REVIEW 4 major objections 5 minor 12 references
The new CSiBORG-Manticore digital twins of the local universe reproduce the thermal Sunyaev-Zel'dovich signals of nearby clusters and their masses well enough to serve as the foundation for future field-level CMB analyses.
Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →
T0 review · deepseek-v4-flash
2026-08-03 08:42 UTC pith:7UYTEFEE
load-bearing objection A solid, well-scoped tSZ validation of BORG digital twins — the headline improvement claim holds up, but the mass-calibration pillar rests on a small matched sample with a post-hoc exclusion that needs addressable fixes before I'd call it fully robust. the 4 major comments →
Validating Digital Twins of the Local Universe with the Thermal Sunyaev-Zel'dovich Signal
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The central claim is that CSiBORG-Manticore, a suite of 50 posterior simulations driven by BORG initial conditions, is an improved generation of digital twins: it places the majority of well-known local clusters (Coma, Perseus, Shapley A3558, etc.) at significant hotspots in Planck's Compton-y map, yields an integrated-Compton-parameter–mass slope of m=1.79±0.28 (close to the self-similar 5/3), and produces halo masses that are statistically consistent with weak-lensing-calibrated eROSITA masses (slope 0.84±0.19, intercept 0.02±0.06, 0.5-sigma joint tension with one-to-one). By contrast, the earlier CSiBORG2 fails to reproduce Perseus and shows a shallower slope and offset masses. The author
What carries the argument
The central object is the BORG (Bayesian Origin Reconstruction from Galaxies) forward-modelling framework, which infers posterior initial conditions from the 2M++ galaxy catalogue and resimulates them as N-body 'digital twins'—CSiBORG-Manticore adds 50 posterior samples with high-resolution zoom-in regions. The validation machinery is a set of three tests: (i) a p_tSZ score that measures whether a simulated halo's sky position lies at a significant Compton-y hotspot relative to random positions; (ii) stacked radial tSZ profiles in mass-ranked bins; and (iii) Bayesian linear-regression fits of Y_tSZ–M and mass–mass relations that marginalise over the posterior distribution of halo masses. Hal
Load-bearing premise
The mass-calibration claims rest on the assumption that the small matched sample of clusters—chosen by strict angular and velocity cuts and then trimmed by removing Shapley A3558—is representative enough that selection does not bias the fitted slope and offset.
What would settle it
Recompute the mass–mass and Y–M scaling relations either including Shapley A3558 or with a completeness-corrected selection model; if the CBM–eROSITA agreement degrades to ≳2σ, the claimed improvement over CB2 would be an artifact of sample selection. Alternatively, verify that the 23/74 matched eROSITA clusters are a random subset of the parent population in mass, distance, and sky coverage.
If this is right
- Future field-level models of CMB secondary anisotropies can use BORG/Manticore as a prior for the local large-scale structure, enabling joint fits with Planck and other CMB data.
- The tSZ data can be used to calibrate the mass–observable relation without an external mass proxy, because the digital twins supply the mass side of the relation.
- The validation framework gives a quantitative way to score future constrained simulations against tSZ observations, complementing velocity-based tests.
- The positional accuracy of BORG twins (typical offsets <1 degree versus >10 degrees for velocity-based approaches) makes them superior anchors for object-by-object cluster studies.
- If the mass calibration holds, BORG halo masses can be used to correct the known ~30% bias in Planck tSZ mass estimates.
Where Pith is reading between the lines
- The strict matching criteria (60 arcmin, 300 km/s) likely bias the matched sample toward the most massive, best-reconstructed clusters; if this is not accounted for, the claimed 0.5-sigma agreement with eROSITA masses could overstate the true fidelity.
- A natural next test would be to apply the same p_tSZ scoring to higher-resolution tSZ maps from other experiments, which would sharpen the positional test and reveal whether the sub-degree offsets hold on smaller scales.
- The authors' suggestion to add tSZ data as a likelihood in BORG could be tested by checking whether the posterior contracts around known clusters in regions where galaxy surveys are sparse, such as the Zone of Avoidance.
- The framework could be converted into a calibration pipeline for future cluster surveys that avoids both hydrostatic and tSZ mass biases, but only if the selection function of the matched sample is modelled explicitly.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper introduces CSiBORG-Manticore (CBM), a new suite of BORG-based constrained simulations of the local Universe, and validates it against Planck tSZ maps and eROSITA X-ray cluster masses in comparison with the previous CSiBORG2 (CB2) generation. Three tests are used: per-cluster angular alignment with Compton-y hotspots (p_tSZ), stacked radial tSZ profiles, and power-law scaling relations between halo mass and tSZ/X-ray observables. The central claims are that CBM places most named clusters at tSZ hotspots, yields a Y_tSZ–M slope closer to the self-similar expectation (m=1.79±0.28 vs m=1.43±0.31), and produces halo masses consistent with weak-lensing-calibrated eROSITA masses (m=0.84±0.19, c=0.02±0.06, 0.5σ from one-to-one). The paper concludes that future field-level CMB models informed by large-scale structure should use BORG/Manticore as their foundation.
Significance. If the central claims hold, the paper provides a valuable new benchmark for constrained simulations and demonstrates a practical route to using digital twins for tSZ analyses. The introduction of CBM is a concrete advance in the BORG programme, and the validation framework (p_tSZ scoring, stacked profiles, mass-scaling fits) is well matched to the goal of object-level comparison. The manuscript provides reproducible code and data links, compares against the SLOW hydrodynamical constrained simulation, and is careful to separate the tSZ-based validation from the mass-calibration comparison. The paper explicitly acknowledges the absence of a formal Bayesian evidence comparison (Section 5.2), which is an honest statement of the current scope. The main quantitative pillar—the eROSITA mass–mass agreement—is genuinely informative if the selection effects and the exclusion of Shapley A3558 are addressed.
major comments (4)
- [Section 4.3, Table 3, Fig. 6] Shapley A3558 is a well-matched cluster (3D separation 1.4 h^-1 Mpc, f_present=0.90, p_tSZ=8.2e-4) but is excluded from the mass-scaling fits because its CBM halo mass (log M=15.24±0.12) exceeds the eROSITA weak-lensing mass (14.81±0.04) by ~0.4 dex. This is a selection on the dependent variable of the fit. The reported 0.5σ agreement for CBM and the statement that 'CBM halo masses are well calibrated' are conditional on removing the clearest mass-calibration failure. Please provide the fits with A3558 included, or a formal selection/outlier model.
- [Section 3.4, Eq. (9), Fig. 8] The matched samples are small (23/74 eROSITA, 15/56 Planck) and the matching fraction is strongly mass-dependent. The likelihood in Eq. (9) ignores the cluster selection; if the probability of passing Δθ<60′, Δcz<300 km/s and the ≥50% membership cut correlates with mass or with the tSZ/X-ray observable, the fitted slope and intercept are biased. A selection model, or at minimum a sensitivity test with looser thresholds and/or weighting by the inverse matching probability, is needed before claiming that the eROSITA agreement is not an artifact of sample selection. The paper itself notes (Section 5.2) that no Bayesian evidence comparison is performed; a selection-corrected fit would strengthen the claim.
- [Section 4.3.1, Table 2] The headline improvement in the Y-M slope (CBM m=1.79±0.28 vs CB2 m=1.43±0.31) is not statistically significant: the difference is 0.36±0.42 (~0.9σ). The lower intrinsic scatter for CBM (0.08±0.07 vs 0.22±0.10) is suggestive, but the two simulations are consistent within uncertainties for both the slope and the mass-mass relations. The claim that CBM 'yields a slope closer to the expected value' should be moderated or supported by a joint fit/tension calculation that accounts for the covariance.
- [Section 6 and Abstract] The concluding recommendation that 'future field-level models of the CMB ... should use BORG—specifically its latest incarnation as Manticore—as their foundation' is broader than the evidence presented. The paper validates cluster positions, stacked tSZ profiles, and mass scaling relations; it does not test field-level CMB modeling or compare against other possible foundations for such models. Please rephrase the conclusion to state what is actually demonstrated, and frame the CMB recommendation as a plausible implication rather than a demonstrated requirement.
minor comments (5)
- [Section 3.1] The mask used for the Compton-y photometry is described only qualitatively ('Galactic plane, point sources, pixels marked as unseen'). Please specify the mask version and the exact point-source mask used, since p_tSZ values depend directly on the random-pointing null distribution.
- [Section 2.5 / Table 1] The text mentions Shapley (A3562) in the discussion of Fig. 2, but A3562 is not listed in Table 1. Clarify whether it is part of the selected cluster sample or just shown as a nearby reference.
- [Section 4.1, Fig. 2] The figure caption states that the red dot marks the local peak of the tSZ map, but the algorithm for identifying that peak is not described in Section 3.1 or elsewhere. A brief description would help reproducibility.
- [References] Some references in the text are abbreviated inconsistently (e.g., 'Collaboration 2016' without the Planck prefix, and 'P. Collaboration' in the bibliography). Please standardize to 'Planck Collaboration' throughout.
- [Section 4.3 / Table 2] The intercept for the eROSITA CB2 L_X–M relation (c=1.46±0.08) looks surprising at first glance because the text does not state the pivot of the X-ray luminosity variable in the same place as the mass pivot. Please state the units and pivot explicitly in the table caption or in Section 3.6.
Circularity Check
No significant circularity; validation uses external tSZ and weak-lensing-calibrated masses, with only mild self-citation.
full rationale
The paper's central validations are performed against data that are not used to construct the simulations: the Planck Compton-y map and the eROSITA cluster masses, the latter calibrated via DES Y3 weak lensing (Grandis et al. 2024). The BORG/Manticore initial conditions are inferred from the 2M++ galaxy density field, not from tSZ or X-ray data, so the per-cluster p_tSZ tests, stacked radial profiles, and mass-scaling fits are external benchmarks rather than re-statements of the simulation inputs. The fitted Y_tSZ-M and mass-mass relations compare simulation halo masses to independent catalog observables; the self-similar slope (m=5/3) and one-to-one relation are external expectations, not fit-derived targets. Self-citations appear (McAlpine et al. 2025 for the Manticore initial conditions and Stiskalek et al. 2025 for velocity-field comparisons), and these are used to motivate the CBM suite and support the broader claim of BORG's fidelity, but the tSZ positional, stacked-profile, and eROSITA mass evidence presented here is newly derived and does not reduce to those citations. The post-hoc exclusion of Shapley A3558 and the small matched samples are legitimate selection-robustness concerns, but they are not circularity: the remaining relation is not equivalent to the simulation inputs by construction, and no parameter of the simulations is fitted to the validation data. No uniqueness theorem, ansatz-by-citation, or renaming-as-unification pattern is present. Hence the derivation chain is substantially self-contained externally; the only mild circularity-adjacent feature is reliance on the authors' own prior work for the simulation suite's standing, which is not load-bearing for the independent tSZ validation.
Axiom & Free-Parameter Ledger
free parameters (6)
- Scaling relation slope m and intercept c for Y_tSZ-M, L_X-M, and mass-mass fits =
CB2 Planck Y: 1.43±0.31, -1.51±0.13; CBM Planck Y: 1.79±0.28, -1.73±0.15; CBM eROSITA M-M: 0.84±0.19, 0.02±0.06; CBM Pla
- Intrinsic scatter σ_int =
CBM Planck Y: 0.08±0.07; CB2 Planck Y: 0.22±0.10; CBM eROSITA M-M: 0.14±0.04
- Matching thresholds Δθ < 60 arcmin, Δcz < 300 km/s =
60 arcmin; 300 km/s
- Halo association parameters: DBSCAN ϵ=1.75 h^-1 Mpc, min membership 9, 0.3 dex mass outlier cut
- Aperture radius for Compton-y photometry = 2θ500c
- Post-hoc exclusion of Shapley A3558 from mass-scaling fits =
excluded
axioms (7)
- domain assumption ΛCDM cosmology with Planck 2020 parameters for CB2 and DES Y3 3x2pt parameters for CBM
- domain assumption BORG forward model (2M++ galaxy density, Poisson likelihood, bias and selection modeling) correctly samples the posterior of initial conditions
- domain assumption Planck PR4 NILC Compton-y map (McCarthy & Hill 2024) traces true integrated electron pressure after CIB deprojection
- domain assumption eROSITA DR1 X-ray masses are unbiased after DES-Y3 weak-lensing calibration
- domain assumption Dark-matter-only BORG halo masses can be compared directly to observed baryonic masses without baryon-feedback correction
- domain assumption Self-similar scalings Y ∝ M^(5/3) and L_X ∝ M^(4/3) are the correct external expectations
- domain assumption Y_5R500 = 1.81 Y_tSZ_500c conversion from the universal pressure profile (Arnaud et al. 2010)
read the original abstract
The thermal Sunyaev-Zel'dovich (tSZ) effect provides a powerful probe of the thermal pressure of ionised gas in galaxy clusters and the cosmic web; constrained simulations reconstruct the mass and velocity fields of the local Universe. We explore how these two may be mutually informative: the tSZ signal provides a benchmark for assessing the fidelity of constrained simulations, and constrained simulations contribute information on the positions, total masses and density profiles of cosmic web structures for use in tSZ studies. We focus on cluster predictions in the Bayesian Origin Reconstruction from Galaxies (BORG) paradigm, introducing CSiBORG-Manticore, a new state-of-the-art suite of digital twins -- data-constrained posterior simulations whose initial conditions are inferred via Bayesian forward modelling. We develop a framework for scoring constrained simulations on their ability to match measured Planck Compton-$y$ maps around clusters, and use it to demonstrate improvement from previous BORG reconstructions. We further validate halo masses against weak-lensing-calibrated X-ray masses from eROSITA. We also show how high-fidelity digital twins offer a practical route to extracting additional information from tSZ data through a novel calibration of the mass-observable relation, and provide a complementary framework to purely statistical analyses of Compton-$y$ maps. This paves the way for integrating the large-scale structure information inherent in constrained simulations into the study of CMB secondary anisotropies.
Figures
Reference graph
Works this paper leans on
-
[1]
R. A. Sunyaev and Y. B. Zeldovich, Astrophysics and Space Science7, 3 (1970). R. A. Sunyaev and Y. B. Zeldovich, Comments on Astrophysics and Space Physics4, 173 (1972). T. Mroczkowski, D. Nagai, K. Basu, J. Chluba, J. Sayers, R. Adam, E. Churazov, A. Crites, L. Di Mascolo, D. Eckert, J. Macias-Perez, F. Mayet, L. Perotto, E. Pointecouteau, C. Romero, F. ...
Pith/arXiv arXiv 1970
-
[2]
Barreiro, J. G. Bartlett,et al., A&A594, A24 (2016a), arXiv:1502.01597 [astro-ph.CO]. T. de Haan, B. A. Benson, L. E. Bleem, S. W. Allen, D. E
-
[3]
Applegate, M. L. N. Ashby, M. Bautz, M. Bayliss, S. Bocquet, M. Brodwin,et al., ApJ832, 95 (2016), arXiv:1603.06522 [astro-ph.CO]. S. Bocquet, J. P. Dietrich, T. Schrabback, L. E. Bleem, M. Klein, S. W. Allen, D. E. Applegate, M. L. N. Ashby, M. Bautz, M. Bayliss,et al., ApJ878, 55 (2019), arXiv:1812.01679 [astro-ph.CO]. M. Hilton, C. Sif´ on, S. Naess, M...
Pith/arXiv arXiv 2016
-
[4]
Shan, L. van Waerbeke, and A. H. Wright, A&A660, A27 (2022), arXiv:2109.04458 [astro-ph.CO]. N. Koukoufilippas, D. Alonso, M. Bilicki, and J. A. Peacock, Monthly Notices of the Royal Astronomical Society (2022), arXiv:1909.09102 [astro-ph.CO]. A. La Posta, D. Alonso, N. E. Chisari, T. Ferreira, and C. Garc ´ ıa-Garc ´ ıa, Physical Review D112, 043525 (202...
Pith/arXiv arXiv 2022
-
[5]
Hudson, D. J. Bartlett, and H. M. Courtois, MNRAS (2025), 10.1093/mnras/staf1960, arXiv:2502.00121 [astro-ph.CO]. S. McAlpine, J. Jasche, M. Ata, G. Lavaux, R. Stiskalek, C. S. Frenk, and A. Jenkins, MNRAS540, 716 (2025), arXiv:2505.10682 [astro-ph.CO]. T. M. C. Abbott, M. Aguena, A. Alarcon, S. Allam, O. Alves, A. Amon, F. Andrade-Oliveira, J. Annis, S. ...
arXiv 2025
-
[6]
Banday, R. B. Barreiro,et al., A&A536, A8 (2011), arXiv:1101.2024 [astro-ph.CO]. M. Sereno, G. Covone, L. Izzo, S. Ettori, J. Coupon, and M. Lieu, MNRAS472, 1946 (2017), arXiv:1703.06886 [astro-ph.CO]. Planck Collaboration, N. Aghanim, Y. Akrami, M. Ashdown, J. Aumont, C. Baccigalupi, M. Ballardini, A. J. Banday, R. B
Pith/arXiv arXiv 2011
-
[7]
Bartolo,et al., A&A641, A6 (2020b), arXiv:1807.06209 [astro-ph.CO]
Barreiro, N. Bartolo,et al., A&A641, A6 (2020b), arXiv:1807.06209 [astro-ph.CO]. E. Bulbul, A. Liu, M. Kluge, X. Zhang, J. S. Sanders, Y. E
-
[8]
Ramos-Ceja, J. Comparat, F. Balzer, K. B¨ ockmann, M. Br¨ uggen, N. Clerc, K. Dennerl, K. Dolag, M. Freyberg, S. Grandis, D. Gruen, F. Kleinebreil, S. Krippendorf, G. Lamer, A. Merloni, K. Migkas, K. Nandra, F. Pacaud, P. Predehl, T. H. Reiprich, T. Schrabback, A. Veronica, J. Weller, and S. Zelmer, A&A685, A106 (2024), arXiv:2402.08452 [astro-ph.CO]. P. ...
Pith/arXiv arXiv 2024
-
[9]
Hansen, M. Reinecke, and M. Bartelmann, ApJ622, 759 (2005), arXiv:astro-ph/0409513 [astro-ph]. A. Zonca, L. Singer, D. Lenz, M. Reinecke, C. Rosset, E. Hivon, and K. Gorski, The Journal of Open Source Software4, 1298 (2019). S. Pfeifer, A. Valade, S. Gottl¨ ober, Y. Hoffman, N. I. Libeskind, and W. A. Hellwing, MNRAS523, 5985 (2023), arXiv:2305.05694 [ast...
Pith/arXiv arXiv 2005
-
[10]
Crawford, R. M. Johnstone, and K. Iwasawa, MNRAS366, 417 (2006), arXiv:astro-ph/0510476 [astro-ph]. C. P. Haines, G. Busarello, P. Merluzzi, K. A. Pimbblet, F. P. A
Pith/arXiv arXiv 2006
-
[11]
Vogt, M. A. Dopita, A. Mercurio, A. Grado, and L. Limatola, MNRAS481, 1055 (2018), arXiv:1807.11525 [astro-ph.CO]. Y. Higuchi, N. Okabe, P. Merluzzi, C. P. Haines, G. Busarello, A. Grado, and A. Mercurio, MNRAS497, 52 (2020), arXiv:2006.08130 [astro-ph.CO]. M. Rossetti, S. Ghizzardi, S. Molendi, and A. Finoguenov, A&A 463, 839 (2007), arXiv:astro-ph/06110...
Pith/arXiv arXiv 2018
-
[12]
Tully, MNRAS495, 4463 (2020), arXiv:2005.08074 [astro-ph.CO]. L. Doeser, M. Ata, and J. Jasche, MNRAS542, 1403 (2025), arXiv:2502.13243 [astro-ph.CO]. B. Horowitz and A. E. Bayer, arXiv e-prints , arXiv:2510.26851 (2025), arXiv:2510.26851 [astro-ph.IM]. N. Battaglia, J. R. Bond, C. Pfrommer, and J. L. Sievers, ApJ 758, 75 (2012b), arXiv:1109.3711 [astro-p...
Pith/arXiv arXiv 2020
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.