{"id":"c7450da6-0b55-4b55-9356-a98eb882f0e7","arxiv_id":"2411.13270","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":13,"one_line_summary":"A new iterative fitting routine deblends unresolved AGN emission from resolved host-galaxy emission in IFS datacubes by jointly modeling a Moffat PSF and a Sérsic host profile per wavelength slice.","lead":"This paper presents an iterative method that separates the bright, unresolved light of an active galactic nucleus from the fainter, extended light of its host galaxy in integral field spectroscopic data, without needing a separately measured point-spread function. It is validated on mock observations and applied to three galaxies from the MaNGA survey.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Mock validation is self-referential: the mocks inject a Moffat PSF for the AGN, the exact model family the method fits, so the central claim is not yet tested against real PSF shapes.","rationale":"The reader's weakest assumption is exactly the most load-bearing point: the method's only physical handle on the unresolved AGN is the 2D profile of each monochromatic slide, and if that profile is not Moffat, the fitted flux integral (Eq. 4) and the residual cube are systematically wrong. The mocks are constructed with exactly a Moffat PSF (constant β, α from stars), so they can only test numerical convergence, noise, centroid recovery, and host-background approximation within the assumed model family; they cannot test whether the assumed family matches real AGN PSFs. The real-galaxy BPT and line-profile results are internally consistent but have no independent ground truth: no PSF from stars, no high-resolution imaging, and no comparison against a non-parametric PSF recovery. A single model-mismatch experiment replacing the Moffat with an empirical PSF would directly probe this assumption. The paper is otherwise clear, publicly coded, and honest about limitations, so the appropriate outcome is the reader's conditional acceptance pending such a test rather than a rejection; the verdict should remain unchanged.","tokens_in":31789,"tokens_out":4236,"duration_ms":48078,"concrete_test":"Generate a mock cube with the same host galaxies and AGN spectra as the HighR-Dom case (Section 3.1), but set the injected AGN PSF equal to the empirical, non-Moffat PSF of a MaNGA standard star at each wavelength (using the same Yan et al. 2019 stellar observations that motivated the Moffat parametrization). Run AGNdecompose and compute the recovered AGN flux residual relative to the input AGN spectrum. If the residual grows from the reported ~10% to ≳25%, or shows a systematic spectral slope, the method's success is contingent on the Moffat PSF assumption and the abstract's claim is not generally supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"All six mocks in Section 3.1 inject the AGN as a Moffat PSF (Eq. 1) with constant β and α(λ) taken from MaNGA stars, while the host is a Sérsic-dominated background; Section 2.1 then fits that same Moffat+Sérsic model family. The Table 1 residuals (10–25%) therefore measure inversion error inside the assumed family, not robustness to real PSF shape. Real MaNGA PSFs are not exactly Moffat (atmospheric seeing, fiber convolution, possible non-Moffat wings); if the true PSF deviates, Eq. 4's infinite integral of the fitted Moffat mis-estimates the AGN spectrum, and step (e) propagates that error directly into the residual host cube. No independent PSF measurement (e.g., from same-field stars) validates the Section 4 real-galaxy decompositions, and no model-mismatch test is reported. The central 'capable of disengaging' claim is therefore conditional on an untested equivalence between the real AGN PSF and the Moffat family.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an automatic, iterative deblending method for integral-field spectroscopic data that simultaneously models the unresolved AGN as a 2D Moffat profile and the resolved host galaxy as a 2D Sérsic profile in each monochromatic slice. The AGN spectrum is obtained by analytically integrating the fitted Moffat profile (Eq. 4), and the residual cube is taken as the AGN-free host. The method is validated on six mock IFU cubes built from three MaNGA galaxies and a quasar composite, with reported continuum residuals of 10-30% and host emission-line flux residuals of 6-40% depending on spatial resolution and AGN dominance. The authors compare the method with QDeblend3D and apply it to three MaNGA Type 1 AGN, showing BPT shifts and recovered stellar absorption features in the host residuals.","tokens_in":32081,"tokens_out":4363,"duration_ms":49027,"significance":"If the central claim is established, this is a useful addition to the AGN-host deblending toolbox for IFS data, since it avoids external PSF characterization from foreground stars and does not require prior AGN/HG flux ratios. The paper is honest in reporting the residuals of its mock tests, makes the code publicly available, and includes a comparison with an established tool. The real-data demonstrations on MaNGA galaxies are valuable sanity checks, and the recovered stellar absorption features and BEL/NC fitting are qualitatively convincing. However, the quantitative validation is currently internal to the assumed Moffat+Sérsic model family, so the headline claim that the method is 'capable of disengaging' real AGN emission is not yet supported at the level the abstract states.","major_comments":[{"comment":"The mock validation is self-consistent in the model family but not a test of the central claim. The injected AGN PSF is exactly the Moffat profile of Eq. (1), with constant beta and alpha taken from MaNGA stars, and the fitter then assumes the same functional family. Consequently, Table 1 and Figures 3-5 measure inversion error inside the assumed model and do not quantify the effect of a real PSF that deviates from a Moffat form. This matters because Eq. (4) integrates the fitted Moffat profile to infinity: any deviation in the PSF wings, fiber-convolution effects, or truncation by the IFU field of view will propagate directly into the recovered AGN spectrum and, through step (e), into the residual host cube. I request a model-mismatch test, for example injecting an empirical MaNGA stellar PSF or a Moffat profile perturbed by non-Moffat wings, and a quantitative statement of how much Moffat mismatch the method can tolerate.","section":"Section 3.1 versus Section 2.1; Eq. (4)"},{"comment":"The real-data comparison with QDeblend3D compares two decomposition codes, not a recovered spectrum against a known truth. The statements that stellar Ca II H&K absorption lines are recovered in the residual and that BPT spaxels move from the AGN region to the composite/SF region are useful sanity checks, but they do not validate the absolute flux scale of the deblended AGN spectrum or the accuracy of the host residual. An independent PSF model derived from same-field stars, or a spectroastrometric or flux-scaling cross-check, would be needed to support the abstract's claim that the method is capable of disengaging real AGN emission from the host continuum and narrow lines.","section":"Section 4.1 and Section 4.2"},{"comment":"The claim that galaxy morphology has little impact on performance is not supported by the mock design. The HighR and IntR hosts contain spirals and bars that are not well described by a single Sérsic profile, while the LowR host is the most Sérsic-like, yet the LowR cases show the largest EELR residuals (Table 1 and Figures 5-7). The differences among mock cases are degenerate with IFU spatial sampling and field-of-view effects, so the stated conclusion that morphology is not important needs a targeted test that varies morphology while holding resolution and AGN/HG flux ratio fixed.","section":"Section 3.6 and Summary"},{"comment":"The assumption that beta, b_n, n_s, and optionally e and theta vary smoothly with wavelength and can be interpolated from coarse windows is not stress-tested against spectral regions with emission lines or continuum breaks. Since beta enters Eq. (4) multiplicatively through the factor 1/(beta-1), an error in the interpolated beta directly scales the recovered AGN flux. A mock test with a non-monotonic wavelength dependence of beta, for example a perturbation localized at H alpha, would quantify this systematic and would also check the claim that the method does not require interpolation of the PSF between spectral regions.","section":"Section 2.1, steps (a)-(b); Eq. (4)"}],"minor_comments":[{"comment":"The captions use 'quasar-dominated' and 'intermediate quasar' while the body text uses 'AGN-dominated' and 'intermediate AGN'; please reconcile the terminology for consistency.","section":"Figures 3-5 captions"},{"comment":"The residual normalization in Figure 12 is described as 'QDeblend3D minus our method over the input flux', which is not identical to the definition in Eq. (5); please specify the exact normalization used in the comparison.","section":"Figure 12 caption and Eq. (5)"},{"comment":"The reported velocity-shift residuals are non-monotonic in spatial resolution (8% for HighR, 30% for IntR, 8% for LowR); a brief explanation of this non-monotonic behavior would help the reader interpret Figure 7.","section":"Section 3.2.2, velocity-shift residuals"},{"comment":"The citation 'Bruce, Dunlop, Mortlock, Kocevski, McGrath & Rosario Kim' is malformed and should be corrected to the proper author list from the reference list.","section":"Introduction, related-work paragraph"},{"comment":"The paper states the code is public on GitHub but does not specify a version or a DOI; adding a versioned release would improve reproducibility.","section":"Data Availability"}],"recommendation":"major_revision","confidential_remarks":"The core issue is that the validation is self-referential: the mocks inject exactly the Moffat PSF family that the method fits, and the real-data applications lack an independent PSF truth. This is a fixable gap rather than a fatal error, but the abstract currently overstates the evidence. I would be willing to recommend acceptance after the authors add a model-mismatch test, temper the abstract's claim, and address the morphology-support issue. The manuscript fits the journal's scope, and the public code is a positive feature."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know: this is a genuinely different AGN-host deblender for IFS. It fits a Moffat profile for the unresolved AGN and a Sérsic for the resolved host per wavelength slice, without needing a foreground-star PSF or a prior AGN/HG flux ratio, and it ships code on GitHub. The novelty is real relative to QDeblend3D and q3dfit.\n\nWhat it does well: the mock suite spans three MaNGA resolutions and two AGN/host contrasts, and the residuals are reported honestly (AGN continuum 10–30%; host EELR fluxes up to 30–40% at low resolution). The real-data checks are meaningful: recovered Ca II H&K absorption in the host residual, BPT spaxels moving into the star-forming region after subtraction, and AGN spectra that fit with specfit. The QDeblend3D comparison is fair in spirit and shows the method at least matches it on the high-resolution mocks.\n\nThe soft spot is real and load-bearing: every mock injects the AGN as a Moffat profile, and the fitter is a Moffat. So the reported residuals measure inversion error inside the assumed family, not robustness to a real PSF shape. If the actual AGN PSF has non-Moffat wings (or is slightly resolved), Eq. 4's integral of the fitted Moffat to infinity will bias the AGN flux, and because the host is the residual, that bias lands in the host cube. The host side is partially tested — the mock hosts are real MaNGA galaxies with bars and spiral arms — but the PSF side is not. An independent PSF from a field star, or a mock with a Gaussian/empirical PSF injected, would close the gap.\n\nMinor issues: the QDeblend3D comparison on mocks uses only the high-resolution cases, and the real-data comparison never states what input flux ratio QDeblend3D was given. There is no error propagation from the MCMC chains to the deblended spectra, and the mock cubes are not shipped.\n\nWho it's for: anyone doing AGN-host separation in MaNGA, LVM, or JWST IFS programs, and tool developers in that space. It deserves peer review. I would send it out and ask for non-Moffat PSF tests, a field-star check, error bars on the deblended fluxes, and a pinned code release with the mocks. With those, it is a solid contribution.","headline":"A useful, code-shipping AGN-host deblender for IFS, but the mock validation never tests whether a real AGN PSF is actually Moffat.","tokens_in":32716,"tokens_out":5847,"would_cite":true,"duration_ms":60003,"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":"An iterative method fits a Moffat point-spread function for the AGN and a Sérsic profile for the host at every wavelength slice of an IFS cube, separating the two without any prior PSF model.","keywords":["AGN-host decomposition","integral field spectroscopy","Moffat profile","Sérsic profile","quasar deblending","PSF modeling","MaNGA","emission-line decomposition"],"falsifier":"Build a mock IFS cube with a host that has a strong bar or spiral arms and a PSF whose wings deviate from the Moffat form, then check whether the recovered AGN spectrum departs from the true input by more than the reported 10–25% residual; alternatively, compare the recovered AGN flux against high-resolution imaging photometry of the same source.","tokens_in":31581,"feed_emoji":"🔭","tokens_out":7399,"duration_ms":66169,"temperature":0.7,"pith_summary":"This paper presents an iterative method for separating the unresolved AGN (quasar) emission from the spatially resolved host galaxy in integral-field spectroscopic (IFS) datacubes. The method fits a two-dimensional Moffat profile for the AGN point source and a two-dimensional Sérsic profile for the host simultaneously per monochromatic slice. It needs no prior PSF characterization from a foreground star, no external host-light model, and no user-supplied AGN/host flux ratio. On mock IFS cubes the recovered AGN and host spectra agree with the input within roughly 10–25 percent of the total flux, with the best performance in AGN-dominated, well-resolved observations. On three MaNGA galaxies, the residual cube shows host stellar absorption and narrow-line emission while the recovered AGN spectrum retains broad and narrow emission lines.","feed_headline":"Deblender pulls AGN and host spectra apart in IFS data","feed_subtitle":"A per-wavelength Moffat and Sérsic fit separates the bright quasar from its host without any external PSF model.","key_machinery":"The central objects are the Moffat profile $F_{\\rm psf}(i,j)=A_t (1+R(i,j)^2/\\alpha^2)^{-\\beta}$ for the unresolved point source and the Sérsic profile $F_{\\rm ext}(i,j)=I_0\\exp[-b_n((R'/R_e)^{1/n_s}-1)]$ for the host. The method first fits both profiles on spectral windows of ten pixels to boost signal-to-noise, then interpolates the smoothly varying parameters ($b_n$, $n_s$, $\\beta$, and optional ellipticity and position angle) to every wavelength pixel, and finally re-fits each monochromatic slice with a reduced free-parameter set. The AGN spectrum follows from analytically integrating the Moffat profile at each wavelength, and the host cube is the residual. This two-stage fitting design is what allows the method to operate without prior PSF or host-galaxy information.","core_discovery":"The central claim is that a datacube of an AGN-host system can be deblended by modelling each monochromatic slice as the sum of a Moffat profile for the unresolved AGN and a circular Sérsic profile for the host, with the host Sérsic index, shape parameter, and PSF wing index interpolated smoothly from coarse spectral windows to full spectral sampling. The recovered AGN spectrum is the analytic total flux of the fitted Moffat profile, $F_{\\rm AGN}(\\lambda)=\\pi\\alpha(\\lambda)^2 A_t(\\lambda)/(\\beta(\\lambda)-1)$, and the host cube is the residual after subtracting the reconstructed AGN model cube. The paper argues that this disengages the bright, nonresolved AGN emission from the host continuum and its narrow emission lines, and that the accuracy is limited mainly by how well the IFS resolves the PSF and by the AGN-to-host flux contrast.","pith_inferences":["The same per-slice Moffat+Sérsic decomposition could in principle be applied to other unresolved sources in IFS data, such as supernovae or tidal disruption events, whenever the host provides a smooth background.","Because the PSF is built from the AGN itself, the method could be repurposed as a wavelength-dependent empirical PSF estimator for AGN-dominated cubes, useful for calibrating other subtraction tools.","The analytic total-flux formula ties the method to a direct measurement of the AGN spectral energy distribution; comparing that flux against independent narrow-aperture photometry would be a clean external validation.","The single circular Sérsic host is the main structural limitation; adding bar, disk, or asymmetric components would likely reduce the 20–40% residuals seen in the low-resolution and intermediate-flux mock cases."],"forward_implications":["Large IFS surveys can be processed without hand-built PSF models, producing AGN-free host cubes and pure AGN spectra automatically.","The residual host cube is clean enough for standard stellar-population and emission-line analyses, as the recovered BPT spaxels move from AGN to star-forming regions after deblending.","The deblended AGN spectra preserve broad and narrow emission-line properties, with broad and narrow line fluxes recovered to within a few percent in the mock tests.","The method's reported accuracy depends on spatial resolution and AGN/host contrast: in AGN-dominated, well-resolved cases the recovered PSF FWHM is within about 5%, while in low-resolution or host-bright cases the residuals grow to 20–40%."],"supporting_citations":[{"why":"Supplies the analytic PSF profile used to model the unresolved AGN emission.","marker":"Moffat 1969"},{"why":"Supplies the host-galaxy surface-brightness profile used as the resolved background.","marker":"Sérsic 1963"},{"why":"Provides the Moffat FWHM relation used to set and evaluate PSF sizes in the mock tests.","marker":"Trujillo et al. 2001"},{"why":"Justifies using a single Sérsic profile to separate AGN from host light.","marker":"Kim et al. 2008"},{"why":"Presents the comparison tool whose AGN spectra are compared with the new method.","marker":"Husemann et al. 2014"},{"why":"Describes the wavelength-dependent PSF update of the comparison tool, establishing the baseline.","marker":"Husemann et al. 2022"},{"why":"Defines the composite quasar spectra used to build the mock AGN inputs.","marker":"Vanden Berk et al. 2001"},{"why":"Provides MaNGA stellar-library PSF measurements used to set realistic wavelength-dependent PSF inputs for the mocks.","marker":"Yan et al. 2019"},{"why":"Supplies the MaNGA survey and its data cubes used both for mocks and for the three real-galaxy applications.","marker":"Bundy et al. 2015"},{"why":"Implements the affine-invariant MCMC sampler used for the two-dimensional surface-brightness fits.","marker":"Foreman-Mackey et al. 2013"}],"fun_headline_variants":["New deblender separates AGN from host in IFS cubes","No PSF model needed to split AGN from host","Iterative Moffat-Sersic deblending for IFS data","Automatic deblending of AGN from host with Moffat-Sersic","Host-AGN disentangler for IFS data, no priors"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The input datacube is well described by a single Moffat profile for the AGN plus a single circular Sérsic profile for the host, sharing one centroid, and the host and PSF shape parameters vary smoothly with wavelength so they can be interpolated from coarse spectral windows.","fun_headline_variants_meta":{"raw":{"variants":["New deblender separates AGN from host in IFS cubes","No PSF model needed to split AGN from host","Iterative Moffat-Sersic deblending for IFS data","Automatic deblending of AGN from host with Moffat-Sersic","Host-AGN disentangler for IFS data, no priors"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000849,"raw_usage":{"total_tokens":3730,"prompt_tokens":1016,"completion_tokens":2714,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":632,"completion_tokens_details":{"reasoning_tokens":2617}},"tokens_in":632,"tokens_out":2714,"duration_ms":20806,"temperature":1.0,"reasoning_tokens":2617,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:37:56.171870+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build a mock IFS cube with a host that has a strong bar or spiral arms and a PSF whose wings deviate from the Moffat form, then check whether the recovered AGN spectrum departs from the true input by more than the reported 10–25% residual; alternatively, compare the recovered AGN flux against high-resolution imaging photometry of the same source.","supporting_citations":[],"review_version":1}