{"id":"8b5a19e2-19b6-4a79-96e2-e2ad675132d1","arxiv_id":"2607.28735","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Host galaxies of 105 changing-look AGNs are predominantly quiescent and similar to matched extended quasars, with a modest post-starburst excess; narrow lines stay stable across the transitions.","lead":"Using DESI and SDSS spectra plus Subaru HSC images, this paper separates the host galaxies of 105 changing-look AGNs from their varying nuclei and compares them with ordinary quasars. It finds these hosts are mostly quiescent and broadly similar to matched quasars, with a modest post-starburst excess, and argues the changes are driven by central black hole accretion.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-epoch model's tied stellar-population assumption may bias host M⋆/sSFR/Hδ_A if radial gradients exist, undermining the matched-quasar comparison.","rationale":"I read the paper in good faith and agree with the reader's conditional verdict. The work is careful: it uses a matched comparison sample, applies the same pipeline to both samples, reports explicit limitations, and provides internal consistency checks. The central claim is appropriately scoped to CL AGNs with measurable extended host emission. However, the weakest link is the two-epoch decomposition's assumption of a single shared stellar population template across the DESI and SDSS apertures. This assumption is not tested for radial gradients, and a violation would bias the very host properties used in the matched comparison—particularly Hδ_A, which drives the only statistically significant difference (post-starburst excess, p=0.020). The paper's mock tests and host-dominated validations do not cover this scenario. The proposed test—examining HSC radial color profiles and, if needed, re-fitting with a two-component stellar population—would directly settle whether the concern lands. Because the reader already identified this assumption as the weakest and chose CONDITIONAL, my read does not change the verdict.","tokens_in":25891,"tokens_out":7243,"duration_ms":78250,"concrete_test":"Measure HSC radial color profiles (g−r, r−i) within the DESI 1.5″ and SDSS 2″/3″ apertures for the 82 extended CL AGNs. If the median color difference between apertures exceeds the photometric uncertainty (e.g., >0.05 mag), the same-SED aperture correction is invalid; then re-run the decomposition allowing a two-component stellar population (inner+outer) and recompute Table 2. If the post-starburst excess or quiescent offset changes materially, the central claim is not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that extended CL AGN hosts are not a distinct population from matched extended quasars—rests on the fidelity of the two-epoch decomposition (Eq. 1, §3.1). The model assumes a single intrinsic host stellar population θ⋆ is shared by the DESI and SDSS epochs, and the SDSS–DESI aperture difference is represented by an 'additional host spectrum' derived from HSC photometry. If the host has radial age/metallicity gradients, the SED shape of the additional light entering the larger SDSS fiber differs from the fiducial model; the decomposition can then compensate by adjusting the AGN continuum or the shared host template, shifting M⋆, sSFR, and Hδ_A. The comparison quasars are fit with a single-epoch pipeline (Sun et al. 2026), so any two-epoch-specific bias is not shared and directly enters the matched differences in Table 2—especially the post-starburst excess (p=0.020). The paper's internal validations (§5.5) use only two host-dominated objects and mock tests from Sun et al. (2026) that do not inject radial gradients; §5.7 lists limitations but does not mention this assumption. Thus the load-bearing assumption is untested.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a host-galaxy study of 105 changing-look (CL) AGNs identified from DESI/SDSS spectral comparisons. A two-epoch spectrophotometric decomposition model (Eq. 1) is used, tying the host stellar population parameters across the DESI and SDSS epochs while allowing the AGN component to vary, with HSC imaging providing external host-flux constraints. The authors report that the CL AGN hosts are predominantly quiescent (75.2%) and that 29.5% show post-starburst signatures. For 82 CL AGNs with extended host emission, matched one-to-one to extended quasars from the same pipeline in redshift and stellar mass, they find no clear offset in log sSFR or Hδ_A and no significant difference in quiescent fraction, but a higher post-starburst fraction (35.4% vs 18.3%, p=0.020). The [O III] and [O II] narrow lines show no population-wide response over rest-frame baselines of 2.4–18.4 yr. The paper interprets the combined host, broad-line, and narrow-line evidence as favoring changes in the central accretion rate or inner disk structure as the main origin of CL transitions.","tokens_in":26155,"tokens_out":4751,"duration_ms":55789,"significance":"If the host-galaxy measurements are reliable, this is a valuable population-level result: CL AGNs with measurable extended host emission do not appear to form a distinct host population from similarly selected extended quasars, strengthening the interpretation that the CL phenomenon is primarily nuclear rather than driven by galaxy-scale transformations. The study has notable strengths: the CL sample and comparison quasars are analyzed with the same pipeline, reducing methodology-induced differences; HSC imaging provides external constraints on the host flux; the two-epoch design uses the stellar population invariance as a physical consistency check; and the paper includes an explicit, honest limitations section. The broad-line/continuum correlations and narrow-line stability provide independent evidence consistent with the central interpretation. However, the central claim rests on an untested modeling assumption—the shared stellar population across epochs with only an SED-scaled aperture correction—and on statistical comparisons that omit individual measurement uncertainties and multiple-testing corrections. These issues are addressable and do not require abandoning the main conc","major_comments":[{"comment":"The tied stellar-population assumption is load-bearing but untested. In Eq. (1), a single θ⋆ is mapped into the two fiber apertures via T^i, with the SDSS–DESI aperture difference represented by an additional host spectrum fit with the same SED model. If the host has radial age or metallicity gradients, the extra light entering the larger SDSS fiber is not drawn from the same SED, and the tied fit may compensate by adjusting the AGN residual or averaging the two host spectra, biasing M⋆, sSFR, and Hδ_A. The internal checks in §5.5 use only two nearly host-dominated objects and mock tests from Sun et al. (2026) that do not inject radial gradients; §5.7 does not list this limitation. A concrete test is needed—for example, injecting radially varying stellar population models into mock HSC+spectral data and recovering them with this pipeline, or comparing a subset with IFU stellar-population","section":"§3.1, Eq. (1)"},{"comment":"The post-starburst excess (35.4% vs 18.3%, p=0.020) is one of four primary comparisons in Table 2, and the paper performs additional morphology, state, and disturbance tests elsewhere without multiplicity control. With a Bonferroni correction for the four Table 2 outcomes (α=0.0125), p=0.020 is not significant. The abstract and summary state that post-starburst hosts are \"more common\" among CL AGNs, which overstates the evidence. Please report the number of tests, provide corrected p-values, or explicitly label p=0.020 as suggestive only.","section":"Table 2, §4.2"},{"comment":"The reported bootstrap 95% CIs and McNemar p-values are computed from pairs without propagating the measurement uncertainties of the individual decomposition quantities. The paper acknowledges this for the intervals (\"The quoted intervals describe sampling uncertainty\"), but the same omission affects the classification fractions and the p=0.020 result; host-measurement scatter can move objects across the quiescent and post-starburst boundaries. Additionally, §5.2 excludes ID 177 and ID 497 from the [O III] population statistics after visual inspection of \"poor local spectral quality\"; this post hoc exclusion should be justified with quantitative S/N or line-fit quality cuts, and the population statistics should be shown with these objects included or with a robust estimator.","section":"§3.5, §5.2"}],"minor_comments":[{"comment":"There are formatting typos: \"T able 1\" in the text, \"GalfitMusing\" in §2.2, and mismatched math-mode/regular text for [O II] and [O III] in several places. These should be corrected.","section":"General"},{"comment":"The transition from 83 extended CL AGNs with at least one candidate comparison to 82 matched pairs is stated, but the reason the last object is dropped is not explained. Clarify whether the 83rd has no unique partner after the without-replacement matching.","section":"§3.5"},{"comment":"The simple scaling of the star-forming fraction by the cosmic SFR-density ratio (1.48) assumes the star-forming fraction scales linearly with the cosmic SFR density. This is a rough heuristic; state that it is not a quantitative evolutionary correction.","section":"§4.3"},{"comment":"Panel (b) would benefit from explicitly marking the matched-pair status of the post-starburst excess, e.g., by annotating the McNemar p-value on the figure, since the p=0.020 result is a headline number in the abstract.","section":"Figure 6"},{"comment":"The text says PS16dtm is \"not included in our final host decomposition sample\" because it is PSF-dominated, but the object appears in the earlier sample counts (§2.1 lists 137 HSC-matched objects). Clarify that the light-curve atlas includes all 137 objects while the host property sample is 105.","section":"§5.1"},{"comment":"The statement that \"the robust result is the absence of a population-wide [O III] or [O II] response\" is stronger than the evidence because the two largest [O III] changes are excluded and the remaining distribution has a median absolute deviation of only 0.087 dex. Consider rephrasing to \"no population-wide response is detected within the limitations of the current spectra.\"","section":"§5.2"}],"recommendation":"major_revision","confidential_remarks":"The comparison sample is from the same group (Sun et al. 2026) and is analyzed with the same pipeline. This is methodologically sensible, but the host measurements are not fully independent; if an external, pipeline-independent host sample or a subset with IFU stellar-population maps is available, it would materially strengthen the claim. The paper is the sixth in a series, and its conclusion depends in part on earlier papers by the same group; that is acceptable as long as the host-galaxy comparison is self-contained, which it largely is. The main revision requests—testing the radial-gradient assumption and tightening the statistics—are feasible within the scope of the manuscript."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the punchline: this is the largest matched study of CL AGN host galaxies yet, and the main result—extended CL AGN hosts look broadly like matched extended quasars, with a modest post-starburst excess—is defensible. The paper is careful, uses the same pipeline for both samples, and is frank about most of its limitations. The real weakness is the two-epoch decomposition's assumption that one shared stellar population underlies both spectra, with the SDSS–DESI aperture difference captured by an HSC-derived 'additional host spectrum' fit with the same SED model. If radial age or metallicity gradients exist, that extra light has a different SED shape, and the fit can compensate by adjusting the AGN or the shared host template, biasing M*, sSFR, and HδA. The comparison quasars are fit with the single-epoch pipeline, so any such bias is not shared and could directly feed the p=0.020 post-starburst excess. The paper validates on only two host-dominated objects and mocks that do not include radial gradients; Section 5.7 does not mention this assumption. So it's an untested, load-bearing assumption—not a demonstrated flaw, but a genuine systematic risk.\n\nOther soft spots are minor and mostly acknowledged: the post-starburst p=0.020 does not survive a multiplicity correction; the [O III] statistics exclude two outliers post hoc; and the bootstrap CIs do not include measurement uncertainties. They state these limitations explicitly, which is good practice.\n\nCredit where it's due: the narrow-line stability over a median 7.5 yr rest-frame baseline is a nice population result, with rough NLR size lower limits. The matched comparison design is a real step up from earlier CL AGN host studies. And the conclusion favoring accretion-rate changes is not uniquely established, but it is consistent with the correlated broad-line/continuum changes and with prior work from the same group.\n\nWho is this for? Anyone working on CL AGN demographics or AGN-host coevolution. It deserves a serious referee. The main thing I would ask the authors to address is the shared stellar population assumption—for example, by allowing the additional host spectrum to have its own SED parameters, or by injecting radial gradients into mocks to bound the bias.","headline":"Largest matched CL AGN host-galaxy study to date; the central result is defensible, but the two-epoch decomposition's untested shared-stellar-population assumption is the main unresolved risk.","tokens_in":26710,"tokens_out":5830,"would_cite":true,"duration_ms":54116,"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":"Changing-look AGNs are not a distinct host-galaxy population; the trigger is likely a change in the black hole's accretion rate.","keywords":["changing-look AGN","host galaxies","quiescent galaxies","post-starburst","spectrophotometric decomposition","AGN accretion rate","narrow line region","DESI survey"],"falsifier":"Spatially resolved integral-field spectroscopy of these CL AGN hosts: if the derived stellar mass, sSFR, or Hδ strength depends on the assumed aperture scaling, or if strong radial age/metallicity gradients are present, the tied-host assumption fails and the population-level conclusions would need revision.","tokens_in":25743,"feed_emoji":"🌌","tokens_out":7965,"duration_ms":67721,"temperature":0.7,"pith_summary":"Changing-look AGNs abruptly switch their broad emission lines and continuum on timescales of months to years, but whether their host galaxies are special has been unclear. This paper analyzes 105 such objects with two epochs of optical spectroscopy and deep imaging, decomposing each spectrum into a constant stellar population and a varying nucleus. The hosts turn out to be overwhelmingly quiescent, and when matched to ordinary quasars of the same redshift and stellar mass, their star formation activity and absorption-line strengths are statistically indistinguishable. The one difference is a higher post-starburst fraction among the changing-look objects (35% vs 18%, p=0.02). Because the narrow forbidden lines do not respond while the broad lines do, the paper concludes that the transitions are driven by changes in the central black hole's accretion rate or inner disk, not by events that reshape the whole galaxy.","feed_headline":"Changing-look AGNs live in ordinary quiet galaxies","feed_subtitle":"Hosts match normal quasars; the switch lies in the black hole's accretion rate, not the galaxy.","key_machinery":"The key mechanism is a two-epoch spectrophotometric decomposition that ties the two spectra to a single shared host stellar population. The model expresses each epoch's flux as an independent AGN component plus the same intrinsic stellar population mapped into the different fiber apertures. Deep multi-band imaging provides an independent constraint on the host flux in each aperture, and an imaging-derived 'additional host spectrum' accounts for the aperture difference. The stellar population parameters (mass, star formation history, dust) are fit jointly across epochs, so the fainter state anchors the host while the brighter state constrains the AGN. The paper also constructs a matched compa","core_discovery":"At the population level, changing-look AGNs with measurable extended host emission do not occupy a distinct host galaxy population from similarly selected extended quasars. Among 105 two-epoch objects (SDSS and DESI spectra plus deep HSC imaging), 75.2% of hosts are quiescent and 29.5% post-starburst. Matched one-to-one against 82 extended quasars at the same redshift and stellar mass, CL hosts show consistent log sSFR (−12.87 vs −12.62), consistent Hδ_A (2.54 vs 1.36 Å; CI [−0.59,+2.94]), and a consistent quiescent fraction (81.7% vs 70.7%, p=0.108). The post-starburst fraction is higher (35.4% vs 18.3%, p=0.020). [O III] and [O II] remain stable over rest-frame baselines up to 18 yr while","pith_inferences":["If CL transitions are purely accretion-rate driven, then any quasar could in principle undergo a changing-look event; the observed CL fraction would be set by survey cadence and sensitivity rather than by a special host property, a prediction testable in unbiased samples.","The modest post-starburst excess might be a selection effect: post-starburst hosts have a bright young stellar continuum that makes the AGN easier to detect in the faint state, artificially boosting their representation among CL AGNs found by spectral comparison.","The narrow-line stability could be turned into an echo-mapping tool: a delayed [O III] response to a known CL transition would directly measure the NLR radius on decade timescales.","A next step is comparing CL AGN hosts to inactive galaxies of the same stellar mass and redshift, not just to quasars, to see whether the apparent consistency with quasars is shared by the general galaxy population."],"forward_implications":["If CL AGN hosts are statistically identical to quasar hosts, the changing-look phenomenon is decoupled from galaxy-scale properties, and CL selection is nearly random with respect to host star formation history.","The stable narrow lines imply the narrow-line region extends beyond a few parsecs in these objects, giving a population-level size constraint from variability alone.","The higher post-starburst fraction, if confirmed, suggests a mild association between recent quenching and CL activity, but not a synchronized event.","The positive broad-line/continuum correlation supports a stratified BLR where inner lines respond before outer lines, consistent with an accretion-rate-change origin."],"fun_headline_variants":["CL AGN hosts match normal quasars, not special galaxies","Changing-look AGNs: hosts ordinary, central engine switched","Post-starburst hints but no distinct host galaxy for CL AGNs","Narrow lines stay put: CL AGN flips come from accretion","105 changing-look AGNs: quiet hosts, active black holes"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that one shared stellar population model can describe the host galaxy at both epochs and in both fiber apertures; if the host has strong radial age or metallicity gradients, or if the imaging-based aperture correction is off, the tied host fit can absorb AGN residuals and bias the derived stellar mass, star formation rate, and Hδ strength.","fun_headline_variants_meta":{"raw":{"variants":["CL AGN hosts match normal quasars, not special galaxies","Changing-look AGNs: hosts ordinary, central engine switched","Post-starburst hints but no distinct host galaxy for CL AGNs","Narrow lines stay put: CL AGN flips come from accretion","105 changing-look AGNs: quiet hosts, active black holes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000124,"raw_usage":{"total_tokens":1007,"prompt_tokens":874,"completion_tokens":133,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":618,"completion_tokens_details":{"reasoning_tokens":42}},"tokens_in":618,"tokens_out":133,"duration_ms":2063,"temperature":1.0,"reasoning_tokens":42,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T00:31:44.190920+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Spatially resolved integral-field spectroscopy of these CL AGN hosts: if the derived stellar mass, sSFR, or Hδ strength depends on the assumed aperture scaling, or if strong radial age/metallicity gradients are present, the tied-host assumption fails and the population-level conclusions would need revision.","supporting_citations":[],"review_version":1}