{"id":"3d758d76-368f-40c5-9db2-052da9466899","arxiv_id":"2509.03798","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A 13,000 square degree weak lensing mass map, the largest to date, with a first demonstration of filament detection from lensing alone.","lead":"Weak lensing from 270 million galaxies in two DECam surveys was combined into the largest cosmic mass map yet, covering 13,000 square degrees. The map also yields a first, preliminary detection of cosmic filaments, validated by the positions of galaxy clusters.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Filament 'first detection' lacks a noise-only null test or significance estimate; the quoted ~0.5 S/N excess of filament pixels is expected for ridge pixels by construction, and the cluster-alignment validation is reported without a p-value.","rationale":"The reader's CONDITIONAL verdict is appropriate. The central map claim is credible: the methods are standard, validation on simulations is presented, and systematic tests on data show no strong contamination. The weakest point is the filament detection, which is a headline scientific claim but is supported only by a tautological S/N statement and an alignment plot without a significance estimate. My stress test sharpens the reader's concern by noting that the ~0.5 S/N excess is expected for any pixels selected as ridges, and by questioning the use of ACT DR6 randoms for the Planck cluster comparison. A noise-only null test is the decisive check: if identical pipeline output on randomized ellipticities produces comparable filament-cluster alignment, then the 'first detection' claim fails. The map itself would remain valuable, so REJECT is too strong; the appropriate recommendation is to keep the conditional acceptance pending the null test and map release. No change to the reader's verdict is needed.","tokens_in":14466,"tokens_out":5997,"duration_ms":64701,"concrete_test":"Generate at least 100 noise-only shear maps by randomly rotating the observed galaxy ellipticities while preserving positions, weights, responses, and the mask. Run the identical Dante Wiener reconstruction and SCONCE filament-finding at NSIDE=128 on each realization. Recompute the SZ cluster-to-nearest-filament distance distributions using area-matched randoms separately for Planck and ACT DR6 footprints, and compare the data against the noise-only ensemble (e.g., Kolmogorov-Smirnov statistic or median distance). Also record the distribution of filament-pixel S/N excess in the noise maps. If the data alignment is not significantly stronger than the noise ensemble at p<0.01, the filament detection claim is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The map construction is well supported by simulation validation and systematic null tests, so the 'largest map' claim is not the weak point. The load-bearing weakness is the filament detection claim ('detecting cosmic filaments directly from the mass map for the first time'). The evidence presented cannot distinguish real filaments from ridges induced by shape noise, smoothing, and the Gaussian Wiener prior. The Wiener MAP is dominated by shape noise (RMS S/N ~1), and SCONCE is applied directly to this map; by construction it traces high-density ridges. Therefore the statement that filament pixels have ~0.5 higher S/N than the rest of the map is a property of the ridge finder, not an independent signal detection. The only external validation, Fig. 4, shows distance-to-filament distributions for Planck and ACT DR6 clusters but provides no p-value or significance level. Moreover, both cluster samples are compared to ACT DR6 randoms, which is not an area-matched null for the all-sky Planck sample. The paper explicitly defers a systematic assessment of masking effects near survey boundaries. Without a noise-only or null-map control, the claimed first filament detection is not quantitatively established, even though the map itself may be sound.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a weak lensing convergence map covering ~13,000 deg^2 by combining DECADE and DES Y3 shear catalogs (270 million galaxies), using both Kaiser-Squires (KS) and Wiener filter reconstructions on the sphere. The map is validated via mock simulations and null tests against observing-condition systematics. As a scientific application, the authors apply the SCONCE ridge finder to the Wiener MAP to identify filamentary structures and report that SZ-selected clusters from Planck and ACT DR6 preferentially lie near these filaments. The main claims are (i) the largest galaxy weak lensing mass map to date and (ii) the first detection of cosmic filaments directly from a weak lensing mass map.","tokens_in":14768,"tokens_out":1918,"duration_ms":23149,"significance":"If fully established, the map itself is a valuable community resource: it triples the area of existing DES Y3 lensing maps while maintaining consistency between the two DECam-based datasets. The validation on mocks and null tests follows standard practice and supports the map construction. The filament detection, however, is a more ambitious claim: weak lensing maps are shape-noise dominated, and the authors themselves note the map has S/N rms ~1 and filament pixels only ~0.5 higher S/N than the rest of the map. The current evidence falls short of a quantitative detection: there is no noise-only null test, no stated significance for the cluster-filament alignment, and masking effects near survey boundaries are deferred. The map claim is sound; the filament claim, as presented, is not yet load-bearing evidence for a 'first detection.'","major_comments":[{"comment":"The paper states that 'filament pixels have an S/N about 0.5 higher than the rest of the map' as evidence of signal. This statement is not a detection statistic: SCONCE is a ridge finder that, by construction, selects locally overdense pixels from a heavily smoothed, Wiener-filtered map that is dominated by shape noise. A higher S/N on ridges is expected even for pure noise. A quantitative claim of filament detection requires a null test, e.g., running SCONCE on maps built from randomly rotated galaxy ellipticities (or on Wiener reconstructions of noise-only mocks) and comparing the number, length, and S/N distribution of ridges. Without such a control, the reported 0.5 S/N excess cannot be interpreted as a detection of physical filaments.","section":"Structures in the reconstructed maps (Fig. 3 and surrounding text)"},{"comment":"The cluster-filament alignment is shown as two distance distributions compared to 'ACT DR6 randoms,' but no p-value or significance is quoted. Moreover, the random sample is not area-matched to the Planck cluster footprint: Planck SZ clusters cover the full sky, while ACT DR6 covers a smaller region, and the randoms are described only as 'ACT DR6 randoms.' The comparison should use an area-matched random distribution (separately for Planck and ACT) and report a two-sample test statistic (e.g., Kolmogorov-Smirnov or permutation p-value). As it stands, the visual 'preferential alignment' is not quantified and could be affected by footprint differences and mask geometry.","section":"Fig. 4 and cluster-filament validation"},{"comment":"The paper explicitly defers 'a systematic assessment of this impact' of masking near survey boundaries to future work. Given that DECADE is a combination of many programs with inhomogeneous depth and coverage, masked or low-density regions can imprint coherent patterns in the Wiener MAP, which SCONCE may trace as spurious ridges. The absence of any masking or boundary null test weakens the filament claim, even though the map itself may be unaffected at the level of the presented systematics tests. A boundary-masked noise realization, or an analysis restricted to the well-covered DES footprint, would be a minimal check.","section":"Masking and survey-boundary effects (end of 'Structures in the reconstructed maps')"}],"minor_comments":[{"comment":"Typos in the extracted text ('W eak', 'DA T A', 'despitetheir') should be corrected. The phrase 'tank-shaped grey area' in the Fig. 1 caption is unclear; presumably 'blank-shaped' or 'masked region' is meant.","section":"Introduction / Abstract"},{"comment":"Eq. (2) writes the log-likelihood with a normalization constant omitted; this is fine, but the text says 'multi-variate Gaussian' while the equation is written for the log; a brief mention of the constant would avoid confusion.","section":"Section 'Mass map inference'"},{"comment":"The choice ℓ<10 removal is stated but not justified; a one-sentence explanation of why these modes are dominated by survey geometry would help. Also, the jackknife error estimation is mentioned but the number of regions and the pixelization used for the systematic maps are not specified.","section":"Section 'Systematic tests on data maps'"},{"comment":"The RMSE and Pearson coefficient for raw KS (0.0371, 0.193) are reported, but the raw KS map is unsmoothed; the comparison with smoothed KS is clear, yet a reader might wonder whether the raw KS RMSE is dominated by small-scale noise. A brief note would help.","section":"Supplementary material, Table I"}],"recommendation":"major_revision","confidential_remarks":"The map construction and null tests are solid and will be useful to the community. The main risk is overclaiming in the filament section: the 'first detection' language is stronger than the evidence. If the authors add a noise-only null test and a quantitative cluster-alignment significance, the paper could be acceptable; without those, the filament claim should be downgraded to 'identification of candidate filaments' or removed from the abstract. The use of the FLAGSHIP cosmology for the Wiener prior is a mild self-consistency loop, but it is not fatal because the map's main claims are validated against systematics and mocks; it should be acknowledged more explicitly as a prior choice rather than a tested assumption."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this paper delivers on the map claim — 13,000 deg^2 from 270 million galaxies, combining DECADE and DES Y3 — and the map construction is careful and validated. The filament \"first detection\" is a weaker claim, and the paper's own text doesn't give it a significance level. I'd trust the map, but treat the filaments as a preliminary application.\n\nWhat's new: the combined dataset is a factor of three larger than the DES Y3 map. The paper is honest about the Wiener filter's Gaussian prior and tests against systematics. The simulation validation with PKDGRAV3 is standard, and the power-spectrum recovery is clearly described. This is a solid data paper.\n\nWhere it gets soft: the filament section. SCONCE is run on the Wiener MAP, which is noise-dominated (S/N RMS ~1). The claim that filament pixels have ~0.5 higher S/N is a property of ridge-finding, not an independent detection. Figure 4 shows cluster distance distributions but no p-value or significance. The comparison to ACT DR6 randoms isn't area-matched for Planck's all-sky sample. Masking effects are explicitly deferred. So the \"first detection\" is not quantitatively established. The map claim is fine; the filament claim needs a noise-only null or a mock with no signal.\n\nAlso, the Wiener prior uses FLAGSHIP cosmology, and the validation uses the same cosmology. That's a mild circularity, not fatal — the prior is standard and the paper notes consistency with cosmic shear constraints.\n\nReproducibility: the map is \"shared upon acceptance,\" so not yet public. That's a limitation for a data-resource paper.\n\nWho it's for: anyone working on mass maps, cross-correlations, or cosmic-web statistics. It deserves a serious referee: the map is worth having, and the filament claim, even if preliminary, is a legitimate test case. My recommendation: send it out, but require the authors to either add a significance estimate or soften the \"first detection\" language.","headline":"A genuinely useful new lensing map, with a filament detection that is more suggestive than demonstrated.","tokens_in":15457,"tokens_out":1453,"would_cite":true,"duration_ms":14624,"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":"The paper combines 270 million galaxies into a 13,000-square-degree lensing map, tripling the covered area, and uses it to detect cosmic filaments directly.","keywords":["weak lensing","mass map","convergence field","cosmic filaments","cosmic web","Wiener filter","Kaiser-Squires inversion","Sunyaev-Zeldovich clusters"],"falsifier":"Take the same shear data, rotate galaxy ellipticities randomly to erase true lensing, rebuild the Wiener MAP, and run SCONCE; if the real filament-cluster distance distribution is not significantly tighter than the noise-only ensemble's, the filaments are not physical. Also rebuild the map with a non-Gaussian prior (e.g., lognormal) and check whether the same filament network and cluster alignment survive.","tokens_in":14409,"feed_emoji":"🕸️","tokens_out":10418,"duration_ms":99191,"temperature":0.7,"pith_summary":"This paper sets out to build the largest map yet of dark matter in the late Universe, reconstructed from the weak gravitational lensing distortions of 270 million galaxies. Combining two surveys taken with the same camera and similar pipeline, it covers 13,000 square degrees—about three times the area of the previous DES Y3 map. The map is made with both Kaiser-Squires inversion and a Wiener filter with a Gaussian prior, and it passes null tests against observing conditions such as seeing, depth, and sky brightness. Then, applying a spherical ridge-finding algorithm to the Wiener map, the paper traces cosmic filaments directly from lensing alone—reportedly for the first time—and shows that Sunyaev-Zeldovich-selected galaxy clusters sit significantly closer to those filaments than random positions. If the filament detection is real, lensing mass maps become a standalone probe of the cosmic web that does not depend on modeling galaxy bias.","feed_headline":"Dark matter map triples to 13,000 square degrees","feed_subtitle":"Made from 270 million galaxies, the map reveals cosmic filaments for the first time.","key_machinery":"The central object is the Wiener-filter MAP estimate of the convergence field on the sphere, computed with the Dante messenger-field sampler under a Gaussian prior with the FLAGSHIP cosmology and a diagonal shape-noise covariance; it yields both the map and posterior samples for uncertainty. The companion machinery is SCONCE, a spherical generalization of the subspace-constrained mean-shift (SCMS) ridge finder that traces filament curves by adaptive gradient ascent on the pixel-weighted convergence map. The Kaiser-Squires E/B-mode inversion plays a supporting role as the flat-prior baseline. The load-bearing move is applying a density-ridge finder designed for galaxy samples directly to the","core_discovery":"On its own terms, the paper claims that the shear catalogs from DECADE and DES Y3—270 million galaxies, all measured from the same camera with the same Metacalibration pipeline—can be stitched into one coherent convergence map covering 13,000 square degrees, roughly three times the sky area of the previous DES Y3 map. The authors validate the map in two ways: against simulations, where the Wiener-filter MAP solution outperforms smoothed Kaiser-Squires in RMSE and pixel correlation, and against the data, where correlations with observing-condition maps are not significant. They then run the spherical ridge finder SCONCE on the Wiener MAP and obtain a network of filament curves whose pixels si","pith_inferences":["The filament detection is more fragile than the map itself: SCONCE runs on a heavily smoothed map whose pixel signal-to-noise is order unity and whose filaments are only about 0.5 S/N above the background, so part of the ridge network could come from the Gaussian prior or the 28-arcmin smoothing. A decisive test is to rerun SCONCE on shape-noise-only reconstructions and on maps made with a non-Gau","If the filament network survives prior changes, the cluster-filament correlation could be turned into a quantitative stacked profile—e.g., mean cluster distance or filament-aligned shear—rather than a distance distribution, giving a cleaner null hypothesis test.","The paper defers the effect of survey masks on filament detection; quantifying completeness and purity as a function of distance to the mask edge would determine how much of the filament web near boundaries is trustworthy and would guide application to the larger Rubin and Euclid footprints."],"forward_implications":["The combined map extends lensing mass maps to 13,000 square degrees, about three times the DES Y3 area, without a visible discontinuity where the two surveys meet.","Null tests against observing conditions indicate the map is clean enough at large scales for cross-correlation studies with CMB lensing, thermal Sunyaev-Zeldovich, and cluster catalogs without per-pixel systematics corrections.","If the filaments are real, the cosmic web can be traced from lensing alone, avoiding the galaxy-bias and selection modeling required by spectroscopic filament finders.","Wiener posterior samples give per-pixel uncertainty estimates, making the map usable for peak counts, moments, topology, and simulation-based inference ahead of LSST and Euclid.","The SZ-cluster-to-filament alignment offers a new, map-based way to test structure growth and gas feedback by stacking Compton-y or other gas tracers along the ridges."],"supporting_citations":[{"why":"Supplies the DECADE galaxy shape catalog and the shear-calibration groundwork that the combined map builds on.","marker":"[6]"},{"why":"Supplies the DES Y3 shape catalogue used as the other half of the source galaxies.","marker":"[45]"},{"why":"Defines the Kaiser-Squires inversion used as the flat-prior baseline mass map.","marker":"[29]"},{"why":"Supplies the curved-sky Wiener reconstruction, the ~20–30 arcmin smoothing scale, and the validation protocol that this 13,000 square degree map extends.","marker":"[31]"},{"why":"Implements the messenger-field Wiener filter that produces the MAP solution and posterior samples used for the filament analysis.","marker":"[56]"},{"why":"Sets the FLAGSHIP cosmology used as the Gaussian prior's power spectrum in the Wiener filter.","marker":"[60]"},{"why":"Defines the SCONCE spherical ridge finder used to extract filaments from the convergence map.","marker":"[66]"},{"why":"Supplies Planck SZ-selected clusters used to validate that the recovered filaments coincide with overdense structure.","marker":"[69]"},{"why":"Supplies the independent ACT DR6 SZ cluster catalog for the same filament validation.","marker":"[70]"}],"fun_headline_variants":["Largest dark matter map yet: 13,000 sq deg from 270M galaxies","Cosmic filaments found in 13,000 sq deg dark matter map","270M galaxies reveal dark matter map 3x larger than before","First direct detection of cosmic filaments in 13,000 sq deg map"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is that the filament ridges visible in the smoothed, noise-dominated lensing map are real cosmic filaments rather than artifacts of the smoothing or of the Gaussian prior used in the reconstruction.","fun_headline_variants_meta":{"raw":{"variants":["Largest dark matter map yet: 13,000 sq deg from 270M galaxies","Cosmic filaments found in 13,000 sq deg dark matter map","270M galaxies reveal dark matter map 3x larger than before","First direct detection of cosmic filaments in 13,000 sq deg map"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000348,"raw_usage":{"total_tokens":1697,"prompt_tokens":657,"completion_tokens":1040,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":401,"completion_tokens_details":{"reasoning_tokens":970}},"tokens_in":401,"tokens_out":1040,"duration_ms":10666,"temperature":1.0,"reasoning_tokens":970,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T10:38:51.649951+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the same shear data, rotate galaxy ellipticities randomly to erase true lensing, rebuild the Wiener MAP, and run SCONCE; if the real filament-cluster distance distribution is not significantly tighter than the noise-only ensemble's, the filaments are not physical. Also rebuild the map with a non-Gaussian prior (e.g., lognormal) and check whether the same filament network and cluster alignment survive.","supporting_citations":[{"cited_title":"Kaiser and G","cited_arxiv_id":null,"evidence_quote":"Defines the Kaiser-Squires inversion used as the flat-prior baseline mass map."},{"cited_title":"Wiener filtering and pure E/B decomposition of CMB maps with anisotropic correlated noise","cited_arxiv_id":"1906.10704","evidence_quote":"Implements the messenger-field Wiener filter that produces the MAP solution and posterior samples used for the filament analysis."},{"cited_title":"SCONCE: A cosmic web finder for spherical and conic geometries","cited_arxiv_id":"2207.07001","evidence_quote":"Defines the SCONCE spherical ridge finder used to extract filaments from the convergence map."}],"review_version":1}