in the abstract that the constraints are 'robust against the reionization uncertainties' goes beyond what the single-example analysis can support.","section":"Sec. II B 2 and Sec. IV"},{"comment":"The E-mode veto is evaluated under an idealized measurement assumption: white noise, a Gaussian beam, full-sky likelihood with fsky scaling, and no foregrounds or 1/f noise. As the paper itself notes, the likelihood approximation in Eq. (10) is not valid for a real cut-sky analysis, and large-scale polarization is dominated by Galactic foregrounds. The discrepancy that is supposed to identify the wrong reionization model appears mainly at l < 10 in Fig. 6, exactly the multipole range where foreground residuals and 1/f noise are most dangerous. The authors should either add a simple foreground-residual noise term and show that the veto survives, or soften the abstract's 'robust' claim to something conditional, such as 'would exclude in an idealized full-sky, white-noise measurement.'","section":"Sec. V and Eq. (10)"}],"minor_comments":[{"comment":"There is a typo: 'TThe other model' should read 'The other model'.","section":"Sec. II B 2"},{"comment":"The section title 'RESUL TS' contains an unintended space; it should be 'RESULTS'.","section":"Sec. IV heading"},{"comment":"The caption labels the exotic models with τ=0.054 (green) and τ=0.08 (red), but the text in Sec. IV refers only to 'the red line'. Please state explicitly that Fig. 4 uses the τ=0.08 exotic model, since this choice is important for interpreting the bias.","section":"Fig. 1 caption"},{"comment":"The arXiv identifier appears as 'arXiv:1009.3204S' with a trailing 'S'; this should be 'arXiv:1009.3204'.","section":"Reference [69]"},{"comment":"The sentence 'an increase (decrease) of r is compensated by decreasing (increasing) δPi with i ≥ 2' would be clearer if it also noted whether this anti-correlation is a result of the reionization bump or of the recombination bump, since the later discussion attributes the small-scale bias to high-redshift ionization.","section":"Sec. IV, discussion of degeneracies"}],"recommendation":"major_revision","confidential_remarks":"The paper's main result is the demonstration that a specific non-tanh reionization history can bias PTPS constraints; that is a valid and useful cautionary result. However, the abstract claims robustness on the basis of an unquantified E-mode veto, and the single adversarial model plus idealized noise assumptions leave a significant gap between the claim and the evidence. The paper can likely be repaired by adding a quantitative rejection test and a more systematic survey of generated reionization models; I would not reject on the current evidence. The novelty is incremental relative to Mortonson & Hu (2007) and Hiramatsu et al. (2018), but the updated experimental assumptions and the PTPS focus are sufficient for a specialized journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a competent, incremental forecast about how reionization-model error affects LiteBIRD-era constraints on a binned primordial tensor spectrum. The exponential-model test is clean, and the paper is honest about its assumptions. But the abstract's claim that large-scale E-modes would 'easily exclude' the biased exotic scenario is not backed by any statistical test. The stress-test note is right on this.\n\nWhat is new: the extension of Mortonson & Hu to a generic PTPS with bin amplitudes, at LiteBIRD sensitivity, with a deliberately adversarial reionization history. The MCMC setup is standard, and the demonstration that a wrong tanh template can bias r and the small-scale bins is a useful caution for the community.\n\nWhere it is soft: (1) The E-mode veto is only a visual discrepancy in Fig. 6. The likelihood in Eq. (10) already includes E-mode data down to lmin=2, and the exotic-model fit still gives tau=0.0954 +/- 0.0006 against a fiducial 0.08. That means the model is a poor fit, but the paper never reports a goodness-of-fit statistic, a delta-chi^2, or a Bayesian evidence. Without a decision rule, 'excluded' is an assertion, not a result. (2) The exotic model is a single hand-picked worst case. That is fine for demonstrating existence of bias, but it does not tell you how often such models slip through. (3) The forecast uses ideal white noise and a simple fsky scaling; the authors list foregrounds and 1/f noise as future work, yet these are the systematics that could bury the E-mode discrepancy.\n\nNone of this undermines the central bias result. It undermines the 'robust' conclusion. A referee should ask for a quantitative rejection test, which is a modest addition.\n\nRecommendation: this deserves peer review. The paper is within its rights to claim a cautionary result, but the robustness claim needs to be earned with a formal test before publication. I would cite the bias demonstration in my own work, not the veto.","headline":"A solid forecast on reionization-model bias for LiteBIRD-era tensor constraints, but the advertised E-mode veto is asserted, not quantified.","tokens_in":13696,"tokens_out":3527,"would_cite":true,"duration_ms":30579,"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":"An incorrect reionization history can bias constraints on primordial gravitational waves inferred from CMB polarization, but the large-scale E-mode power spectrum can identify and exclude such scenarios, keeping the constraints robust for…","keywords":["reionization history","primordial gravitational waves","CMB polarization","B-mode","tensor-to-scalar ratio","LiteBIRD","E-mode power spectrum","reionization bump"],"falsifier":"Run the same forecast with a realistic E-mode noise model that includes Galactic foreground residuals and 1/f noise at l<30; if the discrepancy between the best-fit and fiducial E-mode spectra shrinks below the error bars while the bias in r, δP6, and δP7 persists, the paper's robustness conclusion would fail. Alternatively, generating an ensemble of exotic reionization histories and checking whether some biased cases evade the E-mode veto would directly test the claim.","tokens_in":12615,"feed_emoji":"🌌","tokens_out":3580,"duration_ms":29289,"temperature":0.7,"pith_summary":"This paper asks whether uncertainty in the reionization history—the epoch when neutral hydrogen became ionized—could corrupt a future LiteBIRD-like experiment's measurement of primordial gravitational waves from the cosmic microwave background's polarization. The authors fit mock CMB data generated with two alternative reionization histories using the standard tanh reionization template. They find that an exponential reionization history leaves the recovered tensor-to-scalar ratio and binned tensor power spectrum unbiased, while a purposely generated 'exotic' history biases several parameters by more than 1σ. Crucially, the same exotic history produces a large discrepancy between the observed and best-fit E-mode power spectrum at the reionization bump, so measuring E-modes on large scales would flag and exclude it. If correct, this makes future PGW constraints robust against reionization uncertainty provided the E-mode measurement is clean.","feed_headline":"E-mode veto guards gravity-wave limits from reionization bias","feed_subtitle":"Forecast shows a wrong reionization model breaks constraints unless large-scale E-mode data flag it.","key_machinery":"The argument runs on three objects: (1) a binned primordial tensor power spectrum $P_h(k) = P^{\\mathrm{fid}}_h(k) + \\delta P_i$ over eight logarithmic bins in $k$, defined following Hiramatsu et al. (2018); (2) the tanh reionization template used to fit the data, parameterized by $z_{\\mathrm{reio}}$ and hence the optical depth $\\tau$; and (3) the E-mode reionization bump, which responds to the ionization history at $z \\simeq 5$–$22$ and provides the diagnostic that separates a wrong model from an acceptable one. The forecast uses a Wishart log-likelihood for E and B modes with LiteBIRD-like white noise, $f_{\\mathrm{sky}}=0.7$, and a Gaussian 30 arcmin beam.","core_discovery":"The paper's central claim is that the shape of the primordial tensor power spectrum can be recovered robustly despite uncertainty in the reionization history, as long as the large-scale E-mode polarization is measured. In the forecast, mock data are created with either an exponential reionization model or a randomly generated exotic model and then fitted with a tanh template. With the exponential model, all parameters—the tensor-to-scalar ratio r, the optical depth τ, and the eight k-space bins δP1–δP8—land within the 68% confidence region. With the selected exotic model (τ=0.08), r, δP6, and δP7 are biased by more than 1σ, and τ is biased severely. The same exotic history changes the E-mode spectrum at multipoles l=10–30, producing a best-fit E-mode spectrum that disagrees with the fiducial one by more than the observational errors; the authors conclude that this discrepancy would 'easily exclude' the exotic scenario and make the PGW constraints robust.","pith_inferences":["If Galactic foregrounds, $1/f$ noise, or cut-sky effects degrade the large-scale E-mode measurement, the veto could fail and the >1σ bias could survive; testing the claim with a more complete noise model would be a natural extension of this forecast.","The 'exotic' model is a single sampled history from a random ensemble; scanning the full ensemble would reveal how often biases exceed 1σ and how often the E-mode veto catches them.","The same veto logic could be applied to other parameters that affect the reionization bump, such as a running tensor spectral index, and to other all-sky experiments with different noise levels."],"forward_implications":["For a LiteBIRD-like experiment, constraints on $r$ and on the small-scale tensor bins $\\delta P_6$ and $\\delta P_7$ are robust to reionization uncertainty of the exponential type, with all parameters staying within the 68% confidence region.","A reionization history with high ionization fraction at high redshift, like the exotic example, biases $r$ and the small-scale bins by more than 1σ when the data are fit with a tanh template.","The same data's large-scale E-mode power spectrum provides a consistency check that excludes such biased scenarios, preventing the bias from surviving in the final PGW constraints.","Accurate measurement of the E-mode reionization bump is therefore crucial for robustly constraining the shape of the primordial tensor power spectrum."],"supporting_citations":[{"why":"Mortonson & Hu (2007) is the analysis this paper expands; it established how reionization uncertainty affects constraints on r for earlier experiments.","marker":"[54]"},{"why":"Hiramatsu et al. (2018) supplies the binned primordial tensor power spectrum parameterization and the baseline forecast setup used here.","marker":"[62]"},{"why":"LiteBIRD Collaboration provides the experimental configuration, noise level, and projected sensitivity assumed in the forecast.","marker":"[30]"},{"why":"CLASS is the public code modified to compute CMB polarization power spectra with altered reionization histories and tensor spectra.","marker":"[63]"},{"why":"Planck PR4 E-mode data are used to reject randomly generated exotic models that deviate significantly from current large-scale measurements.","marker":"[66]"},{"why":"Giare et al. motivates the alternative optical depth value τ=0.08 obtained without relying on large-scale E-mode data.","marker":"[43]"},{"why":"CAMB is the source of the exponential reionization model used as one of the two true reionization histories.","marker":"[65]"},{"why":"Lau et al. (2013) is a previous partial-sky study of reionization's impact on r, providing context for the full-sky LiteBIRD forecast.","marker":"[55]"}],"fun_headline_variants":["E-mode data keeps gravity wave limits robust to reionization models","Large-scale E-modes expose bad reionization models for PGW probes","E-mode spectrum vetoes reionization histories that bias gravity waves","Reionization bias in gravity wave limits flagged by E-mode power"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The E-mode veto works only if the large-scale E-mode power spectrum is measured as cleanly as assumed—white noise, no Galactic foregrounds, no 1/f noise, no cut-sky effects—so that the discrepancy identifying the wrong reionization model remains visible.","fun_headline_variants_meta":{"raw":{"variants":["E-mode data keeps gravity wave limits robust to reionization models","Large-scale E-modes expose bad reionization models for PGW probes","E-mode spectrum vetoes reionization histories that bias gravity waves","Reionization bias in gravity wave limits flagged by E-mode power"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000413,"raw_usage":{"total_tokens":2160,"prompt_tokens":997,"completion_tokens":1163,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":613,"completion_tokens_details":{"reasoning_tokens":1086}},"tokens_in":613,"tokens_out":1163,"duration_ms":10787,"temperature":1.0,"reasoning_tokens":1086,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T11:02:06.249077+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same forecast with a realistic E-mode noise model that includes Galactic foreground residuals and 1/f noise at l<30; if the discrepancy between the best-fit and fiducial E-mode spectra shrinks below the error bars while the bias in r, δP6, and δP7 persists, the paper's robustness conclusion would fail. Alternatively, generating an ensemble of exotic reionization histories and checking whether some biased cases evade the E-mode veto would directly test the claim.","supporting_citations":[{"cited_title":"Nakane, M","cited_arxiv_id":null,"evidence_quote":"Mortonson & Hu (2007) is the analysis this paper expands; it established how reionization uncertainty affects constraints on r for earlier experiments."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Hiramatsu et al. (2018) supplies the binned primordial tensor power spectrum parameterization and the baseline forecast setup used here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"LiteBIRD Collaboration provides the experimental configuration, noise level, and projected sensitivity assumed in the forecast."},{"cited_title":"Mukherjee, S","cited_arxiv_id":null,"evidence_quote":"CLASS is the public code modified to compute CMB polarization power spectra with altered reionization histories and tensor spectra."},{"cited_title":"Hiramatsu, E","cited_arxiv_id":null,"evidence_quote":"Planck PR4 E-mode data are used to reject randomly generated exotic models that deviate significantly from current large-scale measurements."}],"review_version":1}