{"id":"61615c54-71a8-40d3-8362-272ddd78090b","arxiv_id":"2607.23339","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Non-linear velocity terms cancel in real-space linear reconstruction, but redshift-space distortions reintroduce a 10–20% small-scale suppression of the stacked kSZ signal.","lead":"This paper tests what the stacked kinetic Sunyaev-Zel'dovich (kSZ) signal actually measures when galaxy velocities are estimated from a galaxy catalogue. It finds that standard redshift-space velocity reconstruction introduces a scale-dependent 10–20% suppression of the signal, a bias that will need to be modelled for next-generation surveys.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Satellite gas stripping assumption is the key unverified input: the ~10% RSD-induced suppression and its satellite attribution rest on ionized gas around satellites tracking the host halo, which is cited but not checked in FLAMINGO.","rationale":"I agree with the reader's identification of the satellite gas stripping assumption as the weakest link. The paper's decomposition is exact, and the qualitative picture—linear real-space reconstruction washes out nonlinear terms, while RSD reintroduces a scale-dependent suppression—is well supported by the simulation analysis. The strongest quantitative claims, however, are the 10% suppression and the 1–2σ significance for current data. Both inherit their normalization and physical interpretation from the choice of host-halo velocities for satellites. The paper itself acknowledges the sensitivity of the large-scale profile to this choice in Appendix A, yet the RSD-reconstructed case is not separately tested. A direct measurement of gas velocities around satellite subhaloes in the simulation would settle whether the stripped-gas assumption is physically realised. This is a testable, concrete concern, not a fatal flaw: the conditional verdict is appropriate, and I would not move it to accept or reject without this check. The paper deserves credit for its transparent decomposition, multiple FLAMINGO variants, and the explicit trade-off discussion; the concern is about the calibration of one physical input, not about the overall argument's soundness.","tokens_in":24211,"tokens_out":9593,"duration_ms":107705,"concrete_test":"In the same FLAMINGO snapshots used for the DESI-like LRG sample at z=0.7, select the satellite subhaloes in the mock catalogue. For each satellite, compute the ionized gas mass within the subhalo's bound radius and the mass-weighted mean line-of-sight gas velocity in that region; compare it with the host halo velocity and with the subhalo velocity. Quantify the fraction of satellites with significant bound gas and the mean gas velocity offset. Then rerun the RSD velocity reconstruction and stacked kSZ analysis under two assumptions: (a) all gas within satellite subhaloes follows the subhalo velocity, and (b) all such gas follows the host halo velocity. If the small-scale suppression changes by more than ~2–3% in amplitude or the satellite contribution shifts materially, the assumption is load-bearing for the quoted 10% and 1–2σ numbers. If the suppression is insensitive, the concern is","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—that RSD in the velocity reconstruction produces a ~10% small-scale suppression driven primarily by satellite Fingers-of-God—depends directly on the treatment of satellite gas in the decomposition. Section 3.3.1 assigns satellite stacked velocities to the host halo rather than the subhalo, with the justification that ionized gas around satellites is largely stripped and therefore moves with the host. This choice enters Eq. 4 through v0 = v_h and controls the size of the velocity-decorrelation term (third term in Eq. 7). If a non-negligible fraction of gas remains bound to satellites and retains their orbital velocity, then v' is not merely a smooth radial decorrelation but includes a coherent satellite-velocity component. The FoG-induced reconstruction error would then couple to this additional momentum, altering both the amplitude of the ~10% suppression and its attribution to satellites. The paper supports the stripping assumption only with citations (He et al. 2026; Contreras et al. 2026) and does not verify it directly in the same FLAMINGO snapshots used for the kSZ stacks. Appendix A shows that using satellite subhalo velocities changes the true-velocity profile by up to ~20% on large scales, but it does not test the RSD-reconstructed case, where the physical mechanism (FoG) is different. Thus the quantitative reach of the headline result is not fully protected without a direct check of satellite gas velocities.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the velocity-weighted stacked kinetic Sunyaev-Zel'dovich (kSZ) signal using the FLAMINGO hydrodynamical simulations and DESI-like mock galaxy samples (LRG, BGS, ELG). The authors derive an exact decomposition of the estimator into a mean bulk-flow term, a density-bulk flow correlation term, and a velocity-decorrelation term. They then evaluate each term for three stacking-velocity choices: true halo velocities, linearly reconstructed velocities in real space, and linearly reconstructed velocities in redshift space. They find that the two non-linear terms individually can be large but cancel substantially; with real-space linear reconstruction the non-linear terms become negligible, while including redshift-space distortions produces a scale-dependent ~10% suppression on small scales that they attribute primarily to satellite Fingers-of-God. They further show that the non-linear terms are insensitive to baryonic feedback at the 5-10% level, with gas-velocity effects below 1%, and they discuss the implications for current and future kSZ analyses, including a trade-off between smoothing scale and signal-to-noise.","tokens_in":1566,"tokens_out":1639,"duration_ms":166889,"significance":"If the results hold, this is a valuable and timely contribution to the interpretation of stacked kSZ measurements. The algebraic decomposition in Eq. (7) is exact and clarifies which physical effects can enter the estimator, and the use of realistic FLAMINGO mocks with multiple feedback variants gives the main conclusions a solid basis. The paper is careful to compute all non-linear terms directly from simulation fields rather than fitting them to the target observable: there are no free parameters tuned to the kSZ signal. The main strength is the clear demonstration that the modelling strategy must depend on how the stacking velocity is constructed, and that real-space linear reconstruction provides a useful theoretical benchmark. The sensitivity of the non-linear terms to baryonic feedback is quantified explicitly. However, the paper's headline result, the ~10% RSD-driven small-scale suppression attributed to satellite Fingers-of-God, rests on a satellite-gas assumption that is not directly verified in the same simulations, and the ELG sample is acknowledged to be under-resolved. These issues need to be addressed before the central quantitative claim can be fully accepted.","major_comments":[{"comment":"The headline small-scale suppression in the RSD case is attributed to satellite Fingers-of-God, and the decomposition that leads to this attribution uses v0 = v_h for satellite galaxies in Eq. (4). The justification is that ionized gas around satellites is largely stripped and therefore moves with the host halo, but this is supported only by citations to He et al. (2026) and Contreras et al. (2026) and is not checked in FLAMINGO, the same simulation used to produce the stacks. If a non-negligible fraction of gas retains the satellite's orbital velocity, v' in Eq. (4) contains a coherent satellite-velocity component, changing the velocity-decorrelation term Cov(p', v_stack) in Eq. (7) and thereby both the amplitude of the ~10% suppression and its satellite attribution. Appendix A tests the alternative subhalo-velocity choice only for the true-velocity stack, where it changes the profile b","section":"Section 3.3.1 / Eq. (4) / Section 4.3.2"},{"comment":"The ELG-like sample is explicitly acknowledged to be under-resolved in FLAMINGO: the adopted mean halo mass is 10^12.4 h^-1 M_sun rather than the target 10^12.2 h^-1 M_sun, and the reconstructed-velocity cross-correlation is r ~ 0.74, differing from the r ~ 0.55 obtained with Abacus HOD catalogs. Despite these caveats, Figure 6 and the reported 0.31-sigma consistency for ELGs are presented as part of the main cross-sample results. The statement that the ELG signal is consistent with the mean bulk-flow model is therefore not robust. The authors should either calibrate the ELG mock against a higher-resolution simulation or an updated HOD model, or explicitly downgrade the ELG-specific quantitative conclusions to indicative rather than quantitative.","section":"Section 3.2 / Section 3.3.2 / Section 5"}],"minor_comments":[{"comment":"No error bars or finite-volume uncertainties are shown for the simulated stacked profiles. Given that the paper makes percent-level quantitative claims, a statement about the statistical precision of the mock stacks (e.g., from bootstrap or realization scatter) would be helpful.","section":"Sections 4.1-4.4 / Figures 1, 4, 5"},{"comment":"The sentence 'One might expect RSD to simply reduce the overall amplitude of the stacked kSZ signal through the lower r' is confusing because Eq. (2) is explicitly normalized by r, and the mean bulk-flow term in Eq. (7) is independent of r. Consider rephrasing to clarify that the concern is about the raw stacked amplitude before normalization.","section":"Section 5.1.1"},{"comment":"The statement that using the centre-of-mass velocity of gas particles produces no appreciable difference should specify that this verification applies to host haloes. For satellites, the corresponding check is not provided, which is directly related to the major comment above.","section":"Section 3.3.1"},{"comment":"The ELG sample caveat should be stated more prominently, ideally in the abstract or the introduction of the cross-sample results, since the current text in Section 3.2 somewhat understates the impact on the ELG-related conclusions.","section":"Section 3.2"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the core derivation and most of the simulation analysis are sound, and the paper is likely to be an important reference for kSZ modelling. The gate for acceptance is a direct test of the satellite-gas assumption in FLAMINGO, or a clear quantitative demonstration that the RSD-induced suppression is insensitive to that assumption. The ELG-resolution issue is real but secondary; it mainly affects the cross-sample breadth of the claims. I would not reject the paper on these grounds, but the current headline statement is too strong without the requested checks."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The key new result: with the standard redshift-space linear velocity reconstruction, the non-linear terms in the stacked kSZ signal do not cancel and instead suppress the small-scale signal by ~10%, driven mainly by satellite fingers-of-God, and this suppression is almost independent of baryonic feedback. That is a concrete and practical finding for the interpretation of current and upcoming kSZ stacks.\n\nWhat the paper actually does well: it writes down the exact decomposition of the velocity-weighted estimator into a mean bulk flow, a density–bulk flow correlation, and a velocity-decorrelation term, then uses FLAMINGO with DESI-like LRG, BGS, and ELG mocks to isolate each contribution. The finding that real-space linear reconstruction reduces the non-linear terms to <2% — that non-linear velocity information is filtered out — is a nice, clean demonstration. The RSD case is the interesting one: reconstruction errors correlate with the halo environment and with the direction of motion, producing the scale-dependent suppression. The paper’s claim that baryonic feedback changes the non-linear terms by less than 5–10% (with gas velocities alone <1%) makes the effect tractable in gravity-only simulations, which is an important practical conclusion.\n\nThe main soft spot is the treatment of satellite gas velocities. The paper assigns satellites the host-halo velocity, with the justification that ionized gas around satellites is largely stripped and does not follow orbital motion (He et al. 2026; Contreras et al. 2026). That assumption is not directly checked in the same FLAMINGO snapshots used for the stacks. Appendix A shows that using subhalo velocities changes the true-velocity profile by up to ~20% on large scales, but it does not test the RSD-reconstructed case, where the FoG mechanism operates. If a non-negligible fraction of gas retains satellite orbital velocity, the velocity-decorrelation term and the satellite attribution of the ~10% suppression would change. This is a legitimate concern, but it is a fixable one: they can measure the gas velocity around satellites directly in FLAMINGO. I would ask for that check rather than treat the number as gospel.\n\nMinor caveats: the ELG sample is at the resolution limit of FLAMINGO, and the paper itself flags an unexplained numerical velocity artifact in the Jet feedback variants. The sigma values inherit covariance normalisation from companion papers, which is fine but should be clearly stated. These do not undermine the qualitative conclusions.\n\nOverall, the central argument holds up. The paper is honest, clearly written, and the main effect is real. It is aimed at anyone modelling stacked kSZ measurements, particularly for DESI and next-generation CMB experiments. It deserves a serious referee with requests to verify the satellite-gas assumption and to clarify the ELG resolution and covariance details.","headline":"Useful simulation-based result: RSD in velocity reconstruction creates a ~10% scale-dependent kSZ suppression, nearly baryon-independent, but the satellite-FoG attribution depends on an unverified stripping assumption.","tokens_in":25013,"tokens_out":5914,"would_cite":true,"duration_ms":56740,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Velocity errors bias stacked kinetic-SZ signal by 10-20 percent","keywords":["kinetic Sunyaev-Zel'dovich effect","velocity reconstruction","redshift-space distortions","Fingers-of-God","baryonic feedback","stacked signal","optical depth","non-linear velocities"],"falsifier":"A direct simulation test would compare the stacked kSZ signal computed when gas bound to satellite subhaloes keeps its own velocity instead of being forced to the host-halo velocity; a change larger than ~10 percent on small scales would falsify the correction mechanism. Alternatively, a future survey achieving sub-3-percent statistical errors on the small-scale profile could check for the predicted ~10 percent suppression (or ~5 percent with large smoothing); its absence would rule out the model.","tokens_in":24077,"feed_emoji":"🔭","tokens_out":8621,"duration_ms":70649,"temperature":0.7,"pith_summary":"This paper shows that the stacked kinetic Sunyaev-Zel'dovich (kSZ) signal—a velocity-weighted measurement of the gas around galaxies—is not simply proportional to the mean gas density. The authors decompose the signal into a dominant bulk-flow term and two non-linear corrections, and show that whether these corrections appear depends entirely on how the stacking velocity is estimated. When velocities are reconstructed from linear theory in real space (an idealized, unobservable case), the non-linear terms cancel and the signal traces the mean optical depth to within a few percent. But with realistic redshift-space distortions, reconstruction errors correlate with the halo environment and the direction of motion, producing a net suppression of about 10 percent on small scales, driven by Fingers-of-God motions of satellite galaxies. These corrections are gravitational in origin and only weakly sensitive to baryonic feedback, and they already reach the 1–2 sigma level for current CMB-galaxy stacks.","feed_headline":"Velocity errors bias stacked kinetic-SZ signal by 10-20 percent","feed_subtitle":"Small-scale suppression of 10 percent already approaches statistical significance for current CMB surveys.","key_machinery":"The key object is the exact decomposition of the velocity-weighted stacked kSZ estimator into three contributions: the mean bulk-flow term (proportional to the mean optical depth times the rms halo velocity), the density–bulk-flow correlation term, and the velocity-decorrelation term. The central identity is the linearized continuity equation used to reconstruct velocities from the smoothed galaxy density field, with the density field built either in real space or redshift space. The argument turns on how these terms couple to the stacking velocity: linear real-space reconstruction filters out the small-scale density–velocity correlations that source the non-linear terms, while redshift-spac","core_discovery":"The central discovery is the identification of the two non-linear velocity terms in the stacked kSZ estimator—the density–bulk-flow correlation and the velocity-decorrelation term—and the demonstration that both are dominated by non-linear gravitational dynamics rather than by baryonic physics. When stacking with true halo velocities, each of these terms individually reaches up to half the total signal but largely cancels, leaving a residual 10–20 percent contribution. When velocities are reconstructed linearly in real space, the reconstructed velocity carries no small-scale non-linear information, so both terms vanish and the signal traces the mean optical depth within a few percent. When r","pith_inferences":["If the satellite gas-stripping assumption is wrong—if satellites retain enough bound gas to share their orbital motion—the small-scale suppression would be larger and depend more strongly on the satellite population; this could be tested directly in simulations that resolve gas around infalling satellites.","A practical strategy for upcoming surveys might be to run two reconstructions, one aggressive for maximizing signal and one conservative for estimation, and to treat the difference as a systematic error budget.","The near-cancellation of the two non-linear terms suggests that analytic models of the stacked kSZ effect that capture only one term may be less accurate than models that ignore both; future effective models should either omit both or include both consistently.","The insensitivity to baryonic feedback implies that any measured disagreement between stacked kSZ and X-ray-based gas fractions cannot be attributed to the non-linear velocity correction, strengthening the case that the discrepancy reflects real baryonic physics."],"forward_implications":["Current measurements are consistent with the mean bulk-flow approximation at the 1–2 sigma level; next-generation CMB surveys will make the non-linear suppression statistically significant, so models must account for it.","The mean optical depth can be recovered unbiasedly either by smoothing the reconstruction to large scales (which reduces the suppression to about 5 percent) or by jointly modelling the two non-linear terms with simulations.","Because the non-linear terms are gravitational in origin, they can be calibrated with gravity-only simulations including baryon-painted densities; the baryonic contribution through gas velocities alone is below 1 percent.","The two non-linear terms must be modelled together; modelling only one introduces errors of roughly 50 percent, while neglecting both introduces errors of only 10–20 percent.","There is a direct trade-off between signal-to-noise and modelling simplicity: aggressive reconstructions boost S/N but require simulation-based modelling, while conservative reconstructions simplify interpretation."],"fun_headline_variants":["Non-linear velocities bias stacked kSZ by 10-20%","kSZ signal suppressed 10-20% by non-linear velocity terms","Velocity reconstruction choice drives kSZ bias up to 20%","Stacked kSZ: non-linear velocity effects cause 10-20% bias"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The result rests on the assumption that ionized gas around satellite galaxies is largely stripped and therefore moves with the host halo rather than with the satellite's orbital velocity; if satellites retain significant bound gas that moves with them, the predicted small-scale suppression changes.","fun_headline_variants_meta":{"raw":{"variants":["Non-linear velocities bias stacked kSZ by 10-20%","kSZ signal suppressed 10-20% by non-linear velocity terms","Velocity reconstruction choice drives kSZ bias up to 20%","Stacked kSZ: non-linear velocity effects cause 10-20% bias"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000317,"raw_usage":{"total_tokens":1656,"prompt_tokens":795,"completion_tokens":861,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":539,"completion_tokens_details":{"reasoning_tokens":795}},"tokens_in":539,"tokens_out":861,"duration_ms":7616,"temperature":1.0,"reasoning_tokens":795,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T23:41:26.925602+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct simulation test would compare the stacked kSZ signal computed when gas bound to satellite subhaloes keeps its own velocity instead of being forced to the host-halo velocity; a change larger than ~10 percent on small scales would falsify the correction mechanism. Alternatively, a future survey achieving sub-3-percent statistical errors on the small-scale profile could check for the predicted ~10 percent suppression (or ~5 percent with large smoothing); its absence would rule out the model.","supporting_citations":[],"review_version":1}