{"id":"ec6368d8-f7f0-4837-a831-78f3a889bc83","arxiv_id":"1908.03720","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"New empirical SED templates for 41 AGNs and 72 Seyferts give photometric redshifts for Bootes AGNs that match or beat a prior library, with smaller flux residuals.","lead":"This paper presents 41 new spectral energy distributions of active galactic nuclei built from archival UV-to-mid-infrared spectra, plus 72 synthetic Seyfert templates. The authors show these templates produce photometric redshifts for Bootes field AGNs that are comparable to or better than existing libraries.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed SED accuracy may rest on the crude assumption that single multiplicative rescaling of spectra taken at different epochs yields a valid continuous SED; a direct test of scaling consistency is needed.","rationale":"The reader's weakest assumption—that multiplicative rescaling of multi-epoch archival spectra produces a valid SED—is indeed the most load-bearing premise. The paper's photo-z validation is genuine and provides strong support for utility, but it is not designed to detect a smooth, wavelength-dependent distortion introduced by the scaling method: broadband filters and the added 10% error can absorb such distortions, and the residual test, while useful, is a population-average that may not isolate this effect. The authors are transparent about the crudeness of the approximation, but the paper does not quantify how close the overlap-region ratios are to constant, nor does it test the sensitivity of the photo-z metrics to this assumption. The proposed test would directly settle whether the rescaling introduces shape errors large enough to affect the claimed accuracy. I agree with the reader's conditional verdict: the paper is a valuable resource and the external validation is credible, but a specific, testable caveat remains. The verdict should remain CONDITIONAL (no change) because the concern, while real, is not demonstrated to be fatal; it is a gap in evidence that the authors could close with an additional analysis.","tokens_in":27325,"tokens_out":6325,"duration_ms":74975,"concrete_test":"For each AGN with overlapping spectra (e.g., UV/optical, optical/NIR), compute the ratio of the two input spectra across the region of overlap before applying any scaling. If the ratio deviates from a constant by more than the claimed ~9% residual on typical scales, the single multiplicative scaling has introduced a wavelength-dependent shape error. Quantify the median and maximum deviation across the 41 AGNs. Then, for a subset where deviations are large, rebuild the SED using a wavelength-dependent scaling function (e.g., linear in log λ) and re-run the EAzY photo-z test on the Bootes sample. If σNMAD or residuals improve, the current SEDs are biased by the scaling assumption; if they do not, the assumption is validated for this purpose.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that the new SEDs are functional models for AGNs, validated by photo-z scatter of σNMAD = 0.096(1+z) and 9% flux residuals—depends on the construction in §3.1, where each archival spectrum is multiplied by a single scalar to produce a continuous SED. The authors explicitly caution that variability and aperture bias have wavelength dependence within individual input spectra. If the true flux ratio between two overlapping spectra varies across the overlap region, a single scaling factor will distort the spectral shape. The photo-z test in §5 uses broadband photometry that averages over wavelength, and EAzY adds 10% flux uncertainty and a template error function that can partially absorb smooth shape distortions. The residual metric is computed at fixed spectroscopic redshift, which is more sensitive, but the residuals are averaged over the Bootes sample and may not isolate rescaling-induced errors. For 7 AGNs the scaling factors exceed a factor of 3, indicating large epoch-to-epoch differences that could plausibly be wavelength dependent. The paper does not quantify how close the overlap-region flux ratios are to a constant. Without this, the claim that the templates are accurate models is not fully established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs spectral energy distributions (SEDs) for 41 AGNs by combining archival spectrophotometry and photometry over 0.09–30 μm, with extensions into X-ray, far-infrared, and radio for some objects. Individual spectra are multiplicatively scaled to form continuous SEDs; gaps are filled with simple models, and 72 Seyfert composites are made by mixing AGN and galaxy SEDs. The authors validate the library by fitting photometric redshifts for 2058 Boötes AGNs and comparing with existing libraries, reporting comparable or better performance (σNMAD = 0.096, median reduced χ² ≃ 1, ~9% typical flux residuals).","tokens_in":27468,"tokens_out":10402,"duration_ms":107414,"significance":"If the SEDs are reliable, they are a valuable public resource with higher resolution and broader coverage than earlier libraries, useful for photo-z, AGN selection, and SED modeling. The photo-z test is a genuine external validation on an independent sample, and the comparison with Ananna et al. (2017) is a strong benchmark. The templates and synthetic photometry are made available via a DOI. The main caveat is the dependence of the SED construction on subjective multiplicative scaling of archival spectra, so the shapes of individual SEDs, particularly those with large epoch-to-epoch flux differences, remain less firmly established.","major_comments":[{"comment":"The construction of the SEDs relies on multiplicative scaling of archival spectra taken at different epochs and with different apertures, and the paper explicitly cautions that this is a crude approximation because variability and aperture bias have wavelength dependence. For 7 of the 41 AGNs the scalings exceed a factor of 3. The external photo-z validation in §5 supports the utility of the templates, but the fitting includes 10% flux uncertainties and a template error function that can absorb smooth shape distortions, and the residual metric is averaged over the sample. Since the central conclusion that these are functional models is broader than photo-z utility alone, the paper should provide a quantitative test that the scaling does not introduce wavelength-dependent biases in the SED shapes. I request either (i) a presentation of the distribution of flux ratios in the overlap regions of the input spectra before scaling, or (ii) a demonstration that the photo-z metrics and fixed-redshift residuals are robust to the removal of the 7 AGNs with the most extreme scalings. Without such a test, the shape fidelity of the SEDs, especially for the extreme-scaling objects, is not fully established.","section":"§3.1, Figure 2"}],"minor_comments":[{"comment":"The word 'polynominals' is a typo and should read 'polynomials'.","section":"§3.1"},{"comment":"The seven AGNs with scalings outside 0.33–3.0 are named in the text but not identified in the figure; labelling them would aid the reader.","section":"Figure 2"},{"comment":"The extinction ranges are asymmetric (E(B−V) ≤ 0.5 for Ananna et al. (2017) and Brown et al. (2014b), but ≤ 0.2 for the new AGN templates). The text should briefly justify this choice, even though it is conservative.","section":"§5, paragraph beginning 'When fitting in EAzY'"},{"comment":"The notation 'σNMAD = 0.096 × (1 + z)' is ambiguous because σNMAD is already defined as a normalized scatter; consider writing 'σNMAD = 0.096 (i.e., a redshift uncertainty of 0.096(1+z))'.","section":"§6"},{"comment":"The 'SED λ range' column uses exponential notation (e.g., '9.0×10−2−6.4×101'); decimal notation (0.09–64) would be clearer.","section":"Table 1"},{"comment":"The greybody model is said to have four free parameters, but the parameters are not listed; please specify them.","section":"§2.3"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a useful contribution and the external photo-z validation is genuinely valuable, but I would like to see the requested robustness test on the scaling procedure before acceptance. The use of Brown et al. (2014b) galaxy SEDs is appropriate, though it is a self-citation by the first author; this does not affect the validity of the results."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version: this is a practical resource paper, and the resource is real. The 41 individual SEDs are the first to use actual spectrophotometry to cover 0.09–30 µm for single AGNs, with X-ray and far-IR extensions where available, and the 72 Seyfert composites fill a genuine gap. The photo-z validation on 2058 Bootes AGNs is a real external test—the templates were not built from Bootes data—and the results (σNMAD = 0.096(1+z), median χ2 ≈ 1, ~9% flux residuals) support the claim that these are usable models for wide-field surveys. They are comparable to Ananna et al. (2017), and a bit better at high z and in residuals.\n\nWhat is most honest about the paper is that it does not hide the construction's weak point: Section 3.1 openly says multiplicative scaling of spectra taken years apart is a crude approximation. That is not a buried caveat, it is in the text. The stress-test worry—that a single scalar cannot capture wavelength-dependent variability or aperture bias, potentially distorting the SED shape—has some teeth. Seven objects required scalings beyond 0.33–3.0, and the paper does not quantify how close the overlap-region flux ratios are to constant. But the photo-z test is exactly the kind of downstream test that would expose serious shape distortion, and the fixed-redshift residuals (9% median, 3% outliers beyond 3σ) suggest the distortion is not catastrophic. The EAzY template error function can absorb some smooth shape error, so the test is not airtight, but the burden is on the user to be careful rather than on the claim to be false.\n\nReal soft spots, in proportion: no per-wavelength uncertainties ship with the SEDs, and users who want to fit these properly need error estimates; the greybody fits use four parameters on ≤6 photometric points, so the far-IR part is illustrative; the red AGN SEDs are few and lean on toy models and polynomial interpolation. None of these sink the central claim.\n\nThe circularity charge does not stick: the Bootes test data are independent of the template construction, and the Brown et al. (2014b) galaxy SEDs are a published, public resource.\n\nWho this is for: anyone doing photo-z or SED fitting for AGN surveys—WISE, eROSITA, SPHEREx. It deserves a serious referee. The main thing I would ask for in revision is SED error estimates, or at least a more automated and reproducible stitching pipeline.","headline":"Genuinely useful SED library that passes a real external photo-z test; the construction is openly hand-stitched and lacks published error estimates, but the central practical claim holds.","tokens_in":28175,"tokens_out":2629,"would_cite":true,"duration_ms":27802,"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":"New empirical SEDs for 41 active galactic nuclei, built from archival spectra and matched-aperture photometry, match the colors of AGNs seen by wide-field surveys well enough to deliver photometric redshifts as good as or better than…","keywords":["galaxies: active","quasars: general","galaxies: Seyfert","quasars: emission lines","galaxies: distances and redshifts","spectral energy distributions","photometric redshifts","AGN templates"],"falsifier":"Compare each rescaled template SED to a contemporaneous, well-sampled SED of the same object (or to the object's measured flux densities at a single epoch across many bands); if the rescaled template deviates from the simultaneous measurement by more than the quoted 9% residual level in ways that correlate with the amplitude of the multiplicative scalings, then the stitching assumption is not safe.","tokens_in":27016,"feed_emoji":"🔭","tokens_out":9943,"duration_ms":86915,"temperature":0.7,"pith_summary":"This paper assembles spectral energy distributions (SEDs) for 41 active galactic nuclei (AGNs) by stitching together archival spectra from the X-ray to the radio, anchoring them to matched-aperture photometry, and filling gaps with greybody and power-law models. It then mixes four Seyfert nuclei with galaxy SEDs to produce 72 additional Seyfert templates. The central claim is that these empirical templates are functional models for AGNs detected by current wide-field surveys: when used to compute photometric redshifts for 2058 AGNs in the Bootes field, they yield a scatter of $\\sigma_{\\mathrm{NMAD}} = 0.096 \\times (1+z)$, median reduced $\\chi^2 \\simeq 1$, and median flux residuals of about 9% over restframe 0.1\\,--\\,3µm. This matters because broad-band AGN colors are shaped by real spectral features\\,—\\,the big blue bump, the 1µm inflection, silicate emission, strong emission lines\\,—\\,that earlier composite templates often smoothed away, and photo-z codes need templates that reproduce those features.","feed_headline":"New AGN templates rival the best photometric-redshift libraries","feed_subtitle":"Built from archival spectra spanning 0.09–30 µm, they match AGN colors with 9% residuals.","key_machinery":"The central object is the SED template itself, built by a specific stitching procedure. For each AGN, archival spectra from many instruments are multiplicatively scaled—by factors usually between 0.5 and 2.0, but occasionally outside 0.33 to 3.0—until they join into one continuous SED, with overlapping wavelength ranges used to fix the relative scalings and matched-aperture photometry used to anchor the absolute level. Gaps are filled by fitting polynomials to the log flux–log wavelength relation, with greybody models for far-infrared thermal dust emission and power-laws or polynomials for radio flux densities. This construction carries the argument because the photo-z test is sensitive to spectral shape: if the rescaling distorts the shape, the template error function and residual statistics would degrade. The test itself uses a template-fitting photometric redshift code that slides each SED in redshift, and the authors report that the templates produce the redshift posterior, scatter, and residual statistics quoted above.","core_discovery":"The discovery the paper argues for is that a continuous, spectrophotometric SED of a single AGN can be built from archival data taken years apart, and that the resulting library outperforms or matches existing SED libraries when used for photometric redshift estimation. In the authors' presentation, AGN photometric redshifts determined with the new AGN SEDs plus galaxy SEDs have typical scatter $\\sigma_{\\mathrm{NMAD}} = 0.096 \\times (1+z)$, median reduced $\\chi^2$ values of about 1, and typical flux density residuals of 9%, with the exact value depending on wavelength. Compared to fitting with a galaxy-only library, the new templates reduce model colour biases by a factor of roughly 2 to 5 and cut the median absolute residual from 18% to 9%; compared to a recent AGN-optimised library they are comparable or better, particularly at $z>2.5$. The paper's framing is that previous AGN templates traded spectral resolution for coverage, washing out features like silicate emission that matter for broadband colours, while these SEDs preserve those features at the cost of being tied to individual, variable objects.","pith_inferences":["A testable extension the paper leaves implicit: because several templates include far-infrared and radio extensions, they could predict the sub-mm and radio colours of radio-selected AGNs in new surveys, an application the paper only mentions in passing.","The striking $z\\sim4$ feature where Hα enters a broad 3.4µm infrared band turns the templates into a possible redshift discriminator for mid-infrared selected samples; the paper notes the colour effect but does not exploit it as a stand-alone diagnostic.","The Seyfert composites are built from $z\\sim0$ galaxy SEDs, so building analogous composites from higher-redshift galaxy SEDs would naturally extend the library to host-dominated AGNs at $z>1$; the authors explicitly leave this to future work.","The scaling factors themselves encode information about AGN variability amplitude as a function of wavelength; restricting photo-z fits to low-scaling templates would test whether the 9% residual figure depends on the most heavily rescaled spectra."],"forward_implications":["For X-ray selected AGNs at $z<2.5$, photometric redshifts from the new templates are comparable to those from a recent AGN-optimised library, with improved scatter and outlier fractions at $z>2.5$.","For optically and mid-infrared selected AGNs, the new library performs comparably to the X-ray case, indicating it covers the colour space of AGNs selected by different survey techniques.","Median absolute flux residuals drop from 18% (galaxy-only templates) to 9% (new AGN+galaxy library), and the template error function is reduced most strongly at restframe wavelengths below 2000 Å and near 1µm.","Combining the templates with their own error function and a magnitude prior reduces robust scatter from $\\sigma_{\\mathrm{NMAD}}=0.12$ to $0.095$ and the outlier fraction from 37% to 31%, showing the quoted performance is not yet optimised."],"supporting_citations":[{"why":"Supplies the optical/UV spectra that anchor the short-wavelength end of many of the SEDs.","marker":"Shang et al. 2005"},{"why":"Provides near-infrared spectrophotometry used to bridge the 1\\,2.5µm gap in the SEDs.","marker":"Landt et al. 2008"},{"why":"Adds further near-IR spectra, extending coverage and resolution.","marker":"Landt et al. 2013"},{"why":"Galaxy SED library used both to build the 72 Seyfert composites and as the galaxy-only comparison set in the photo-z test.","marker":"Brown et al. 2014b"},{"why":"The recent AGN-optimised template library the new SEDs are directly compared against in the Bootes photo-z test.","marker":"Ananna et al. 2017"},{"why":"Supplies the template-fitting photometric redshift code used for all photo-z measurements in the paper.","marker":"Brammer et al. 2008"},{"why":"Defines the optical, infrared and X-ray AGN selection criteria that build the Bootes test sample.","marker":"Duncan et al. 2018a"},{"why":"Supplies the broadband X-ray spectral analysis and spectra used to extend the SEDs into the X-ray.","marker":"Ricci et al. 2017"},{"why":"Classic composite AGN SEDs used as the shape benchmark for the new templates.","marker":"Elvis et al. 1994"}],"fun_headline_variants":["Archival AGN SEDs cut photometric redshift errors by half","New AGN templates from archival spectra sharpen redshifts","AGN SED library matches best, with 9% flux residuals","Photometric redshifts for AGNs improved with new templates","Broader AGN SEDs reduce model colour biases 2-5x"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole library stands or falls on the assumption that multiplicatively scaling spectra taken at different times and through different apertures yields one faithful SED of the AGN, even though variability and aperture bias act differently at different wavelengths.","fun_headline_variants_meta":{"raw":{"variants":["Archival AGN SEDs cut photometric redshift errors by half","New AGN templates from archival spectra sharpen redshifts","AGN SED library matches best, with 9% flux residuals","Photometric redshifts for AGNs improved with new templates","Broader AGN SEDs reduce model colour biases 2-5x"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000451,"raw_usage":{"total_tokens":2297,"prompt_tokens":999,"completion_tokens":1298,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":615,"completion_tokens_details":{"reasoning_tokens":1209}},"tokens_in":615,"tokens_out":1298,"duration_ms":11438,"temperature":1.0,"reasoning_tokens":1209,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:03:44.463806+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare each rescaled template SED to a contemporaneous, well-sampled SED of the same object (or to the object's measured flux densities at a single epoch across many bands); if the rescaled template deviates from the simultaneous measurement by more than the quoted 9% residual level in ways that correlate with the amplitude of the multiplicative scalings, then the stitching assumption is not safe.","supporting_citations":[],"review_version":1}