{"id":"c82e2ade-bc97-48c6-9870-b09327d73577","arxiv_id":"2608.02755","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"fCAMIRA measures photometric redshifts for 129 shear-selected clusters with mean bias about 0.005 and scatter about 0.008, and forecasts that these systematics have negligible impact on cosmological constraints for HSC-Y3-like samples.","lead":"This paper develops fCAMIRA, a method to confirm 129 galaxy clusters found purely by weak gravitational lensing and to measure their distances with about 0.5% accuracy. By adding reliable distances, it makes these gravity-selected clusters usable for precision cosmology, which previously was limited by missing redshift information.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed 0.008 photo-z scatter is validated against spectroscopic redshifts of galaxies selected by the same RS model, making the test circular and unable to catch misidentified counterparts.","rationale":"The reader's weakest_assumption (richness-mass linearity) is real, but it is only damaging if a wrong choice of counterpart goes undetected. In this paper it goes undetected by construction, because the validation sample is drawn from the same RS model. Equation (12) defines membership with the RS model and the fCAMIRA redshift; Section 5.1 then estimates z_cl,spec from these members. This is a circular validation: the test and the measurement share the same redshift-dependent selection. The paper's own independent check, the comparison with z_Chen25, shows roughly 3% scatter and an outlier fraction of about 7% to 15% depending on threshold, which is not consistent with the claimed 0.008 scatter if z_Chen25 were taken as a reference. The authors attribute the difference to systematics in external catalogs, but that attribution is not tested. The cosmological forecast in Section 6 underscores the issue: it models photo-z systematics as a constant bias plus Gaussian scatter and random shuffles. If real misidentifications are correlated with WL peak height or with line-of-sight large-scale structure, the 'negligible impact' conclusion does not follow. The paper is still a valuable methods contribution, and the fCAMIRA photo-z may well be excellent; however, the precision numbers in the abstract are not yet supported by an independent validation. This keeps the verdict CONDITIONAL rather than ACCEPT or REJECT: the pipeline is sound in design, but the redshift validation needs an independent member-selection step before the precision claim can be accepted.","tokens_in":61537,"tokens_out":6513,"duration_ms":65296,"concrete_test":"Recompute z_cl,spec for the 106 shear-selected clusters with at least three spec-z members using members selected by an independent method, such as redMaPPer or a DNN-based photo-z catalog in the HSC footprint, instead of w_mem from Equation (12). Re-evaluate the bias and scatter of z_cl,phot relative to this independent z_cl,spec. If the scatter remains at or below 0.01 and the outlier fraction at or below 8%, the circularity concern is resolved. If the scatter increases to roughly 0.02 or larger, the claimed sub-percent precision is an artifact of the self-consistent member selection, and the cosmological impact forecast in Section 6 would need to be redone with the true redshift error distribution.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing weakness is not the richness-mass linearity per se but the fact that the headline photo-z validation cannot detect a failure of that assumption. In Section 5.1, the 'spectroscopic cluster redshift' z_cl,spec is built from galaxies selected by w_mem > 0.1 in Equation (12). That membership weight uses the same calibrated RS model and the same fCAMIRA redshift z_cl,phot that is being tested. If the RS model has a systematic color-redshift offset, or if the lensing-score ranking in Equation (11) picks a foreground or background group, then the selected 'members' are simply the galaxies that agree with the chosen redshift; their spec-z will confirm z_cl,phot by construction. The comparison in Figures 8-9 therefore measures internal consistency, not absolute accuracy. The independent cross-match z_Chen25 in Figure 8 shows roughly 3% scatter and 7-15% outliers, far larger than the claimed 0.008 scatter, so the self-consistent validation is not capturing the true error distribution. The cosmological forecast in Section 6 then injects only Gaussian scatter and random redshift shuffles, which is the benign version of misidentification; real projection-induced misidentification is correlated with richness, mass, and line-of-sight structure and is not represented in the mock. Until z_cl,spec is derived from members selected independently of the RS model, the central 'bias 0.005, scatter 0.008' claim is unproven.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops fCAMIRA, a forced-mode CAMIRA algorithm for optical confirmation of 129 weak-lensing shear-selected clusters from HSC-Y3. The red-sequence model is calibrated in two steps: a metallicity-luminosity relation from eFEDS X-ray clusters out to z≈1.3, and color offsets and intrinsic scatter from spectroscopic galaxies. fCAMIRA builds two richness maps, identifies redshift peaks at the WL centers, ranks optical counterparts by a lensing score (Eq. 11), and assigns the best-matched counterpart's refined redshift and richness. The paper reports a cluster photo-z bias of about 0.005 and scatter of about 0.008 relative to spectroscopic cluster redshifts, an outlier fraction of roughly 7–8% relative to external cross-matching, and uses mock forecasts to argue that these systematics have negligible impact on HSC-Y3-like N(ν,z) abundance cosmology.","tokens_in":61845,"tokens_out":6746,"duration_ms":61531,"significance":"If the claimed photo-z performance holds, this is a valuable ingredient for shear-selected cluster cosmology: it adds redshift information to the WL-selected sample, enabling N(ν,z) modelling that substantially improves constraints over N(ν) alone. The data-driven RS calibration is well conceived, the galaxy-level photo-z validation on a held-out spectroscopic sample is a real strength, and the planned public release of the confirmed cluster catalog is good practice. However, the cluster-level spectroscopic validation is not independent of the RS model, and the mock forecast does not reproduce the correlated nature of projection-induced misidentification. The central quantitative claims therefore need additional support before the paper can serve as the basis for precision cosmology.","major_comments":[{"comment":"The spectroscopic cluster redshift z_cl,spec used in Figures 8 and 9 is constructed from galaxies selected with w_mem > 0.1, where w_mem (Eq. 12) is built from the same calibrated RS model and evaluated at the fCAMIRA redshift z_cl that is being tested. If the RS colors have a systematic offset, or if the lensing score in Eq. (11) selects a foreground or background group, the selected \"members\" are preferentially the galaxies that agree with the trial redshift, so their spectroscopic redshifts will confirm z_cl by construction. The measured bias of about 0.005 and scatter of about 0.008 therefore largely quantify internal consistency rather than absolute accuracy. The only independent comparison in Figure 8, z_Chen25 versus z_cl,spec, shows roughly 3% scatter and outlier fractions of 7–15% depending on threshold, which is much larger than the claimed 0.008 scatter. Please re-derive z_cl,spec from members selected independently of the RS model, for example by taking all spectroscopic galaxies within a projected radius and a narrow velocity window, or by using an external cluster catalog, and re-report the bias, scatter, and outlier fraction for the same clusters.","section":"Section 5.1, Eq. (12)"},{"comment":"The forecast scenario \"perturbed z_cl with outliers\" adds Gaussian scatter and randomly shuffles cluster redshifts to produce an 8% outlier fraction. This is the benign, uncorrelated limiting case. In the real sample, projection-induced misidentification occurs because a rich, massive line-of-sight halo outranks the true counterpart in Eq. (11); the resulting redshift outliers are therefore correlated with richness, mass, and flens, and they affect the selection function in ways that random shuffling does not capture. The paper itself acknowledges at the end of Section 6 that projection effects in shear-selected clusters are more complex than in optically selected samples. As written, the mock does not validate the claim that the measured outlier population has negligible cosmological impact. Please inject outliers drawn from the actual flens-ranked distribution or from a mock containing projected halos, or explicitly restrict the conclusion to random contamination.","section":"Section 6, Eq. (14), Fig. 10"},{"comment":"The lensing score S_opt assumes N_mem ∝ M when ranking optical counterparts and choosing z_cl. The cited richness–mass relations are power laws with slope consistent with unity, but the paper does not test the sensitivity of the ranking to deviations from exact linearity. In particular, the roughly 16% of clusters with flens ≤ 0.4 are precisely the cases where the ranking must decide between candidates, and the self-consistent spectroscopic validation in Section 5.1 cannot detect a ranking failure because the members are selected using the same z_cl. A concrete test would be to repeat the counterpart selection with N_mem^α for α = 0.8 and 1.2 and report how many best-matched counterparts change and how the outlier fraction changes.","section":"Section 4.2, Eq. (11)"}],"minor_comments":[{"comment":"\"fCMAIRA\" appears as a typo for fCAMIRA; the same section also contains \"light-of-sight\" where \"line of sight\" is intended.","section":"Section 7, Conclusions"},{"comment":"\"potometric redshift\" should be \"photometric redshift\".","section":"Section 3.3"},{"comment":"The axis label \"mRS(z, )\" is incomplete; it should identify the plotted quantity, e.g., δm_RS as a function of rest-frame wavelength for the labeled redshifts.","section":"Figure 2"},{"comment":"\"the dimensionless aperture aperture mass peak\" contains a duplicated word (\"aperture aperture\").","section":"Section 6"},{"comment":"The abstract states \"approximately 8% of the total sample exhibits redshift discrepancies greater than 0.15,\" while Section 5 reports 6.7% at the 0.15 threshold and 13–15% at thresholds of 0.10 and 0.08; the abstract and Section 7 should use the same definition and value as Section 5.","section":"Abstract and Section 5"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the circularity of the cluster-level spec-z validation is the deciding issue. If the authors can provide an independent spec-z test, even for a subset, or clearly qualify the accuracy claim, the paper could become acceptable. I do not see a novelty or scope concern; the fCAMIRA method and its application to HSC-Y3 are appropriate for this journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: fCAMIRA is a genuinely useful tool and the paper deserves a serious referee, but the headline numbers are not as airtight as they look. The 0.008 scatter is a measure of internal consistency, not absolute accuracy, and the cosmological forecast rests on the benign version of projection effects.\n\nWhat's actually new: forced-mode counterpart identification at WL peak positions, dual-filter richness maps (TI20 and CAMIRA) matched to the WL aperture, and the lensing-score ranking with f_lens. The data-driven RS calibration using eFEDS clusters out to z~1.3 is a solid step beyond fixed-template red sequences. The validation on 67,000 held-out spectroscopic galaxies (bias 0.2%, scatter 2.5%) shows the RS model itself is well calibrated. The paper is honest about limitations: Section 6 explicitly says the mocks assume perfect knowledge of the selection function and are optimistic forecasts.\n\nThe main problem is that the cluster photo-z validation is circular. In Section 5.1, z_cl,spec is computed from galaxies selected with w_mem>0.1, which is built from the same RS model and the same z_cl,phot being tested. The selected members are those that agree with the candidate redshift, so the resulting 0.008 scatter cannot reveal a systematic color-redshift offset or a misidentified counterpart. The comparison with z_Chen25 (direct positional matching to external catalogs) shows ~3% scatter and 7-15% outliers depending on threshold—much larger than 0.008. The paper acknowledges the outliers as projection effects, but then the forecast in Section 6 models outliers as random Gaussian scatters and random shuffles. That is the benign version; real misidentification is correlated with richness, mass, and line-of-sight structure, and is not represented in the mock. So the claim that redshift systematics have 'negligible impact' is only established for the benign case, and the 'precision cosmological sample' title is premature given that the actual abundance modeling in (nu,z) is not done here—only a forecast.\n\nA minor point: the lensing score assumes richness scales linearly with mass. The paper cites power-law relations with index consistent with unity, so it is a reasonable first-order assumption, but it is not tested for the specific TI20 richness. That is a minor concern compared to the circularity.\n\nWho this is for: cluster cosmologists working on shear-selected samples, particularly for future Stage-IV surveys. The fCAMIRA method itself is worth building on. With proper independent validation—e.g., using spec-z galaxies selected without the RS model, or a fully simulated projection treatment—this could become the standard confirmation tool. As is, I would not take the 0.008 scatter as the real accuracy, but the paper deserves a serious referee and heavy revision.\n\nRecommendation: send to peer review. The method is novel and the authors are transparent about limitations. The referee should push for an independent redshift validation and a more realistic outlier model before publication.","headline":"Useful and honestly limited method paper, but the headline photo-z scatter is circularly validated and the cosmological forecast only tests the benign version of projection effects.","tokens_in":62458,"tokens_out":4548,"would_cite":true,"duration_ms":40010,"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":"This paper shows that a forced-mode red-sequence finder, fCAMIRA, measures photometric redshifts for 129 weak-lensing shear-selected galaxy clusters with bias ≈0.005 and scatter ≈0.008, and that these redshifts are precise enough for…","keywords":["weak gravitational lensing","galaxy clusters","photometric redshifts","red sequence","cluster cosmology","shear-selected clusters","projection effects","Hyper Suprime-Cam"],"falsifier":"Take the 129 clusters, obtain complete spectroscopic redshifts for their candidate member galaxies (or for the roughly half with f_lens below 0.5), and compare the photo-z outlier fraction in the low-f_lens subset with the overall 8% rate; a significantly higher outlier fraction would indicate that the lensing-score ranking misidentifies projected halos as the true clusters.","tokens_in":61335,"feed_emoji":"🔭","tokens_out":8069,"duration_ms":69726,"temperature":0.7,"pith_summary":"The paper's goal is to give a clean redshift to each of the 129 weak-lensing shear-selected galaxy clusters found in the HSC-Y3 aperture-mass maps, so that cluster abundance can be modelled as a joint function of signal-to-noise and redshift instead of signal-to-noise alone. The tool it builds, fCAMIRA, forces the CAMIRA red-sequence finder to run at the fixed sky positions of the shear-selected peaks, calibrates a red-sequence galaxy model in a data-driven way (a metallicity–luminosity relation from X-ray-selected eFEDS clusters out to z≈1.3, then color-offset calibration with large spectroscopic samples), and ranks all optical counterpart candidates along the line of sight by their lensing scores. Against spectroscopic cluster redshifts, the resulting photometric redshifts show a mean bias of about 0.005 and scatter of about 0.008, with an outlier fraction of roughly 8% attributed to projection effects causing mis-identification of the optical counterpart. Mock abundance forecasts show that these systematics are negligible for HSC-Y3-like samples of about 220 clusters, which is why this matters: such a sample can now move from constraining only S8 to constraining Omega_m and sigma_8 simultaneously.","feed_headline":"Sub-percent cluster redshifts unlock the HSC shear-selected sample","feed_subtitle":"A forced-mode red-sequence finder measures cluster distances precisely enough to sharpen dark-matter constraints.","key_machinery":"fCAMIRA (forced-mode CAMIRA) is the algorithm that carries the argument: given a WL peak position, it builds two red-sequence richness maps in fine redshift bins—one, N_TI20, using the same truncated-isothermal filter that defines the aperture-mass maps, and the other, N_CAMIRA, using a spatial filter matched to a typical cluster size of R≈0.8 $h^{-1}$ Mpc—then finds redshift peaks at the WL center, locates optical cluster candidates within 7 arcmin, and assigns each candidate a lensing score S_opt = N_opt D_A(z_opt) D_A(z_opt, z_src)/D_A(z_src) with z_src≈1.3. Ranking candidates by S_opt assumes that richness scales linearly with halo mass (N_mem ∝ M), so the top-ranked counterpart is the halo that dominates the lensing signal; the ratio f_lens = S1/(S1+S2+S3) then quantifies how much of the WL signal comes from the best-matched halo, giving a per-cluster measure of projection contamination.","core_discovery":"The central claim is that a forced-mode optical confirmation pipeline, built on a red-sequence model calibrated entirely from data, can assign photometric redshifts to shear-selected clusters with sub-percent accuracy and precision. Applied to the 129 clusters of the HSC-Y3 WL-selected sample, fCAMIRA achieves a mean redshift bias of ≈0.005 with scatter ≈0.008 relative to spectroscopic cluster redshifts, after correcting a mild bias; about 8% of the sample disagrees with positional cross-match redshifts by more than 0.15, which the authors attribute to projection effects along the line of sight rather than to failures of the red-sequence model. The authors further demonstrate, with mock cluster catalogs generated from the same WL selection, that a bias of 0.01, a scatter of 0.01, and an outlier fraction of 8% have negligible impact on cosmological constraints on Omega_m, sigma_8, and S8 for samples of about 220 clusters, while these systematics become comparable to statistical uncertainties for Stage-IV-like samples of about 2400 clusters.","pith_inferences":["The f_lens ranking could be turned directly into a sample-selection cut: restricting to f_lens above a threshold such as 0.5 or 0.8 should reduce the projection-induced outlier fraction at the cost of sample size; the paper quantifies the f_lens distribution but does not re-run the cosmological forecasts under such a cut.","The same forced-mode confirmation logic should transfer to X-ray- or SZ-selected cluster samples, providing redshift and richness estimates on the same data-driven red-sequence model without entangling the selection functions of external catalogs.","The forecast that Stage-IV sample sizes will be sensitive to these redshift systematics suggests that future analyses will need to marginalize over an outlier plus redshift-scatter model or add spectroscopic calibration samples, rather than treating photo-z errors as negligible.","The agreement between fCAMIRA redshifts and spectroscopic BCG redshifts points to a low-cost validation path: a modest spectroscopic campaign targeting member galaxies of the low-f_lens clusters could directly test whether projection mis-identification is the dominant outlier channel."],"forward_implications":["The 129 HSC-Y3 shear-selected clusters now have uniformly measured photometric redshifts and richnesses from a single optical-confirmation procedure, providing the input needed to model cluster abundance as N(ν,z) rather than N(ν) alone.","The measured bias of about 0.005 and scatter of about 0.008 mean cluster-redshift systematics will not dominate parameter errors in HSC-Y3-like samples; the paper's forecasts show Ωm, σ8, and S8 each constrained to roughly 0.04 with about 220 clusters.","About 70 per cent of the sample has f_lens ≥ 0.5, and around 40 clusters have f_lens ≥ 0.9, indicating that the majority of WL detections are dominated by a single massive halo rather than by line-of-sight projections.","For Stage-IV-like samples of about 2400 clusters, a photo-z bias of 0.01, scatter of 0.01, and 8% outliers shift cosmological constraints by about 1 sigma, so next-generation surveys will need further improvement in cluster redshift measurements.","The lensing fraction f_lens is introduced as a practical tool to flag and potentially reduce line-of-sight contamination in shear-selected cluster samples."],"supporting_citations":[{"why":"Supplies the CAMIRA cluster-finding algorithm and the chi-square/number-parameter formalism that fCAMIRA adapts to forced-mode optical confirmation.","marker":"Oguri (2014)"},{"why":"Provides the optical-confirmation prescription and the baseline composite stellar population model for the red sequence.","marker":"Liu et al. (2015)"},{"why":"Constructed the HSC-Y3 aperture-mass maps and the 129 shear-selected clusters that form the sample.","marker":"Oguri et al. (2021)"},{"why":"Gives the previous abundance-only cosmological analysis of these clusters that the new redshift information extends to N(ν,z).","marker":"Chiu et al. (2024)"},{"why":"Provides the selection-function model and the positional cross-match redshifts z_Chen25 used for comparison with fCAMIRA redshifts.","marker":"Chen et al. (2025)"},{"why":"Anchors the absolute luminosity calibration of the red-sequence model through local cluster luminosity functions.","marker":"Lan et al. (2016)"},{"why":"Supplies the spectral energy distribution template library used to generate the passively evolving galaxy templates.","marker":"Bruzual & Charlot (2003)"},{"why":"Cites the power-law richness-mass relation that supports the N_mem ∝ M linearity assumption behind the lensing score.","marker":"Murata et al. (2019)"}],"fun_headline_variants":["fCAMIRA cuts cluster photometric redshift bias to 0.005","Forced-mode optical finder nails cluster redshifts to sub-percent","Precision cluster redshifts from HSC shear-selected sample","Projection effects explain 8% redshift outliers in fCAMIRA","Sub-percent cluster redshifts via data-driven red-sequence model"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The method's ranking of which galaxy cluster corresponds to a detected lensing signal assumes that the number of red galaxies in a cluster grows in direct proportion to the cluster's mass; if that proportionality fails for a meaningful share of clusters, the assigned cluster distance and the claimed outlier rate could both be biased.","fun_headline_variants_meta":{"raw":{"variants":["fCAMIRA cuts cluster photometric redshift bias to 0.005","Forced-mode optical finder nails cluster redshifts to sub-percent","Precision cluster redshifts from HSC shear-selected sample","Projection effects explain 8% redshift outliers in fCAMIRA","Sub-percent cluster redshifts via data-driven red-sequence model"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00089,"raw_usage":{"total_tokens":3928,"prompt_tokens":1126,"completion_tokens":2802,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":742,"completion_tokens_details":{"reasoning_tokens":2728}},"tokens_in":742,"tokens_out":2802,"duration_ms":19695,"temperature":1.0,"reasoning_tokens":2728,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:00:40.213501+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the 129 clusters, obtain complete spectroscopic redshifts for their candidate member galaxies (or for the roughly half with f_lens below 0.5), and compare the photo-z outlier fraction in the low-f_lens subset with the overall 8% rate; a significantly higher outlier fraction would indicate that the lensing-score ranking misidentifies projected halos as the true clusters.","supporting_citations":[],"review_version":1}