{"id":"7544c9af-989d-43bb-97a1-1d638be9bdde","arxiv_id":"2506.17379","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A 7.2-sigma kSZ detection and a 4.1-sigma patchy screening signal around DESI LRGs, with the screening signal attributed mostly to CMB lensing, leaving a 95% upper limit tau < 2.5e-4 for the gas optical depth.","lead":"This paper stacks microwave maps from the Atacama Cosmology Telescope around red galaxies from the DESI survey to study two faint gas signals: a Doppler shift called kSZ and a screening effect. It shows the screening signal is mostly contamination from gravitational lensing, and after correction sets an upper limit on the gas content of these galaxies.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The lensing template is the load-bearing element: it supplies roughly two-thirds of the patchy signal and is subtracted without a data-side validation.","rationale":"Agreeing with the reader's weakest assumption, I find the simulated lensing template to be the load-bearing pillar of the paper's central claim. The kSZ detection at 7.2σ is robust because it uses a velocity-weighted estimator that cancels additive foregrounds and does not require a lensing correction. The patchy signal is detected at 4.1σ, but its interpretation as an optical-depth measurement after lensing removal, the quoted upper bound, and the claimed agreement with kSZ all depend on the lensing template being correct in both amplitude and shape. The paper does not provide a data-side check: while it cites ACT DR6 lensing maps, it does not stack them on the same DESI LRG sample to validate the AbacusSummit prediction. The template's dependence on the halo population (mean mass, satellite fraction, miscentering, photo-z scatter) is acknowledged qualitatively but not marginalized over; a single simulation and a single Gaussian CMB realization are used. The transfer-function breakdown at small scales (App. A) is a separate but related limitation: the tau theory curve used for the amplitude fit is least reliable in the very bins where Fig. 5 shows the SNR is largest. This does not invalidate the kSZ measurement, and the concern is addressable, so a conditional verdict is appropriate; no change from the reader's verdict is needed.","tokens_in":47987,"tokens_out":6903,"duration_ms":76474,"concrete_test":"Stack the public ACT DR6 convergence map (refs [77,78]) at the same DESI Extended LRG positions and measure <kappa>(theta) in the same radial bins as the patchy profiles. Compare this measured profile with the AbacusSummit convergence profile used to generate the lensing template in Sec. IV B, including systematic shifts from photometric redshift outliers and halo-mass calibration. If the two differ by more than the combined statistical and systematic error, the lensing subtraction is biased and the tau upper bound must be revised; if they agree within errors, the lensing attribution is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central interpretation—that the 4.1σ patchy-screening signal is in excess of kSZ and that this excess is CMB lensing—rests entirely on the simulated lensing template of Sec. IV B. The template is built from AbacusSummit convergence maps and a random Gaussian CMB realization, stacked on halos matched only in mean mass (log M ≈ 13.3). It is then subtracted from the data, and the residual is used for the tau upper bound (Eqs. 18–20) and for the 'perfect agreement' with kSZ in Fig. 1. No cross-check against the actual lensing field over the same DESI LRGs is presented, even though ACT DR6 lensing maps are public and cited in the paper. A bias in the template amplitude or shape—from photometric-redshift outliers, satellite fraction, miscentering, or halo-mass calibration—propagates directly into the residual tau profile, the fitted amplitude A, and the 95% limit tau < 2.5e-4. The transfer-function limitation noted in App. A (r(k) much less than 1 below roughly 1 arcmin) affects the tau model rather than the lensing subtraction itself, but Fig. 5 shows that the first radial bins carry most of the SNR, so small-scale template errors are not harmless. The companion paper [91] that would independently derive the lensing bias is still in preparation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a joint measurement of the kinematic Sunyaev-Zel'dovich (kSZ) and 'patchy screening' (anisotropic screening) effects on the same sample of DESI Extended LRGs at z~0.7, using ACT DR5 temperature maps. The authors detect kSZ at 7.2 sigma and a patchy-screening signal at 4.1 sigma, finding the latter in excess relative to kSZ. They attribute approximately two-thirds of the patchy signal to CMB lensing contamination estimated from AbacusSummit N-body simulations, and after subtracting this lensing template they place a 95% upper bound of tau < 2.5e-4 on the sample's mean optical depth. They also compare the residual signal with IllustrisTNG and Illustris simulations, inferring strong baryonic feedback, and discuss prospects for measuring the rms line-of-sight velocity from the ratio of the two effects.","tokens_in":48365,"tokens_out":3591,"duration_ms":43295,"significance":"If the lensing attribution is correct, this is a useful demonstration that real-space patchy-screening estimators are strongly contaminated by CMB lensing and that the contamination can be quantified with simulations; the first self-consistent kSZ-plus-patchy measurement on the same galaxy sample also provides a path toward breaking the optical-depth/velocity degeneracy. Strengths of the paper include the use of public ACT DR5 and DESI DR9 data, the ThumbStack pipeline, and state-of-the-art simulations (AbacusSummit, IllustrisTNG, Illustris), as well as explicit caveats about transfer-function limitations and covariance underestimation. The central scientific conclusions, however, rest on a simulation-based lensing template that is not validated against observed lensing maps, and the tau upper bound and feedback comparison inherit that template's systematic uncertainty.","major_comments":[{"comment":"The paper itself notes that the transfer-function method used to paint gas onto N-body simulations breaks down on the smallest radial bins, where the cross-correlation coefficient r(k) becomes much smaller than unity; Fig. 5 shows that for the high-pass-filtered estimators most of the signal-to-noise is concentrated in the first one to two radial bins. Since the amplitude fit A and the feedback comparison in Figs. 1 and 2 rely on these small-scale bins, the analysis should either restrict the fit to scales where the transfer function is calibrated (r(k) > ~0.95) or include a model for the small-scale systematic error. As written, the central amplitude conclusions are partly supported by theory curves that are acknowledged to be unreliable in the regime where the data carry the most weight.","section":"Appendix A and Fig. 5"}],"minor_comments":[{"comment":"The text contains a typo: 'single-drequency' should be 'single-frequency'.","section":"Section II B"},{"comment":"The cutoff at ell ~ 2000 for the low-pass filter is motivated by preserving the sign of the primary CMB, but the threshold |T_lo| > 40 micro-K in Section III C is introduced without justification; a brief explanation of how this threshold was chosen would help the reader assess foreground contamination.","section":"Section III A"},{"comment":"The minus sign in the patchy estimator in Eq. (13) is stated without a derivation; since the sign determines whether the measured signal has the expected positive tau profile, a one-sentence explanation of the sign convention would improve clarity.","section":"Eq. (13)"},{"comment":"The statement that lensing makes up 'about two thirds' of the signal refers to the amplitude of the high-pass-filtered profile, but the profile is a function of radius; the paper should specify whether this fraction is evaluated at the peak bin, as an integrated quantity, or as an amplitude ratio.","section":"Section IV A"},{"comment":"The text says the CMB map is generated from a 'random Gaussian realization from the lensed C_l,' but does not specify whether this is the lensed primary CMB power spectrum computed with CAMB; please clarify the exact input spectrum and realization procedure.","section":"Appendix A"},{"comment":"Two references that are central to the method (the analytic lensing treatment and the transfer-function validation) are marked 'in preparation'; the manuscript should either provide the essential results in an appendix or state explicitly which conclusions would need revision if those works change.","section":"References [91] and [99]"}],"recommendation":"major_revision","confidential_remarks":"The paper's central scientific claim depends on a lensing template that is not yet validated against observed lensing data, and the quantitative results are tied to a companion paper [91] that is in preparation. This is a substantive correctness-risk issue rather than a presentation issue, so I recommend major revision rather than rejection: the measurements themselves, especially the kSZ detection, appear robust, and the lensing validation is a concrete, feasible addition. The editor may also wish to consider whether the manuscript's scope is appropriate given that 'self-consistent' comparison is not fully achieved until the lensing contribution is measured with the same data or justified with a published, independent estimate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nQuick take: this is a competent observational paper, and the central kSZ result is solid. The genuinely new thing is the self-consistent comparison of the two gas tracers on the same DESI LRG sample, together with the first quantification of how much CMB lensing contaminates the real-space stacked patchy screening estimator—roughly two thirds of the measured signal. That number matters for anyone interpreting patchy screening stacks, and the paper is upfront about it.\n\nWhat it does well: the kSZ detection at 7.2 sigma does not depend on the lensing model, so it stands on its own. The lensing contribution is estimated from external AbacusSummit simulations and added to the kSZ prediction, so the central comparison is not circular. The use of public pipelines (ThumbStack, pyrecon, AbacusSummit, IllustrisTNG) and the explicit discussion of estimator details make the analysis reproducible in spirit. The paper also correctly notes that the high-pass filter erases shape information, so amplitude is where feedback and satellite fraction enter.\n\nWhere I part company with the authors' framing: the lensing template is load-bearing and is subtracted without a direct validation against the actual lensing field over the same galaxies. ACT DR6 lensing maps are public and cited in the paper, so an obvious cross-check is missing. A bias in the template amplitude or shape—from photo-z outliers, satellite fraction, miscentering, or halo-mass calibration—propagates into the residual tau profile, the fitted amplitude A, and the quoted 95% upper bound tau < 2.5e-4. The template uncertainty is not propagated, and the tau conversion itself depends on an assumed sample mean optical depth of 1.6e-4. The transfer-function breakdown below roughly an arcmin, noted in Appendix A, is acknowledged, but the first radial bins carry most of the SNR, so it is not a negligible caveat. Calling the post-subtraction agreement with kSZ \"perfect\" is also too strong; the significance after subtraction is much lower.\n\nThe citation pattern is appropriate; the kSZ measurement on this sample was in Ref. [61], and the paper says so. The reliance on a companion paper [91] for the analytic lensing treatment is a bit uncomfortable, but the simulation-based estimate here is enough for a first look.\n\nBottom line: this paper deserves peer review. The robust kSZ measurement and the lensing contamination estimate are worth refereeing even if the tau upper bound tightens after a data-side lensing validation. I would ask the authors to add an ACT-DR6-based check of the lensing template and to propagate its uncertainty before acceptance. It is a useful paper for CMB-LSS people, and I would bring it to our reading group.","headline":"A useful paper that makes the first same-sample kSZ/patchy screening comparison and shows CMB lensing dominates the stacked patchy estimator; the lensing subtraction needs a data-side check before the tau upper limit is trusted.","tokens_in":48807,"tokens_out":2247,"would_cite":true,"duration_ms":22695,"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 patchy-screening signal around DESI luminous red galaxies is dominated by CMB lensing, not gas.","keywords":["kinematic Sunyaev-Zel'dovich effect","patchy screening","CMB lensing contamination","optical depth","DESI luminous red galaxies","ACT DR5","baryonic feedback","CMB secondary anisotropies"],"falsifier":"Stack the same patchy screening estimator around the same DESI LRGs, but subtract a lensing convergence map reconstructed directly from CMB observations instead of the simulated template; if the residual profile changes significantly or no longer agrees with the kSZ profile, the simulated lensing template is biased, and the $\\tau<2.5\\times10^{-4}$ bound would need revision.","tokens_in":47748,"feed_emoji":"🌌","tokens_out":11529,"duration_ms":113542,"temperature":0.7,"pith_summary":"This paper attempts to put the kinematic Sunyaev-Zel'dovich (kSZ) effect and the patchy screening effect on the same footing: both are imprints of free electrons on the cosmic microwave background, and both should measure the same optical depth $\\tau$ of gas around galaxies. Stacking ACT temperature maps on DESI luminous red galaxies at $z\\approx0.7$, the authors detect kSZ at $7.2\\sigma$ and a patchy screening signal at $4.1\\sigma$, and argue that about two thirds of the patchy signal is gravitational lensing of the CMB rather than gas. After subtracting a lensing template built from $N$-body simulations, the two probes agree, and the data yield a 95% upper limit $\\tau<2.5\\times10^{-4}$ for the sample. The kSZ effect requires a reconstructed velocity field, whereas patchy screening does not, so their agreement is a check on systematics and on the gas density itself. If the lensing correction holds, comparing the two breaks the degeneracy between optical depth and velocity, allowing tests of baryonic feedback and velocity-sensitive cosmological models.","feed_headline":"Two-thirds of patchy-screening signal is CMB lensing","feed_subtitle":"After subtracting it, kSZ and screening agree on DESI gas density: tau < 2.5e-4.","key_machinery":"The load-bearing machinery is a pair of stacked estimators applied to the same filtered CMB maps. The kSZ estimator (Eq. 12) averages the small-scale temperature decrement around each galaxy weighted by the reconstructed line-of-sight velocity, normalized by the velocity rms and a cross-correlation coefficient $r$; the patchy screening estimator (Eq. 13) averages the small-scale temperature decrement weighted by the sign of the large-scale primary CMB temperature, normalized by the mean absolute large-scale temperature. Non-overlapping high-pass ($\\ell\\gtrsim2350$) and low-pass ($\\ell\\lesssim2000$) filters separate the relevant scales. The lensing contamination is isolated by stacking a convergence map generated from $N$-body simulations at the same halo positions, and a transfer-function method paints gas onto the $N$-body density field to predict the pure optical depth profile.","core_discovery":"The central claim is that the apparent patchy screening signal around DESI luminous red galaxies is not primarily gas damping the primary CMB: it is dominated by CMB lensing. The stacked estimator averages a small-scale temperature decrement weighted by the sign of a large-scale temperature fluctuation, and gravitational lensing couples those two scales, producing a profile that mimics screening. Simulated lensing maps stacked at the same halo positions show that this contamination makes up roughly two thirds of the measured $4.1\\sigma$ signal. Once the lensing contribution is removed, the residual optical depth profile agrees with the kSZ-derived profile, so the paper quotes $\\tau<2.5\\times10^{-4}$ at 95% confidence for a sample with mean $\\tau\\approx1.6\\times10^{-4}$. The comparison with hydrodynamical simulations also indicates that baryonic feedback around these galaxies is stronger than fiducial models predict, and closer to a simulation with aggressive feedback.","pith_inferences":["A direct test of the paper's lensing interpretation would replace the simulated lensing template with a lensing convergence map reconstructed from the CMB itself; if the residual $\\tau$ profile is template-dependent, the two-thirds lensing attribution is not yet settled.","Because the high-pass filter erases most profile-shape information, the feedback comparison rests almost entirely on the amplitude of the stacked profile; a matched or multi-band filter that preserves shape could separately constrain halo mass, satellite fraction, and feedback strength.","The paper's upper bound is only about 1.6 times the predicted mean optical depth, so a modest increase in survey area or depth should either detect the gas or force a substantial revision of the predicted halo gas content."],"forward_implications":["The measured $4.1\\sigma$ patchy screening signal should not be read as a gas detection until lensing is subtracted; after subtraction it is consistent with the kSZ-derived profile, but the gas-only significance drops below a detection.","For the Extended DESI LRG sample, the optical depth is bounded as $\\tau<2.5\\times10^{-4}$ at 95% confidence, so current data cannot distinguish the predicted mean $\\tau\\approx1.6\\times10^{-4}$ from a more gas-poor population.","Running both estimators on identical maps and the same galaxy sample gives a direct systematics cross-check, since additive foregrounds such as the cosmic infrared background and thermal SZ cancel differently in the two estimators.","If both signals are measured at high signal-to-noise, their amplitude ratio is proportional to the root-mean-square line-of-sight velocity of the host halos, which can constrain velocity-sensitive models such as modified gravity and phantom dark energy."],"supporting_citations":[{"why":"Introduces the patchy screening estimator and filtering scheme (v1) that this paper applies to ACT and DESI data.","marker":"[62]"},{"why":"Provides the temperature-inversion estimator variant and methodology for detecting patchy screening.","marker":"[63]"},{"why":"Argues that CMB lensing generally biases patchy screening estimators and can exceed the signal, motivating the lensing subtraction.","marker":"[90]"},{"why":"Supplies the velocity reconstruction cross-correlation coefficient used to normalize the kSZ optical depth estimate.","marker":"[84]"},{"why":"Provides a previous ACT-times-DESI kSZ measurement of the same galaxy sample, giving the baryonic feedback comparison.","marker":"[61]"},{"why":"Calibrates the halo occupation distribution and mean halo mass, log M about 13.3, for the DESI LRG sample.","marker":"[88]"},{"why":"Supplies the light-cone catalogues from N-body simulations used to build lensing and optical depth maps.","marker":"[92]"},{"why":"Introduces the transfer-function method used to paint gas and optical depth onto N-body density fields.","marker":"[101]"}],"fun_headline_variants":["Lensing dominates patchy screening: tau < 2.5e-4","Patchy screening is 2/3 lensing; gas tau < 2.5e-4","After lensing, kSZ and screening agree on gas","CMB lensing skews patchy screening; gas bound tightens"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument stands on the assumption that the lensing template built from $N$-body simulations matches the true CMB lensing contamination of the stacked patchy screening signal in both amplitude and shape, since that contaminant makes up about two thirds of the measured signal and is subtracted from it.","fun_headline_variants_meta":{"raw":{"variants":["Lensing dominates patchy screening: tau < 2.5e-4","Patchy screening is 2/3 lensing; gas tau < 2.5e-4","After lensing, kSZ and screening agree on gas","CMB lensing skews patchy screening; gas bound tightens"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001185,"raw_usage":{"total_tokens":4994,"prompt_tokens":1146,"completion_tokens":3848,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":762,"completion_tokens_details":{"reasoning_tokens":3763}},"tokens_in":762,"tokens_out":3848,"duration_ms":26561,"temperature":1.0,"reasoning_tokens":3763,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T19:09:59.776224+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Stack the same patchy screening estimator around the same DESI LRGs, but subtract a lensing convergence map reconstructed directly from CMB observations instead of the simulated template; if the residual profile changes significantly or no longer agrees with the kSZ profile, the simulated lensing template is biased, and the $\\tau<2.5\\times10^{-4}$ bound would need revision.","supporting_citations":[{"cited_title":"An Improved Forecast of Patchy Reionization Reconstruction with CMB","cited_arxiv_id":"1106.4313","evidence_quote":"Argues that CMB lensing generally biases patchy screening estimators and can exceed the signal, motivating the lensing subtraction."}],"review_version":1}