{"id":"b137c4a6-df45-42e4-b9f2-0cdac06be212","arxiv_id":"2608.00164","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Stacked tSZ measurements around DESI LRGs show FLAMINGO simulations overpredict the thermal pressure of galaxy groups by about a factor of two, including the feedback variant that matches kSZ gas densities.","lead":"Using ACT microwave maps and weak lensing data, the authors measure the thermal pressure of hot gas around 10^13–10^14 solar-mass galaxies and compare it to the FLAMINGO simulations. The simulations predict roughly twice the observed pressure, suggesting the simulated gas is too hot or is missing extra non-thermal support.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 1.6′ SED separation is underdetermined (3 channels, 4 free parameters); the claimed upper envelope of y may depend on fixing T_dust=24 K and β=1.7, so the factor-of-two deficit could shrink under a broader dust/radio SED family.","rationale":"The reader's weakest-assumption diagnosis is in the right place: the SED decomposition is the most fragile link. I agree that the flat-spectrum-radio/dust-shape worry is real, but I would sharpen it: the deeper issue is that the decomposition is underdetermined, so the 'conservative' upper limit on y is not a measured upper envelope of the full physical SED family. The paper's own tests are reassuring in that plausible perturbations of the fiducial model move y downward, making the deficit larger, and the fiducial model is deliberately conservative. However, those tests do not sample the full T_dust–β degeneracy or include multi-temperature dust/flat-spectrum radio, which are both physically motivated for LRGs at z≈0.6. Since the claimed discrepancy is only ~2× and dust dominates the lowest-mass bin, a dedicated test with a broader SED family is the single check that would settle whether the deficit is real or an artifact of the assumed foreground shapes. The paper otherwise makes a careful case: the GGL-calibrated like-with-like comparison is well designed, the simulation variants bracket the mass/satellite systematics, and the kSZ-based density argument adds independent support for the thermal interpretation. Maintaining the conditional verdict is appropriate until the SED-family test is done.","tokens_in":35589,"tokens_out":13657,"duration_ms":172174,"concrete_test":"Re-run the 1.6′ SED fits for all 16 mass/redshift bins with T_dust sampled over 15–40 K, β over 1.0–2.5, an optional second dust component at T=40 K, and a flat-spectrum radio component with amplitude prior set by the 3σ VLASS non-detection at 3 GHz. Compare the resulting 95% upper limit on y with the fgas-8σ FLAMINGO prediction bin-by-bin; if any low-mass bin's upper limit reaches or exceeds the simulation, the factor-of-two deficit is not robust to the SED family, whereas if all remain below it, the concern is retired.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the 1.6′ SED separation, not the stacking or GGL matching. In §3.3 the model has three ACT photometric measurements (90, 150, 220 GHz) and four free parameters (log y, log A_dust, log A_radio, α), with the dust shape fixed to T_dust=24 K and β=1.7. The authors acknowledge the problem is “highly degenerate” and deliberately use the upper end of the y posterior; Fig. 4 shows dust contributes >60% of the 150 GHz signal in the lowest-mass bin. The Appendix B robustness tests vary β with T fixed, omit radio, or add a VLASS prior, but they do not jointly vary T_dust and β, nor include a second dust temperature component or a flat-spectrum radio population — both physically plausible for z≈0.6 LRGs. With three data points and four parameters, the posterior upper limit is set partly by the prior volume and by the fixed SED family, not by an independently measured component separation. If the true dust SED is warmer (T_dust≈30–40 K with corresponding β) or if flat-spectrum radio sources below the VLASS threshold contribute at 90/150 GHz, the positive contamination can trade against the negative tSZ decrement and raise the allowed y. Because the claimed deficit is only ~2×, even a modest SED-driven shift in y is enough to erode it. Thus the central claim rests on the assumed SED family being exhaustive rather than merely self-consistent.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents new stacked thermal Sunyaev-Zel'dovich (tSZ) measurements around DESI photometric LRGs using ACT DR6 single-channel maps, together with galaxy-galaxy lensing (GGL) measurements from HSC. The authors model the stacked 90/150/220 GHz photometry as tSZ plus dust plus radio emission, and compare the inferred Compton-y signal with forward-modeled predictions from the FLAMINGO simulations, with simulated galaxy selections calibrated to match the observed GGL profiles. The central claim is that current hydrodynamical simulations overpredict the tSZ signal of group-mass halos by roughly a factor of two at apertures ≲3′, and that even the strongest-feedback FLAMINGO variant, which reproduces the kSZ-inferred gas density, still overpredicts the thermal pressure. The authors interpret this as evidence for missing non-thermal pressure support or significant departures from hydrostatic equilibrium in the simulations.","tokens_in":35994,"tokens_out":5216,"duration_ms":70628,"significance":"If the central result holds, this is a significant step toward a coherent observational picture of the thermodynamic state of gas in galaxy groups: it combines a new foreground-modeled tSZ measurement with GGL-calibrated simulation comparisons, and it connects the tSZ deficit to the existing kSZ-based evidence for gas depletion. The paper is careful in several respects: the tSZ signal is isolated via galaxy-informed SED modeling rather than blind map-level component separation; the simulation comparison is like-with-like, forward-modeling miscentering, satellites, masking, and two-halo contributions; and many robustness tests are reported (free β, radio omitted, VLASS prior, cluster-mask variants, extreme simulation selections). The authors also make conservative choices, e.g., focusing on the upper end of the Compton-y posterior and using a stricter cluster mask in the simulations. These strengths make the claimed deficit credible within the adopted model family. However, as detailed below, the central claim depends on the assumed dust/radio SED family being sufficiently broad, and the paper does not yet provide a fully quantitative significance for the factor-of-two deficit.","major_comments":[{"comment":"The load-bearing SED separation is underdetermined: three ACT channels are fit with four free parameters (log10 y, log10 A_dust, log10 A_radio, α) plus fixed T_dust=24 K and β=1.7. The robustness tests in Appendix B vary β with T fixed, omit radio, or add a VLASS prior, but they do not jointly vary T_dust and β, include a second dust temperature component, or add a flat-spectrum radio population. Since dust contributes >60% of the 150 GHz signal in the lowest-mass bin (Fig. 4), the posterior upper envelope of y—which the paper deliberately adopts—may be set by the fixed SED family rather than by the data. Warmer dust or flat-spectrum radio below the VLASS threshold could partially fill the tSZ decrement and raise the allowed y, shrinking the claimed factor-of-two deficit. Please quantify this: e.g., fit with T_dust in a physically plausible range (say 20–40 K) with β free, and/or with an","section":"§3.3, Fig. 4, Appendix B"},{"comment":"The central claim is that the strongest-feedback FLAMINGO simulation overpredicts the observed tSZ signal by about a factor of two, yet the paper refrains from quoting a formal significance. Because the statement is the headline result, the manuscript should provide a quantitative summary of the deficit—e.g., per-bin posterior distributions of y_data/y_sim, a combined probability that y_data < y_sim, and an explicit accounting of the systematic contributions (SED, GGL mass calibration, cluster masking) that dominate over statistical errors. Currently the reader cannot assess whether the factor of two is a 2σ effect or a 10σ effect, nor how the systematic uncertainties are folded into the claim.","section":"§5, Fig. 6"},{"comment":"It is not fully clear whether the GGL measurements and the tSZ stacking are performed on the same galaxy samples. GGL uses HSC Y3, whose footprint is smaller than the ACT-DESI overlap; the tSZ stacking appears to use the full ACT-DESI overlap. If the GGL-calibrated simulation selection is derived from a subset of galaxies that differs from the tSZ sample (e.g., in stellar mass distribution, depth, or environment), the like-with-like comparison could be biased. Please state explicitly whether the tSZ analysis is restricted to the HSC overlap for the GGL-calibrated bins, or demonstrate that sample differences between the footprints are negligible for the tSZ signal.","section":"§4.3, Appendix D"},{"comment":"The interpretation that the tSZ deficit reflects an overpredicted gas temperature (rather than gas density) relies on the external kSZ result that the strongest-feedback FLAMINGO simulation reproduces the gas density of the same galaxy sample. The paper should propagate the uncertainty of this kSZ anchor into the temperature-deficit claim. For example, if the kSZ normalization is uncertain at the 10–20% level, does this change the required non-thermal pressure fraction? A quantitative propagation would strengthen the physical conclusion and separate it from the robust-but-less-interpreted statement that simulations overpredict tSZ.","section":"§6.1"}],"minor_comments":[{"comment":"The sentence 'We therefore fix T_dust = 24 K throughout this work, without loss of model flexibility' is too strong: β and T_dust are degenerate but not perfectly so, and the Appendix B results show that the inferred y can shift when the dust model is changed. Please rephrase to 'without substantial loss of model flexibility within the adopted dust family' or similar.","section":"§3.2"},{"comment":"The gray bands for the CIB+dβ deprojected maps are described in the caption but are visually dense and sometimes confused with statistical error bars. Consider labeling the band directly in the figure or using a hatched region, and clarify in the caption that the band spans the range of assumed dust SEDs, not a statistical uncertainty.","section":"Fig. 3 / Fig. 5"},{"comment":"There is no explicit data/code availability statement. Given the reproducibility value of the SED-fitting and forward-modeling pipeline, please add one (or state that code will be released upon acceptance).","section":"General"},{"comment":"The notation 'f_gas −8σ' is used without defining whether σ is the observational uncertainty on the mean gas fraction relation. It is clear later, but a one-line definition at first use would help.","section":"§4.4"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses an important question and the authors have clearly invested in robustness tests and careful forward modeling. The main risk is the SED separation: the central factor-of-two claim could erode under a broader but physically plausible dust/radio model family. I recommend major revision with emphasis on (i) explicitly quantifying the SED-model dependence of the upper-envelope y, and (ii) providing a quantitative significance for the deficit. If those can be addressed, the paper would be a strong contribution. No concerns about novelty or scope; the fit to a cosmology/astrojournal is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a solid, careful measurement paper. The key new result—that the FLAMINGO variant that matches the kSZ-inferred gas density still overpredicts the stacked tSZ signal by roughly a factor of two at small apertures—is worth taking seriously. The combination of galaxy-redshift-informed SED decomposition on ACT DR6 maps with GGL-calibrated forward modeling is a genuine improvement over previous work; the kSZ anchor is the right way to isolate the pressure question.\n\nThe robustness section is genuinely good. They vary the radio treatment, free beta, drop the radio component, switch to a radio-quiet subsample, change the cluster masking, and even use absurdly narrow mass distributions. In each case the deficit persists or grows. Their honesty in refusing to quote a formal significance because systematics dominate is exactly the right call.\n\nNow the soft spots. The load-bearing step is the 1.6' SED decomposition, and there the worry is not the stacking or GGL calibration but the assumed dust/radio SED family. With three ACT channels and four free parameters, the fit is necessarily degenerate. They fix T_dust=24K and beta=1.7 from a radio-quiet subsample and from literature, which is reasonable, and freeing beta shifts y down, making the deficit larger. But they do not jointly vary T_dust and beta, and they do not include a second dust temperature component or a flat-spectrum radio population below the VLASS threshold. Those are physically plausible for z~0.6 LRGs, and if real they would raise the allowed y and shrink the deficit. This does not kill the result—the conservatism of the analysis helps—but it is a genuine hole in the robustness argument.\n\nThe second soft spot: the broad claim about 'current hydrodynamical simulations' rests on a single suite. FLAMINGO is well calibrated, and the kSZ matching is a strong point, but the headline should be scoped to FLAMINGO until another simulation is forward-modeled the same way. Third: no data products or code are released, which limits independent verification of the SED fitting and the GGL selection.\n\nWho is this for: anyone working on baryon feedback, tSZ-based cluster cosmology, or forward-modeling of CMB secondaries. It deserves a serious referee. Recommend peer review, and ask the authors to explore the extended dust/radio SED space and release the pipeline.","headline":"Credible, carefully guarded evidence for a ~2x thermal pressure deficit at group masses, but the SED separation at 1.6' needs a broader dust/radio model before the headline is bulletproof.","tokens_in":36498,"tokens_out":2765,"would_cite":true,"duration_ms":34691,"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":"Current hydrodynamical simulations overpredict the thermal pressure of galaxy groups by roughly a factor of two, pointing to missing non-thermal pressure support or departures from hydrostatic equilibrium.","keywords":["thermal Sunyaev-Zel'dovich effect","galaxy groups","intracluster medium","galaxy-galaxy lensing","hydrodynamical simulations","baryon feedback","non-thermal pressure support","Compton y parameter"],"falsifier":"Measure the tSZ signal around the same DESI LRG sample with a CMB experiment that has more frequency channels (e.g., six channels from 30 to 280 GHz) and check whether the derived Compton y at 1.6' changes by more than the quoted statistical uncertainty; alternatively, obtain spatially resolved X-ray spectra of the same groups to directly measure gas temperature and non-thermal pressure, and see whether the observed temperature is indeed about half of the simulated prediction.","tokens_in":35466,"feed_emoji":"🌌","tokens_out":4343,"duration_ms":47005,"temperature":0.7,"pith_summary":"This paper stacks the thermal Sunyaev-Zel'dovich (tSZ) signal around a large sample of DESI Luminous Red Galaxies to measure the thermal pressure of gas in galaxy groups, and compares these measurements to state-of-the-art hydrodynamical simulations (FLAMINGO). By also measuring galaxy-galaxy lensing for the same galaxy samples, the authors calibrate the halo masses and satellite fractions in the simulations like-with-like, removing a major source of ambiguity. They find that the fiducial simulation overpredicts the stacked tSZ signal at small scales, and even the simulation variant with the strongest gas expulsion—which already reproduces the gas density inferred from kinetic SZ measurements—overpredicts the thermal pressure by about a factor of two. Because the strong-feedback simulation already matches the gas density, the authors argue the remaining deficit is not due to gas depletion but points to the simulated gas being too hot, requiring either significant non-thermal pressure support or departures from hydrostatic equilibrium.","feed_headline":"Simulations double the thermal pressure of galaxy groups","feed_subtitle":"Stacked tSZ and lensing data show feedback alone can't explain the missing pressure; non-thermal support may be needed.","key_machinery":"The central observable is the Compton y parameter, which is proportional to the line-of-sight integral of electron pressure (n_e T_e). The measurement pipeline uses compensated aperture photometry (CAP) to stack ACT single-channel temperature maps, and a multi-frequency SED model with three components—tSZ, a modified black-body dust emission (T_dust=24 K, beta=1.7), and a power-law radio component—to isolate the tSZ signal. For the simulation comparison, galaxy-galaxy lensing (GGL) profiles are used to select a sample of simulated galaxies that matches the observed halo mass distribution and satellite fraction, and the tSZ and GGL profiles are forward-modeled from the simulations, accounting","core_discovery":"The authors measure the tSZ effect around DESI LRGs using ACT single-channel maps, modeling the stacked SED as a combination of tSZ, dust, and radio emission, and compare the resulting Compton y profiles with forward-modeled predictions from the FLAMINGO simulations. After calibrating the simulated galaxy selection with galaxy-galaxy lensing, they find that the fiducial FLAMINGO simulation overpredicts the tSZ signal by roughly a factor of two or more across 10^13–10^14 Msun and z=0.4–1, and the strongest feedback variant, which best matches the kSZ-inferred gas density, still overpredicts it by about 2x at 1.6' apertures. The paper concludes that the deficit in thermal pressure cannot be ex","pith_inferences":["If non-thermal pressure support is the cause, this would affect not only group gas fractions but also cluster mass calibration and the interpretation of cluster counts and tSZ power spectrum measurements in cosmological analyses.","The galaxy-informed SED modeling approach used here could be extended to other galaxy samples and redshift ranges, especially with upcoming multi-frequency CMB experiments (e.g., Simons Observatory) that can better break the tSZ-dust-radio degeneracy.","The factor-of-two temperature overprediction in simulations may also imply that the gas in groups is more multiphase or clumpy than simulated, or that the energy injection from feedback is more efficiently converted into non-thermal forms than currently modeled."],"forward_implications":["If the deficit is real, the thermal pressure of gas in galaxy groups is about half of what current simulations predict, so the gas is either supported by non-thermal pressure (turbulence, cosmic rays, magnetic fields) or is not in hydrostatic equilibrium.","Additional gas depletion alone cannot resolve the discrepancy because the strongest feedback simulation already matches the kSZ-inferred gas density, so future simulation development must include additional physics rather than merely stronger feedback.","The tSZ power spectrum at small scales is expected to be suppressed relative to simulation predictions, potentially explaining observed deficits at high multipoles.","X-ray measurements of the same galaxy groups could directly test the predicted temperature deficit by measuring gas temperature and non-thermal pressure support.","Mass calibration of groups and clusters using X-ray or SZ observables may be biased if non-thermal pressure support is significant and unaccounted for in hydrostatic equilibrium analyses."],"fun_headline_variants":["Galaxy groups show half the thermal pressure simulations predict","Missing pressure in galaxy groups: simulations overpredict by 2x","Thermal pressure deficit in galaxy groups defies feedback models","Simulations overpredict group gas pressure even with max feedback","Galaxy groups: thermal pressure only half of simulations' value"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The three-channel SED decomposition assumes the stacked signal is exactly tSZ plus a single modified-black-body dust component (T_dust=24 K, beta=1.7) plus a single power-law radio component; if the true dust or radio SEDs differ—especially flat-spectrum radio sources not detected by VLASS—the inferred Compton y, and thus the pressure deficit, could shift.","fun_headline_variants_meta":{"raw":{"variants":["Galaxy groups show half the thermal pressure simulations predict","Missing pressure in galaxy groups: simulations overpredict by 2x","Thermal pressure deficit in galaxy groups defies feedback models","Simulations overpredict group gas pressure even with max feedback","Galaxy groups: thermal pressure only half of simulations' value"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000753,"raw_usage":{"total_tokens":3254,"prompt_tokens":876,"completion_tokens":2378,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":620,"completion_tokens_details":{"reasoning_tokens":2295}},"tokens_in":620,"tokens_out":2378,"duration_ms":15260,"temperature":1.0,"reasoning_tokens":2295,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T01:06:01.998914+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the tSZ signal around the same DESI LRG sample with a CMB experiment that has more frequency channels (e.g., six channels from 30 to 280 GHz) and check whether the derived Compton y at 1.6' changes by more than the quoted statistical uncertainty; alternatively, obtain spatially resolved X-ray spectra of the same groups to directly measure gas temperature and non-thermal pressure, and see whether the observed temperature is indeed about half of the simulated prediction.","supporting_citations":[],"review_version":1}