{"id":"c5604947-af3d-4f1c-9378-9455287a10db","arxiv_id":"1908.05303","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Simulations show that polarized foreground leakage can bias global 21 cm signal extraction by 20-30% and, in a selected subset of low rotation-measure realizations, produce an EDGES-like absorption trough.","lead":"Polarized galactic foregrounds, Faraday-rotated and leaking into a single-polarization dipole, can bias the recovered global 21 cm cosmic signal and can even reshape a standard absorption trough into one resembling the anomalous EDGES detection. The paper matters because it offers a systematic, non-exotic explanation for a headline cosmological anomaly and informs how global-signal experiments should handle polarization.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The EDGES-mitigation claim hinges on the 'low φ' Faraday-depth cutoff (φ<5 rad/m²) and a hand-tuned 10% contamination amplitude; neither choice is independently calibrated, and the all-φ model fails to produce the effect.","rationale":"The paper's central claim is that unaccounted polarized foreground contamination can produce an EDGES-like absorption trough and overestimate the signal amplitude by ~30%, mitigating the need for exotic physics. The load-bearing link is the model of the polarized foreground itself. In the LF band, the all-φ S18 template yields too much contamination for the extraction to converge, so the EDGES-like result depends entirely on the 'low φ' variant (Figure 8). The low-φ cutoff at 5 rad/m² is motivated by qualitative references to frequency-dependent depolarization, but no quantitative validation or sensitivity analysis is given. Similarly, the ~30% amplitude bias is derived after reducing the template to 10% of its value (Figure 9), a normalization chosen to match EDGES residuals rather than from a physical depolarization model. If either of these choices is unrealistic, the headline mitigation of exotic physics is unsupported. I agree with the reader's weakest_assumption. The paper does, however, include a no-contamination control (Figure 6) and the rotated-antenna consistency check, which are real pieces of evidence that the general threat of polarization leakage is plausible. But the EDGES-specific claim is a conditional demonstration, not a robust result. My concern does not move the reader's conditional verdict; it reinforces it. The suggested test—a parameter scan over the Faraday-depth cutoff and the amplitude factor—would settle whether the EDGES-like profile is an artifact of the specific choices.","tokens_in":14343,"tokens_out":7808,"duration_ms":69389,"concrete_test":"Perform a sensitivity scan over the two model choices: (i) Faraday-depth cutoff φ_max ∈ {2, 5, 10} rad/m² (plus the full S18 distribution) and (ii) contamination amplitude factor f ∈ {5%, 10%, 20%} of the template. For each combination, rerun the LF-band signal extraction from Section 3 with the fiducial Gaussian input and flattened-Gaussian model. Record the fraction of converging realizations and the mean reconstructed amplitude/shape. If the EDGES-like trough and the ~30% bias occur only for φ_max≈5 rad/m² and f=10%, the central claim is determined by the tuning; if they occur over a broad range, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the LF band, the EDGES-like distorted trough (Figure 8) appears only when the polarized foreground template is the 'low φ' variant (φ<5 rad/m²); with the full S18 Faraday-depth distribution ('all φ'), the extraction fails to converge and the effect disappears (Section 3, second bullet). The physical justification for the 5 rad/m² cutoff rests on qualitative references (Haverkorn et al. 2004; Bernardi et al. 2009; Lenc et al. 2016) and no sensitivity analysis is provided. In addition, the quantitative bias estimates underlying the 'amplitude overestimated by ~30%' statement are computed after reducing the polarized spectrum to 10% of the template (Section 2.4), a value chosen 'in qualitative agreement with the magnitude of the residual rms in the Bowman et al. (2018a) observations' rather than derived from a depolarization model. Because the all-φ and low-φ cases bracket the possible sky, and the 10% reduction is a free normalization, the central claim that unaccounted polarized foregrounds 'can produce an enhanced and distorted 21 cm absorption trough similar to the anomalous profile' is not robust to plausible variations in the assumed foreground model.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents sky-averaged 21 cm signal simulations with a dipole beam at the Murchison Radio-astronomy Observatory, including total-intensity foregrounds, a 21 cm signal (Gaussian, flattened Gaussian, or tanh reionization model), noise, and polarized foregrounds from the Spinelli et al. (2018) Faraday-depth simulations. The polarized foregrounds are considered in an 'all φ' and a 'low φ' (φ < 5 rad/m^2) variant, and in some cases are scaled to 10% of the template amplitude. The authors use the hibayes Bayesian code to extract the signal. They show that recovery is unbiased when foregrounds are smooth (Figure 6), whereas polarized leakage biases the reconstructed amplitude and shape or prevents convergence. They report that, for the low-φ case, a standard Gaussian input reconstructed with a flattened-Gaussian model can appear as an enhanced, EDGES-like absorption trough, and that when the EDGES flattened Gaussian is both input and model, the recovered amplitude is biased high by roughly 20-30%.","tokens_in":14530,"tokens_out":7869,"duration_ms":73122,"significance":"If the polarization-contamination mechanism is validated, the paper would supply an astrophysical systematic that can mimic or distort the EDGES Cosmic Dawn absorption trough, reducing the need for exotic-physics explanations. The quantitative bias estimates are potentially important for global-signal experiments, and the rotated-dipole test argument (Section 4) is a useful caution: such tests do not automatically exclude polarized contamination. The simulation machinery is appropriate, and the unbiased-recovery control and the rms distributions are valuable. However, the EDGES-specific conclusion rests on several modeling choices whose sensitivity is not demonstrated, and the body of the paper contains two distinct scenarios that are conflated in the abstract. These issues currently weaken the central claim.","major_comments":[{"comment":"The EDGES-like trough in Figure 8 is obtained only for the 'low φ' (φ < 5 rad/m^2) variant. Section 3 states that in the 'all φ' case almost all realizations are discarded for the Gaussian-input test. The cutoff is justified only qualitatively through references, and the 10% amplitude rescaling is chosen to match the EDGES residual rms rather than derived from a depolarization model. No sensitivity analysis is provided for either parameter. Because the effect disappears under the alternative 'all φ' model, the claim that unaccounted polarized foregrounds can produce the EDGES-like profile is not robust to plausible variations in the assumed foreground model.","section":"§2.4, §3"},{"comment":"The Gaussian-input/flattened-Gaussian-extraction test lacks a control run without polarized contamination. A flattened-Gaussian fit to a Gaussian input can itself introduce amplitude and width biases; without that control, the enhanced and distorted reconstructed profile cannot be cleanly attributed to polarized contamination. The authors should show the same extraction with the same priors and model but with T_Q = 0.","section":"§3, second bullet; Figure 8"},{"comment":"The abstract states that the reconstructed EDGES amplitude can be overestimated by around 30%, mitigating the need for exotic physics. In the body, the ~20-30% amplitude bias is reported for the case where the EDGES flattened Gaussian is both the input and the model (Section 4, Figure 7), not for the standard-Gaussian-input case in Figure 8, where the reconstructed amplitude is far larger than the input. These are logically distinct assertions: one assumes the anomalous EDGES signal exists and is partially contaminated, while the other claims contamination can create the signal from a standard input. The abstract conflates them, and the '30%' figure does not support the trough-generation claim. The wording should be corrected.","section":"Abstract; §4"},{"comment":"In the Gaussian-input test, only about 30% of the low-φ realizations are retained after discarding cases with high-frequency troughs or non-convergence. The selection criteria are stated, but the analysis should report how the resulting EDGES-like profile depends on the retention thresholds and should demonstrate that the reported bias is not driven by selecting the most extreme realizations. A quantitative statement of the fraction of realizations that produce an EDGES-like profile under plausible selection criteria would make the 'can produce' claim more meaningful.","section":"§3, second bullet"}],"minor_comments":[{"comment":"The log-polynomial in Eq. (14) is written with sum over n = 1 to N, but Table 1 lists p0 through p4. If the intended order is N = 4, the table should list only p0..p3; otherwise the sum upper limit should be N = 5 or the indexing should be adjusted.","section":"Eq. (14), Table 1"},{"comment":"The priors for the flattened-Gaussian parameters w and τ are not specified in the text, unlike the Gaussian parameters. Please state these priors explicitly.","section":"§3"},{"comment":"The HF-band beam is obtained by linearly scaling the 100 MHz model up to 200 MHz, with no justification. This is a significant simplification for a band that is central to part of the analysis; it should at least be noted as a limitation.","section":"§2.1"},{"comment":"Figure 5 shows rms distributions from 100 realizations, whereas the signal-extraction runs use 50 realizations. Please clarify whether the same realizations are used and whether the 50 are a subset of the 100.","section":"§2.4, Figure 5"}],"recommendation":"major_revision","confidential_remarks":"The paper contains useful simulations and an important cautionary message, but the EDGES-specific conclusion needs a control without polarized contamination, a sensitivity analysis of the low-φ cutoff and the 10% amplitude rescaling, and a correction of the abstract's conflation of the two scenarios. I would recommend requiring these revisions before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper is worth reading for the general result — unmodeled polarized foregrounds can bias global 21 cm parameter recovery at levels comparable to the signal. The EDGES-specific claim, that such contamination can mimic the observed trough and remove the need for exotic physics, is plausible but not yet robust. I'd send it to a serious referee, with the expectation that the EDGES section gets tightened.\n\nWhat's new: Switzer & Liu (2014) treated polarization leakage analytically; Spinelli et al. add Faraday-depth-resolved all-sky polarized maps, a realistic LWA dipole beam, and a full Bayesian extraction pipeline. That combination allows them to quantify the bias on recovered 21 cm parameters, not just the leakage level. The control case with smooth foregrounds (their Figure 6) confirms the pipeline is unbiased without contamination. The rotated-antenna consistency point is genuinely useful: they show that xx and yy extractions can differ by only 10-20% even when contaminated, so the EDGES rotated-antenna check does not exclude this systematic.\n\nSoft spots, in order of seriousness. First, the EDGES-mitigation result depends on the 'low φ' (φ<5 rad/m2) polarized sky and on reducing the contamination to 10% of the template. The low-φ cutoff is motivated by qualitative references but no sensitivity analysis is shown; with the full Faraday-depth distribution the extraction mostly fails. The 10% amplitude is chosen to match EDGES residual rms, not derived from a depolarization model. These choices bracket the plausible sky, so the 'can produce an EDGES-like trough' statement is real but not yet quantitative. Second, the 1.5σ tension with the input is weak, and the ~30% retention rate after selection plus the need to discard high-frequency troughs means the reported mean profile comes from a filtered subset. That's not fatal — they are transparent about it — but it means the headline claim is a demonstration of possibility, not a probability. Third, no code or data released; exact reproducibility is limited, though the method is described in enough detail to re-implement.\n\nBottom line: the warning about polarization as a general systematic for single-dipole experiments is well supported and should influence design of future global signal experiments. The EDGES-specific explanation is a hypothesis that deserves testing with independent polarized sky models and a depolarization treatment. This paper deserves a proper peer review; I'd recommend a major-revision path that either strengthens the EDGES section or reframes it more clearly as illustrative.","headline":"A careful simulation study whose general caution about polarized leakage is solid, but whose EDGES-mitigation claim rests on a hand-tuned 10% contamination level and a selected subset of realizations.","tokens_in":15180,"tokens_out":1709,"would_cite":true,"duration_ms":16892,"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":"This paper argues that unaccounted polarized foreground emission can reproduce the anomalous 21 cm absorption trough reported by the EDGES experiment, possibly inflating its amplitude by about 30%.","keywords":["21 cm cosmology","global signal","polarized foregrounds","Faraday rotation","cosmic dawn","EDGES anomaly","foreground contamination","Bayesian parameter estimation"],"falsifier":"Measure the actual Stokes Q and U sky at 50–100 MHz over degree-to-tens-of-degree scales with a calibrated polarimetric instrument, compute the leakage spectrum that a single-polarization dipole would see, and compare its amplitude and Faraday-depth structure with the model predictions; if the real contamination is far below about 150 mK rms or is spectrally smooth, the claimed EDGES-mimicking bias would not occur.","tokens_in":14017,"feed_emoji":"📡","tokens_out":9051,"duration_ms":80332,"temperature":0.7,"pith_summary":"Global 21 cm experiments that measure the sky-averaged signal rely on foregrounds having smooth spectra, but a single-polarization antenna also picks up linearly polarized Galactic emission that Faraday rotation corrugates across frequency. This paper simulates realistic dipole observations with polarized full-sky maps and shows that this contamination biases the reconstructed 21 cm signal in both the 50–100 MHz Cosmic Dawn band and the 100–200 MHz reionization band. The main result is that, for a restricted Faraday-depth model of polarization, unaccounted contamination can turn a standard ~150 mK Gaussian absorption trough into an enhanced, broadened profile resembling the anomalous 78 MHz trough reported by the EDGES experiment, with the recovered amplitude inflated by roughly 30%. If correct, this would weaken the case for exotic physics (for example dark-matter interactions) built on the unusually deep EDGES absorption.","feed_headline":"Polarized foregrounds can mimic the EDGES 21 cm trough","feed_subtitle":"Simulated Faraday-rotated emission deepens the absorption trough by ~30%, easing the need for exotic physics.","key_machinery":"The mechanism is polarization leakage in a single-polarization dipole: the measured total-intensity spectrum includes Stokes Q, and Faraday rotation of Galactic synchrotron emission makes that contamination oscillate with frequency rather than follow a smooth power law. The simulation machinery consists of (i) all-sky Stokes Q and U maps from the Spinelli et al. (2018) simulations, with two variants—full Faraday-depth range and 'low φ' (only Faraday depths below 5 rad/m²); (ii) a parametric analytic dipole beam model to integrate the sky over the observing window; and (iii) a Bayesian Monte-Carlo fit that models foregrounds as a fourth-order log-polynomial and the 21 cm signal as a Gaussian or flattened Gaussian. The 'low φ' variant is the one that yields EDGES-like profiles; the 'all φ' case is usually too contaminated to allow signal extraction.","core_discovery":"The paper's central claim is that polarized foreground contamination—emission measured because a single-polarization dipole couples to Stokes Q via the antenna response—has a frequency structure that is not smooth after Faraday rotation, so standard smooth-foreground subtraction cannot remove it. In simulations that inject a fiducial Gaussian 21 cm absorption profile at 78.3 MHz and then fit a flattened Gaussian, the recovered profile in the 'low φ' polarized-sky model is systematically deeper and wider; its amplitude is in mild tension (~1.5σ) with the input profile and can mimic the EDGES detection. The authors further note that under the contamination hypothesis the reconstructed EDGES signal amplitude can be overestimated by around 30%, which mitigates the need to invoke exotic physics such as dark-matter cooling or an excess radio background. They also show that at 10% contamination the biases persist in both bands, and that a 90-degree antenna rotation does not remove them.","pith_inferences":["If the true low-frequency polarized sky is closer to the 'all φ' realization, the EDGES-like profile would probably not survive; a direct measurement of the Faraday-depth distribution at 50–100 MHz over an observing field would discriminate between these regimes.","The same bias mechanism should affect any single-polarization global-signal measurement from any site, so amplitudes and shapes inferred by other experiments in this band may carry a comparable systematic.","A testable consequence: a dual-polarization global-signal antenna with accurate relative calibration—so that Stokes Q and U leakage is removed—should recover a shallower and narrower absorption trough than the reported EDGES profile if contamination is the cause."],"forward_implications":["If the contamination hypothesis holds, the reported EDGES trough amplitude is likely overestimated by roughly 30%, so the deviation from standard astrophysical predictions is much smaller than claimed.","Global 21 cm experiments must either model polarized leakage explicitly or reduce it (for example by dual-polarization differencing and careful calibration) even when the contamination is only about 10% of the current polarized-sky estimate.","In the 100–200 MHz band, polarized leakage biases the recovered midpoint of reionization by up to about 10% and can underestimate the duration of reionization by up to a factor of about 10.","Antenna-rotation consistency checks are not sufficient to rule out polarized contamination, because both orthogonal polarizations show the same qualitative bias in the simulations."],"supporting_citations":[{"why":"Supplies the full-sky polarized foreground maps (Stokes Q and U with Faraday-depth structure) that generate the contaminating spectra.","marker":"Spinelli et al. (2018)"},{"why":"The EDGES detection being tested; its anomalous flattened-Gaussian profile is the target of the mimicry and its amplitude is the one inflated by ~30%.","marker":"Bowman et al. (2018a)"},{"why":"Provides the fourth-order log-polynomial total-intensity foreground model, the dipole beam model, and the fiducial Gaussian 21 cm signal parameterization used in the simulations.","marker":"Bernardi et al. (2015)"},{"why":"Supplies the Bayesian extraction approach used to reconstruct the 21 cm signal from the simulated spectra.","marker":"Bernardi et al. (2016)"}],"fun_headline_variants":["Faraday-rotated foregrounds can fake EDGES signal","Polarized emission mimics EDGES 21 cm anomaly","Polarized foregrounds may explain EDGES depth","EDGES signal could be polarized foreground artifact","Polarized foregrounds bias 21 cm absorption trough"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The modeled polarized sky—especially the low Faraday-depth variant that produces the EDGES-like profile—must match the actual polarized Galactic emission a dipole sees at 50–100 MHz, and the optimistic case additionally assumes that contamination is reduced to 10% of the model.","fun_headline_variants_meta":{"raw":{"variants":["Faraday-rotated foregrounds can fake EDGES signal","Polarized emission mimics EDGES 21 cm anomaly","Polarized foregrounds may explain EDGES depth","EDGES signal could be polarized foreground artifact","Polarized foregrounds bias 21 cm absorption trough"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000199,"raw_usage":{"total_tokens":1418,"prompt_tokens":1038,"completion_tokens":380,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":654,"completion_tokens_details":{"reasoning_tokens":298}},"tokens_in":654,"tokens_out":380,"duration_ms":3870,"temperature":1.0,"reasoning_tokens":298,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:19:31.929366+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the actual Stokes Q and U sky at 50–100 MHz over degree-to-tens-of-degree scales with a calibrated polarimetric instrument, compute the leakage spectrum that a single-polarization dipole would see, and compare its amplitude and Faraday-depth structure with the model predictions; if the real contamination is far below about 150 mK rms or is spectrally smooth, the claimed EDGES-mimicking bias would not occur.","supporting_citations":[],"review_version":1}