{"id":"3604f212-b511-43e9-91c8-19895fbfafab","arxiv_id":"1908.08956","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Using EAGLE and SKIRT radiative transfer, the paper shows that a galaxy's dust attenuation strength and curve slope are largely set by its dust surface density, and that adding a birth-cloud term reproduces the observed attenuation slope versus optical depth relation.","lead":"This paper builds a simple model of how dust reddens galaxies, trained on about 100,000 simulated galaxies from the EAGLE cosmological simulation, and shows that the model reproduces an observed relationship between the shape and strength of dust attenuation. The usefulness is that astronomers can now apply a physically motivated dust-correction recipe to real galaxies, semi-analytic models, and other simulations.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The birth-cloud 'reproduction' of the Salim et al. relation is not yet validated: model δ–A_V is fit to noise-free attenuation, while the observed relation comes from SED inference; the same pipeline must be applied to EAGLE spectra.","rationale":"The reader's weakest_assumption flags the fixed dust-to-metal ratio (f_dust=0.3) and fixed Milky Way extinction curve. That is a genuine limitation for quantitative application to real galaxies, and the paper itself acknowledges it (footnote 5, §5), but the EAGLE-internal relations are not invalidated by it: within the model, Σ_dust is well defined by construction. The more load-bearing gap is the observational comparison for birth clouds. The central novelty is the first reproduction of the observed δ–A_V relation, yet the paper compares a noise-free model fit to an observationally inferred relation without applying the same inference procedure to the model spectra. This is not an accusation of misconduct; it is a validation gap the authors themselves state (§5). A concrete forward-modeling test could settle whether the agreement is physical or methodological. Because the reader already issued CONDITIONAL and noted the comparison is qualitative, this analysis does not move the verdict.","tokens_in":23641,"tokens_out":9155,"duration_ms":97456,"concrete_test":"Generate mock SDSS photometry and/or spectra from the birth-cloud-corrected EAGLE SEDs used in §5 at z=0.1, adding realistic noise and observational selection, then run these mocks through the same SED-fitting code used by Salim et al. (2018, e.g. CIGALE) to infer δ and A_V. Compare the inferred δ–A_V distribution with both Salim et al. and the noise-free fits in Fig. 5. As a secondary check, vary t_disp in Eq. (2) from 5 to 20 Myr to test whether the claimed robustness extends to the birth-cloud dispersal time. If the inferred relation reproduces the observed scatter and median, the claim is validated; if it does not, the current reproduction is method-dependent.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's most novel claim is that adding a birth-cloud prescription to EAGLE reproduces the observed relation between attenuation slope and optical depth 'for the first time in a cosmological simulation' (abstract, §5). The model relation in Fig. 5 is obtained by fitting Eq. 1 directly to noise-free EAGLE attenuation values with representative SDSS errors (Appendix C). The Salim et al. (2018) relation, by contrast, is the product of full SED fitting with its own degeneracies, priors, and sample selection. The authors explicitly note that the same inference procedure applied to observations has not been applied to EAGLE spectra, and they leave this for future work (§5). Until such forward-modeling is done, agreement between a noise-free model relation and an observationally inferred relation can reflect differences in definitions and fitting methodology rather than physics. Additionally, the claimed insensitivity to birth-cloud parameters is demonstrated only for η_BC and f_τ; t_disp is fixed at 10 Myr, so 'merely requiring a boosted attenuation' is conditional on that fixed dispersal time. This is the weakest link in the paper's headline claim; the complementary Σ_dust–attenuation relation for EAGLE itself is comparatively robust and well supported by the simulations.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper distills SKIRT radiative transfer calculations for roughly 100,000 EAGLE galaxies at redshifts z = 0.1, 0.5, 1 and 2 into a two-component screen model of galaxy attenuation. The ISM component is parametrized by the dust surface density, Σ_dust, and the paper reports tight, nearly redshift-independent relations between Σ_dust and both the V-band attenuation strength and the power-law slope of the attenuation curve. A birth-cloud term is then added in post-processing to attenuate stars younger than 10 Myr, and the resulting δ–A_V relation is compared with the observed relation of Salim et al. (2018). The paper claims that this is the first cosmological simulation to reproduce that relation, and that the result is insensitive to the assumed birth-cloud attenuation properties as long as infant stars receive a boosted attenuation. The paper is careful to state several caveats, including the fixed dust-to-metal ratio, the fixed input extinction curve, and the fact that a full forward-modeling comparison to observed SED fits is left to future work.","tokens_in":23937,"tokens_out":6487,"duration_ms":67441,"significance":"If the central claims hold, the Σ_dust-parametrized ISM attenuation model is a valuable practical tool for semi-analytic models and for interpreting galaxy observations, and the redshift-independence of the relations would be an interesting emergent property of the EAGLE model. The study's strengths include the large statistical sample, the explicit treatment of scatter and orientation effects, and the convergence tests in Appendix A, which are a genuine strength. The birth-cloud comparison with Salim et al. (2018) is promising and the paper is honest about its limitations; however, as written, the headline claim that the observed δ–A_V relation is reproduced is not yet fully supported because the model relation and the observed relation are derived through different inference procedures. The robust core of the paper is the EAGLE ISM attenuation–Σ_dust relation itself, which stands independently of the birth-cloud modeling.","major_comments":[{"comment":"The comparison to Salim et al. (2018) is between a model relation obtained by fitting Eq. (1) directly to noise-free SKIRT attenuation values, using representative SDSS errors only as fit weights (Appendix C), and an observational relation obtained from full SED fitting with its own degeneracies, priors, and sample selection. The paper itself states in §5 that applying the same inference procedure to EAGLE spectra is left for future work, so the abstract's claim that the observed δ–A_V relation is reproduced 'for the first time in a cosmological simulation' is stronger than the evidence presented. Without forward-modeling the EAGLE spectra through the same SED-fitting pipeline used for the observations, the agreement could reflect differences in how δ and A_V are defined rather than physical agreement. Please either carry out such a test or explicitly temper the claim to say that the model is consistent with the observed relation under a simple birth-cloud boost.","section":"§5.2, Fig. 5"},{"comment":"The stated robustness of the birth-cloud result is demonstrated only for variations in the birth-cloud spectral slope η_BC and the optical-depth ratio f_τ. The birth-cloud dispersal time t_disp is fixed at 10 Myr, as acknowledged in §6, and the paper notes in §5.2 that variation in t_disp is not explored. Because t_disp determines which stellar populations receive the birth-cloud boost, and because Fig. 4 shows that the infant-star flux fraction varies rapidly with wavelength and sSFR, a different dispersal time could change the shape of the δ–A_V relation, especially at low A_V where infant stars contribute a larger fraction of the light. The conclusion that the result 'merely requires' a boosted attenuation should therefore be restricted to the explored parameter space, or the t_disp dependence should be tested.","section":"§5.2, §6"}],"minor_comments":[{"comment":"The text 'Sloane galaxies' should read 'Sloan galaxies'.","section":"§1"},{"comment":"The sentence 'assuming ηBC = ηshell ISM' appears to contain a typo; it should presumably be 'ηBC = ηshellBC' or 'ηBC = −1.3'.","section":"§5.2"},{"comment":"The phrase 'birth cloud dspersal time' contains a typo and should read 'birth cloud dispersal time'.","section":"§6"},{"comment":"The phrase 'Rec25 differs rom RefHi25' should read 'Rec25 differs from RefHi25'.","section":"Appendix A"},{"comment":"The inset label 'Modified C01 fit' should read 'Modified C00 fit' for consistency with Calzetti et al. (2000).","section":"Appendix C"},{"comment":"The right-hand panel shows a λ^0.2 offset in η_ISM for Recal25 relative to Ref100; this is described as 'moderately well converged' in the text, but given that η_ISM is one of the two model parameters, it would be helpful to state this offset explicitly as a systematic uncertainty in the model's slope calibration.","section":"Fig. A1"}],"recommendation":"major_revision","confidential_remarks":"The paper is a good fit for MNRAS and the EAGLE ISM attenuation relation is a solid, useful contribution. The main issue is the framing of the birth-cloud comparison with Salim et al. (2018): the authors themselves know that a fair comparison requires applying the same inference procedure to EAGLE spectra, and they leave that for future work. This is acceptable as a caveat, but it conflicts with the abstract's 'reproduce the observed relation for the first time' claim. A major revision that either adds the forward-modeling test or rewrites the claim to match the actual evidence would resolve the concern."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the thing. The headline is the ISM calibration; the birth-cloud reproduction is real but not yet proven. What Trayford et al. actually deliver is a two-parameter screen for the ISM attenuation of ~100k EAGLE galaxies, where the V-band optical depth and spectral slope are set by Σ_dust. That relation is tight, nearly redshift-independent, and they show convergence across resolution using the 25 cMpc runs. That part holds up. The scatter they provide, including the inclination dependence, is a useful addition; earlier SKIRT-EAGLE work didn't fold this into a simple predictor.\n\nThe weaker part is the Salim et al. comparison. The model δ–AV curve is obtained by fitting Eq. 1 to noise-free EAGLE attenuation values with representative SDSS errors. Salim et al. is the product of full SED fitting with priors and degeneracies. The authors themselves write that a fairer comparison would apply the same inference to EAGLE spectra, and leave it for future work. So when they say they reproduce the observed relation 'for the first time,' I read that as a plausible but unvalidated claim. The insensitivity to birth-cloud parameters is also only demonstrated for η_BC and f_τ, with t_disp fixed at 10 Myr; 'merely requiring a boost' is conditional on that choice.\n\nThe fixed Zubko extinction curve and fdust = 0.3 are genuine caveats, but the paper is transparent about them, and the main Σ_dust trend is emergent geometry, not an input. I don't see circularity here. No public data or code is released, which slows adoption, but the EAGLE database is public and the method is reproducible.\n\nBottom line: if you work on SED fitting, semi-analytic models, or high-redshift attenuation, the Σ_dust–AV/η relations are worth taking seriously. The birth-cloud claim should be treated as a hypothesis for now. I'd send it to a serious referee: the central result is solid, the weak link is localized and clearly acknowledged, and a good referee could push for the forward-modeling test or a clear caveat in the abstract.","headline":"The Σ_dust–attenuation calibration is solid and useful; the birth-cloud 'reproduction' of Salim et al. is plausible but not yet demonstrated.","tokens_in":24463,"tokens_out":2321,"would_cite":true,"duration_ms":24349,"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":"A galaxy's dust attenuation curve is set by its average dust surface density, a relation that is tight and nearly redshift-independent, and adding a birth-cloud screen reproduces the observed attenuation trend.","keywords":["galaxy attenuation","dust attenuation","radiative transfer","EAGLE simulation","dust surface density","birth clouds","attenuation curve slope","Charlot-Fall screen model"],"falsifier":"Measure dust surface densities from resolved far-infrared dust maps and attenuation slopes from independent SED fits for galaxies at $z \\approx 2$ and $z \\approx 0.1$, and check whether the same $\\Sigma_{\\rm dust}$ produces the same attenuation curve; a systematic offset between redshifts, or a change in slope at fixed $\\Sigma_{\\rm dust}$ beyond the predicted orientation scatter, would falsify the relation.","tokens_in":23461,"feed_emoji":"🌌","tokens_out":13907,"duration_ms":123058,"temperature":0.7,"pith_summary":"This paper distils radiative-transfer calculations for roughly 100,000 simulated EAGLE galaxies at redshifts $z = 2$, $1$, $0.5$ and $0.1$ into a simple statement: a galaxy's ISM attenuation curve---how its own dust dims and reddens its starlight---is well described by a two-parameter power law whose strength and slope are set by the galaxy's average dust surface density, $\\Sigma_{\\rm dust}$. The relation between $\\Sigma_{\\rm dust}$ and attenuation is tight, with scatter attributable to morphology and orientation, and it is nearly the same at every redshift studied. Adding a sub-resolution birth-cloud screen, with boosted attenuation for stars younger than 10 Myr, makes the model reproduce the observed relation between attenuation slope and optical depth, something earlier cosmological simulations had not done. The authors offer the model as a practical replacement for idealised screen or slab geometries when interpreting observations and when assigning dust attenuation in semi-analytic models.","feed_headline":"Dust surface density sets galaxy attenuation curves","feed_subtitle":"The same dust surface density predicts how much and how reddeningly galaxies obscure their starlight from z=2 to today.","key_machinery":"The load-bearing object is a two-component screen model in the style of Charlot & Fall (2000): a diffuse ISM power-law screen, $\\tau_{\\rm ISM}(\\lambda) = \\tau_{\\rm ISM}^{550}(\\lambda/5500\\,{\\rm \\AA})^{-\\eta_{\\rm ISM}}$, plus a birth-cloud screen that applies only to stars younger than 10 Myr. The paper distils the full SKIRT radiative-transfer output by fitting this power law to each galaxy's ugriz-band attenuation, then studies the fitted parameters against the average dust surface density $\\Sigma_{\\rm dust}$, measured either within the projected stellar half-mass radius or as a stellar-mass-weighted average. $\\Sigma_{\\rm dust}$ is the single predictor that carries the argument: it encodes the geometric effects of star-dust mixing that make attenuation differ from a uniform screen, and it explains both the normalisation and the slope of the attenuation curve. The birth-cloud term is added in post-processing below the simulation resolution, with $\\tau_{\\rm BC} = f_\\tau \\tau_{\\rm ISM}$ and a steeper slope $\\eta_{\\rm BC} = -1.3$, and the observed slope-attenuation relation is recovered for any reasonable $f_\\tau \\gtrsim 2$.","core_discovery":"The central claim is that the effective attenuation curve of a galaxy, computed by full radiative transfer through the star-dust geometry of EAGLE galaxies, collapses onto a two-parameter family: $\\tau_{\\rm ISM}(\\lambda) = \\tau_{\\rm ISM}^{550}(\\lambda/5500\\,{\\rm \\AA})^{-\\eta_{\\rm ISM}}$. Fitting this power law to the ugriz attenuation of each galaxy, the authors find that the V-band optical depth $\\tau_{\\rm ISM}^{550}$ and the slope $\\eta_{\\rm ISM}$ both track the average dust surface density $\\Sigma_{\\rm dust}$: more dust per square kiloparsec means stronger and flatter, greyer attenuation, approaching the Milky Way extinction curve at the highest surface densities. The relation is tight enough that $\\Sigma_{\\rm dust}$ alone predicts the ISM attenuation, with residual scatter set by galaxy orientation and morphology, and it is nearly independent of redshift over $z = 2$ to $0.1$. When an unresolved birth-cloud screen with boosted attenuation for stars younger than 10 Myr is added, the model reproduces the observed relation between attenuation slope and V-band attenuation found by Salim et al. (2018), and this result is insensitive to the birth-cloud attenuation prescription as long as the infant-star attenuation is boosted.","pith_inferences":["If real dust-to-metal ratios vary with metallicity or redshift, the same $\\Sigma_{\\rm dust}$--attenuation relation would shift, and mapping that variation could turn the model into a probe of dust production and destruction.","A direct observational test would be to compare resolved $\\Sigma_{\\rm dust}$ measurements with attenuation slopes from SED fitting across galaxy orientations, checking the paper's prediction that edge-on galaxies are greyer at fixed attenuation.","The same $\\Sigma_{\\rm dust}$ parameter could help explain the scatter in the infrared-excess--UV-slope relation, since that scatter is partly driven by the geometric effects the paper quantifies.","The greying-with-dustiness trend suggests that SED fits to very dusty high-redshift galaxies may systematically require shallower attenuation curves than the Calzetti law allows."],"forward_implications":["SED fitting codes can replace arbitrary screen or slab attenuation geometries with a relation keyed to a measurable physical quantity, $\\Sigma_{\\rm dust}$.","Semi-analytic models and other hydrodynamic simulations can assign wavelength-dependent attenuation from gas mass, metallicity and galaxy size, instead of from fixed idealised geometries.","The near redshift-independence implies that the simulated evolution of galaxy attenuation is not driven by evolving dust physics but by galaxies populating different parts of a static $\\Sigma_{\\rm dust}$--attenuation relation.","The predicted scatter means orientation and morphology leave a distinctive imprint: at fixed attenuation, edge-on galaxies should show flatter, greyer attenuation curves.","The birth-cloud result implies that the observed slope--attenuation relation requires only boosted attenuation of infant stars, not fine-tuned birth-cloud dust properties."],"supporting_citations":[{"why":"Provides the two-component screen model---diffuse ISM plus birth clouds---that the paper adapts and tests against SKIRT output.","marker":"Charlot & Fall (2000)"},{"why":"Sets up the SKIRT radiative-transfer processing of EAGLE galaxies, including the fixed dust-to-metal fraction f_dust = 0.3 and the MAPPINGS-III treatment of young stars.","marker":"Camps et al. (2016)"},{"why":"Supplies the SKIRT photometry and property maps from which the ISM attenuation curves and dust surface densities are derived.","marker":"Trayford et al. (2017)"},{"why":"Gives the Milky Way grain extinction curve assumed for the dust, the fixed input extinction for every radiative-transfer run.","marker":"Zubko et al. (2004)"},{"why":"Provides the observed relation between attenuation slope and optical depth that the birth-cloud model is designed to reproduce.","marker":"Salim et al. (2018)"},{"why":"Gives line-of-sight attenuation versus dust surface density measurements in nearby galaxies used as an observational comparison to the EAGLE relation.","marker":"Kreckel et al. (2013)"},{"why":"Demonstrates theoretically how differential obscuration of young and old stars shapes the attenuation curve; the paper compares its ISM-only relation to this result.","marker":"Narayanan et al. (2018)"},{"why":"Generalises the Charlot-Fall screen model with separate ISM and birth-cloud spectral slopes, the basis for the adopted spectral slope of -1.3.","marker":"Wild et al. (2007)"},{"why":"Defines the starburst attenuation law whose modified form, with a power-law tilt, is fitted to the EAGLE curves to obtain attenuation strength and slope.","marker":"Calzetti et al. (2000)"}],"fun_headline_variants":["Dust density alone predicts how galaxies obscure starlight","Galaxy dimming and reddening set by dust surface density","One parameter drives galaxy attenuation curves in EAGLE","Attenuation curve slope and depth track dust surface density","Dust surface density collapses galaxy attenuation to a line"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The model assumes dust always makes up 30 per cent of the metal mass in gas and that dust grains scatter and absorb light like Milky Way dust at every redshift, so a real variation in the dust-to-metal ratio or in grain properties would shift the predicted attenuation relations.","fun_headline_variants_meta":{"raw":{"variants":["Dust density alone predicts how galaxies obscure starlight","Galaxy dimming and reddening set by dust surface density","One parameter drives galaxy attenuation curves in EAGLE","Attenuation curve slope and depth track dust surface density","Dust surface density collapses galaxy attenuation to a line"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000462,"raw_usage":{"total_tokens":2342,"prompt_tokens":1010,"completion_tokens":1332,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":626,"completion_tokens_details":{"reasoning_tokens":1251}},"tokens_in":626,"tokens_out":1332,"duration_ms":11098,"temperature":1.0,"reasoning_tokens":1251,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:23:55.795843+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure dust surface densities from resolved far-infrared dust maps and attenuation slopes from independent SED fits for galaxies at $z \\approx 2$ and $z \\approx 0.1$, and check whether the same $\\Sigma_{\\rm dust}$ produces the same attenuation curve; a systematic offset between redshifts, or a change in slope at fixed $\\Sigma_{\\rm dust}$ beyond the predicted orientation scatter, would falsify the relation.","supporting_citations":[],"review_version":1}