{"id":"37467811-3570-41fd-8b76-7c21747fecbc","arxiv_id":"2411.15682","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A visual census of the TNG50 simulation finds 807 ringed galaxies (59% inner rings), which are redder, more metal-rich, and less star-forming than matched non-ringed galaxies, with inner rings at the half-mass radius and outer rings at 1.5 times it.","lead":"This paper builds a catalog of 807 ringed galaxies from the TNG50 cosmological simulation, sorting them into inner, outer, combined, and partial rings. It compares their star formation, gas, color, and metallicity against matched non-ringed galaxies and measures where rings sit relative to galaxy size.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The catalog's headline number depends on unvalidated visual classification, and Table 3's bar counts sum to 790 rather than 807; until the catalog is released and classification repeatability/complete counts are demonstrated, the sample statistics are not on firm ground.","rationale":"The reader's weakest-assumption analysis correctly points to the unvalidated visual classification as the load-bearing element of the paper. I agree that this is the primary concern: every headline number, from the 807-galaxy count to the 64% bar fraction and the control-sample differences, depends on the reliability of classifications made by eye from synthetic, dust-free face-on images. The paper's own quantitative radial-profile analysis is a genuine strength, and it supports the internal consistency of the classifications by showing density bumps near r50 and 1.5 r50, but it does not test whether the classification would be reproduced by other classifiers or whether the catalog is complete. The additional internal inconsistency in Table 3, where bar counts sum to 790 rather than 807, reinforces the need for a released catalog with per-galaxy identifiers. This does not require rejection; the paper can be published as a conditional catalog if the authors provide reproducibility checks, a completeness statement, and corrected tables. The recommended verdict therefore remains CONDITIONAL, matching the reader's assessment.","tokens_in":22409,"tokens_out":10595,"duration_ms":106799,"concrete_test":"Release the per-galaxy catalog (subhalo ID, snapshot, ring class, bar flag) and have two independent classifiers reclassify a random, blinded subset of about 100 images from the same selection; then compute inter-rater Cohen's kappa per ring type and bar presence, and verify from the released IDs that Table 3 sums to 807. If kappa is below 0.6 for any class, or if 17 galaxies cannot be assigned a bar measurement, rerun the affected statistics on the full sample and state the classification completeness.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is the construction of a catalog of 807 ringed galaxies and the derived statistics on ring-type fractions, bar fraction, and control-sample differences. The weakest load-bearing premise is that the visual classification in Section 2.3 is accurate and repeatable. No inter-rater agreement, blinded test, or quantitative ring-detection algorithm is reported, and the only internal quantitative check (Section 5.2) stacks profiles of galaxies already classified as ringed, so it validates that selected objects have density bumps but cannot test completeness or classification bias. A systematic bias in what is called a ring would propagate directly into Table 1 percentages, the SDSS comparison in Section 2.5, and all property comparisons in Section 4. There is also an objective internal inconsistency: Table 3 reports 507 barred + 283 non-barred = 790 galaxies, while Section 2.3 and Table 1 use 807. No explanation is given for the missing 17 systems, and the quoted '64% (507 galaxies)' is 64% of 790, not 64% of the full 807-galaxy sample (62.8%). This makes the bar fraction and all barred/unbarred subsample analyses unreliable until the complete catalog and bar classification are released and reconciled.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a visually classified catalog of 807 ringed galaxies in the TNG50 simulation at 0.01 < z < 0.1, split into inner, outer, inner+outer, and partial/pseudo-rings. It reports ring-type fractions, a bar fraction of 64% using the kinematic bar catalog of Zana et al., and compares the ring-type fractions with the SDSS-DR14 ringed-galaxy catalog of Fernandez et al. (2021). A control sample of five nearby non-ringed galaxies per ringed galaxy is matched in redshift, stellar mass, and local environmental density. Using this control sample, the authors report lower sSFR and gas fractions, redder colors, and higher metallicities for ringed galaxies, with further differences between barred and unbarred subsamples. Radial stellar-mass profiles are stacked in r50 units, and the paper claims that inner rings lie approximately at r50, outer rings at 1.5 r50, and that i+o galaxies have more compact and massive inner rings. The results are compared with the S4G/Arrakis observational ring-size catalog.","tokens_in":22663,"tokens_out":5795,"duration_ms":53322,"significance":"If the catalog and its statistics are reliable, this is a potentially useful resource: TNG50 provides homogeneous, high-resolution simulated galaxies, and the matched control-sample design gives the property comparisons a sound statistical basis beyond simple mean differences. The use of KS tests and bootstrap bands, the consistency checks of the control-sample matching, and the external comparison with Arrakis are strengths. The catalog also yields a falsifiable prediction about ring locations relative to r50. However, the central caveat is that the classification is entirely visual and its reliability is not demonstrated; the catalog counts, the SDSS comparison, and all property trends inherit this uncertainty. The bar-count arithmetic also needs reconciliation before the barred/unbarred results can be accepted.","major_comments":[{"comment":"The catalog is defined solely by visual classification of synthetic images, but no reliability assessment is reported: there is no inter-rater agreement test, no blinded re-classification, and no quantitative ring-detection criterion. The stacking analysis in Section 5.2 only shows that the selected galaxies have density bumps at the claimed radii; it cannot test completeness or measure selection bias. Because every headline number in Table 1 and all ringed-versus-control comparisons in Section 4 inherit this classification, please add a repeatability measurement (e.g., re-classification of a random subset by a second classifier and by the same classifier after a time interval, with agreement statistics) or an equivalent quantitative check.","section":"Section 2.3"},{"comment":"The barred and non-barred entries sum to 790 galaxies (507 + 283), not the stated 807; consequently the quoted 64% (507 galaxies) is 64% of 790, while 507/807 = 62.8%. No explanation is given for the missing 17 systems, and if those systems lack bar classifications from the Zana et al. catalog this should be stated and the subsample analyses in Sections 4.1–4.3 should be repeated with the correct sample definition.","section":"Table 3 and Section 2.4"},{"comment":"The comparison of ring-type fractions between TNG50 and SDSS-DR14 is not an apples-to-apples comparison because the parent samples are selected differently: the SDSS catalog uses apparent magnitude, axial-ratio, and concentration cuts, while the TNG50 sample uses stellar mass, r50, and sSFR cuts. The differences in inner-ring and i+o fractions (59% versus 49% and 7% versus 21%) could partly reflect these selection differences. Please either apply the SDSS selection criteria to the simulation or clearly state this limitation before drawing conclusions from the comparison.","section":"Section 2.5 and Table 1"},{"comment":"The headline quantitative statements that inner rings are located approximately at r50 while outer rings are at 1.5 r50 are based on visual inspection of residual bumps; the paper gives no peak-detection algorithm, no uncertainty interval on the peak positions, and no quantitative measure of the claimed central mass deficit. The bootstrap error bars in Fig. 11 show scatter of the mean residual but do not propagate to the peak location. Please quantify the ring radii (e.g., fit the residual bumps and report fitted centers and uncertainties) and state how the 'more compact and massive' inference for i+o inner rings is measured.","section":"Section 5.2 and Fig. 11"}],"minor_comments":[{"comment":"The abstract states '0.01 < z < 0.1' while Section 2.2 states '0.01 ≤ z ≤ 0.1'; make the redshift range consistent.","section":"Abstract and Section 2.2"},{"comment":"There are repeated typographical issues such as 'di fferent' in the abstract and Section 2.2, and 'Percentil range' in the Fig. 3 caption; these should be corrected.","section":"Throughout"},{"comment":"'Front projection' and 'front perspective' should be replaced with 'face-on projection' and 'face-on view' for clarity.","section":"Section 5.1 and Fig. 10"},{"comment":"The paper is titled a catalog paper, but no catalog table, machine-readable file, or Data Availability statement is provided; please state where the catalog will be released.","section":"General"},{"comment":"The per-redshift totals are not stated; readers must infer the sample size in each column from the quoted percentages, so please add a total row or column.","section":"Table 2"},{"comment":"The Spearman coefficients for the control sample appear only in the figure; please state in the text that the control sample has rS = 0.41 with p = 1.20e-148.","section":"Section 4.3 and Fig. 9"},{"comment":"The conversion r50 ≈ 3.08 R25 is introduced from Leroy et al. (2021), but the direction of the 4% offset from Muñoz-Mateos et al. is not clearly explained; clarify the sign and source of the offset.","section":"Section 5.2"}],"recommendation":"major_revision","confidential_remarks":"The Table 3 arithmetic mismatch and the absence of any classification-reliability check are the two issues that most need attention before this paper can be accepted. The mismatch looks like a correctable data-table error, and a classification-reliability study is a standard and feasible addition for a visual-morphology catalog. If those are addressed, the paper's contributions would be solid. There are no apparent citation or scope concerns. I would not reject the manuscript, but I cannot recommend acceptance in its current form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is the first systematic catalog of ringed galaxies in a cosmological simulation, and the control-sample design is sound. But the paper currently has an internal arithmetic problem in the bar counts, and the visual classification needs at least one repeatability check before the catalog should be treated as a resource.\n\nWhat's actually new: previous ringed catalogs (CSRG, GZ2-CNRG, F21, Arrakis) are all observational; earlier simulations were mostly idealized or single-galaxy models. Applying visual classification to TNG50 synthetic images and measuring ring locations relative to r50 in stacked profiles is a legitimate step forward. The matched control sample (five nearest non-ringed neighbors in Mstar-sigma5 space per redshift) is well chosen, and the KS tests show the matching works. The property trends — lower sSFR, lower gas fractions, redder colors, higher metallicity — are compared against that control sample, so they are not forced by construction, and they line up with the authors' own F21 observational results. The stacked profile analysis is a nice quantitative addition: the residual method (subtracting the control median slope profile) is sensible, and the result that inner rings sit near r50 and outer rings near 1.5 r50 is testable against S4G via the Leroy conversion.\n\nWhere it gets soft. The arithmetic first: Table 3 sums to 790 barred+unbarred galaxies, not the 807 in Table 1 and the abstract. The quoted \"64% (507 galaxies)\" is actually 64% of 790, not 64% of the full sample (which would be 62.8%). Seventeen galaxies are unaccounted for, and the bar fraction and all barred/unbarred subsample comparisons rest on that incomplete table. This is a fixable error, but it has to be fixed before the numbers are used.\n\nSecond, the visual classification has no reliability check. No inter-rater agreement, no blinded test, no quantitative ring-detection. The stacked profiles in Section 5.2 confirm that objects classified as ringed have density bumps at the claimed radii, but that cannot test completeness or classification bias. For a simulation where you have the full 3D data, a quantitative check is feasible, and the catalog would be much stronger with it. Third, the catalog itself is not released or referenced as available, which is an odd omission for a \"catalog\" paper. Finally, minor: the TNG50 ring positions are quoted without a spread, while the S4G comparison has uncertainties; a bootstrap scatter on the stacked bump location would be easy to add.\n\nOverall: the central claims are plausible and the trends are consistent with observations, so I don't think the flaws are load-bearing. But the bar-fraction inconsistency and missing validation data need to be addressed. This deserves peer review, with the expectation of a revised version. I'd be happy to referee it myself.","headline":"First TNG50 ringed-galaxy catalog with a sound control-sample design, but the bar fraction rests on a 790-vs-807 inconsistency and the visual classification lacks a repeatability check.","tokens_in":23207,"tokens_out":3478,"would_cite":true,"duration_ms":30371,"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":"Visual classification of simulated galaxy images yields 807 ringed galaxies, with inner rings at the half-mass radius and outer rings at 1.5 times it.","keywords":["ringed galaxies","TNG50 simulation","galaxy morphology","galactic bars","secular evolution","stellar mass surface density","visual classification","galaxy evolution"],"falsifier":"Hand the same dust-free, face-on synthetic images to independent classifiers, or run an automated ring-detection algorithm on the stellar mass maps, and compare the recovered number and type distribution with the paper's 807; a large disagreement would show the catalog is not robust to how rings are defined.","tokens_in":66,"feed_emoji":"🌌","tokens_out":10425,"duration_ms":145822,"temperature":0.7,"pith_summary":"This paper builds a catalog of ringed galaxies in the TNG50 cosmological simulation by visually classifying synthetic, dust-free, face-on images, and uses it to ask what rings reveal about galaxy evolution. It reports 807 ringed galaxies among star-forming disks with $M_\\star > 10^9\\,M_\\odot$ in the redshift range $0.01 < z < 0.1$, with 59% inner rings, 22% partial rings, 12% outer rings, and 7% inner-plus-outer rings, and 64% barred hosts. Compared with matched non-ringed controls, ringed galaxies show lower specific star formation rates, lower gas fractions, redder colors, and higher metallicities, with larger metallicity scatter at fixed stellar mass. Stacked stellar mass surface-density profiles place inner rings near $r_{50}$ and outer rings at $1.5\\,r_{50}$, preceded by a central mass deficit. A sympathetic reader would care because these are quantitative, simulation-side counterparts to observational ring catalogs, offering a place to test resonant ring-formation theories.","feed_headline":"807 ringed galaxies mapped in TNG50 simulation","feed_subtitle":"Ringed galaxies are redder, gas-poor, and metal-rich; inner rings sit at the half-mass radius.","key_machinery":"The load-bearing object is the visual classification of synthetic, dust-free, face-on TNG50 images, using ring categories adopted from F21 and Buta (2017). The quantitative analysis then rests on three pieces: a control sample that matches each ringed galaxy to its five nearest non-ringed neighbors in normalized ($\\sigma_5$, $M_\\star$) space at the same redshift; the projected stellar mass surface-density profile $\\Sigma_\\star(r)$ computed in six azimuthal sectors so that a bar does not dominate the signal; and the residual of the slope profile relative to the median control profile, $e = [\\Delta_h[M_{\\star,\\mathrm{RG}}](r/r_{50}) - \\mathrm{Me}(\\Delta_h[M_{\\star,\\mathrm{CS}}](r/r_{50}))]/h$, whose oscillation marks the ring radius and the central mass deficit. This combination converts a visual morphology into a measurable radial signal.","core_discovery":"On its own terms, the paper claims that resonant ring structures are common and physically meaningful in simulated disk galaxies: 807 of the selected TNG50 galaxies show rings, and their host galaxies differ systematically from otherwise similar non-ringed disks. The distribution is 58.85% inner, 22.18% partial, 11.77% outer, and 7.18% inner+outer rings, and 64% of hosts are barred. Ringed galaxies form stars less efficiently, have lower gas fractions, redder $M_g - M_r$ colors, and higher gas-phase metallicities than controls, and they scatter more widely in metallicity at fixed stellar mass. When stellar mass surface-density profiles are stacked in units of $r_{50}$, inner rings sit at approximately $r_{50}$ and outer rings at approximately $1.5\\,r_{50}$, with a central mass deficit just inside the ring; in inner+outer systems the inner ring is more compact and massive, and partial rings extend beyond $r_{50}$ with deeper central profiles. These results are presented as broadly consistent with the observed SDSS ringed-galaxy catalog, the main difference being fewer inner+outer systems in the simulation.","pith_inferences":["Our inference: because the classification is purely visual and no inter-rater or automated check is reported, the catalog percentages should be treated as provisional until independent classifiers or a quantitative ring-finder confirm them.","Our inference: if inner and outer rings really sit at $r_{50}$ and $1.5\\,r_{50}$, then observed galaxies should show the same scaled ring radii in large samples, and measuring ring sizes in existing surveys could distinguish this resonance-scaling picture from alternative ring formation channels.","Our inference: the central mass deficit seen ahead of each ring suggests ring formation is tied to inward mass redistribution, which could be tested by tracking individual TNG50 galaxies across snapshots to see whether rings appear after central mass concentration rises.","Our inference: the rise of partial rings toward higher redshift implies they are transient, so tracing the same galaxies through the TNG50 merger trees should show partial rings either dissolving or maturing into outer rings."],"forward_implications":["If the catalog is representative, ring presence marks a later, more quiescent evolutionary stage among disk galaxies: the ringed sample is systematically redder, more metal-rich, and more gas-poor at fixed stellar mass.","The measured locations place inner rings at $r_{50}$ and outer rings at $1.5\\,r_{50}$, so future simulations and observations can check directly whether resonance radii scale with the stellar half-mass radius.","The lower fraction of inner+outer rings in TNG50 compared with SDSS (7% versus 21%) implies the simulation either forms fewer double-ring systems or disrupts them, pointing to resolution or feedback effects.","Partial rings rise from 3% of the sample at $z=0.01$ to 25% at $z=0.1$, which the paper reads as evidence that partial rings are transient and may mature into stable outer rings."],"supporting_citations":[{"why":"Supplies the SDSS-DR14 ringed-galaxy sample and selection criteria used to define the TNG50 target and to compare ring-type percentages and properties.","marker":"F21 (Fernandez et al. 2021)"},{"why":"Provides the ring classification definitions (inner, outer, inner+outer, partial) adopted for the visual labels.","marker":"Buta (2017)"},{"why":"Introduces the TNG50 simulation whose snapshots and galaxy catalogs provide the sample.","marker":"Nelson et al. (2019a)"},{"why":"Defines the synthetic dust-free images in JWST-like filters that the classification is performed on.","marker":"Vogelsberger et al. (2020)"},{"why":"Supplies the kinematic bar identifications used to divide ringed galaxies into barred and unbarred.","marker":"Zana et al. (2022)"},{"why":"Establishes the mass-size relation and exponential-disk-plus-bulge profile description used to set $r_{50}$ and stack profiles.","marker":"Wang & Lilly (2023)"},{"why":"Provides the S4G observed ring catalog and ring sizes that the simulated ring locations are compared against.","marker":"Comerón et al. (2014)"},{"why":"Supplies the observed mass-metallicity relation used as a reference for the simulated ringed and control galaxies.","marker":"Tremonti et al. (2004)"}],"fun_headline_variants":["807 ringed galaxies in TNG50: redder, gas-poor","Ringed galaxies sit at half-mass radius, TNG50 shows","TNG50: rings trace central mass deficit, metal-rich hosts","Simulated rings: inner at r50, outer at 1.5r50"],"cache_read_input_tokens":25344,"weakest_assumption_plain":"The catalog rests entirely on the authors' visual identification of ring structures in synthetic images; if that identification is noisy or biased, the ring counts, the comparison with SDSS, and every ringed-versus-control property difference would shift.","fun_headline_variants_meta":{"raw":{"variants":["807 ringed galaxies in TNG50: redder, gas-poor","Ringed galaxies sit at half-mass radius, TNG50 shows","TNG50: rings trace central mass deficit, metal-rich hosts","Simulated rings: inner at r50, outer at 1.5r50"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00026,"raw_usage":{"total_tokens":1697,"prompt_tokens":1163,"completion_tokens":534,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":779,"completion_tokens_details":{"reasoning_tokens":453}},"tokens_in":779,"tokens_out":534,"duration_ms":5115,"temperature":1.0,"reasoning_tokens":453,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:00:38.449069+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Hand the same dust-free, face-on synthetic images to independent classifiers, or run an automated ring-detection algorithm on the stellar mass maps, and compare the recovered number and type distribution with the paper's 807; a large disagreement would show the catalog is not robust to how rings are defined.","supporting_citations":[{"cited_title":"2021, A&A, 653, A71","cited_arxiv_id":null,"evidence_quote":"Supplies the SDSS-DR14 ringed-galaxy sample and selection criteria used to define the TNG50 target and to compare ring-type percentages and properties."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the ring classification definitions (inner, outer, inner+outer, partial) adopted for the visual labels."},{"cited_title":"2022, MNRAS, 515, 1524 14","cited_arxiv_id":null,"evidence_quote":"Supplies the kinematic bar identifications used to divide ringed galaxies into barred and unbarred."},{"cited_title":"& Lilly, S","cited_arxiv_id":null,"evidence_quote":"Establishes the mass-size relation and exponential-disk-plus-bulge profile description used to set $r_{50}$ and stack profiles."}],"review_version":1}