{"id":"b0d37c2b-af25-4211-a0ec-fda52e168627","arxiv_id":"1908.03423","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"The Shark semi-analytic galaxy formation model, combined with EAGLE-informed attenuation curves and energy-conserving FIR re-emission, reproduces galaxy luminosity functions, number counts, UV slopes, SMG redshift distributions, and cosmic SEDs from z=0 to z=10 with a universal IMF.","lead":"Astronomers combined two computer models, one that simulates how galaxies form and one that generates their light, to predict what galaxies look like from ultraviolet to far-infrared wavelengths across 10 billion years of cosmic history. The result is a single framework that reproduces many observed properties of galaxies, including the disputed population of bright submillimetre galaxies, without changing the stellar initial mass function.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Simultaneous UV-to-FIR match relies on an untested assumption that the local dust-to-metal scaling holds unchanged to z=10; the paper itself flags this as its central fragility.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: the paper's strongest claim, that Shark reproduces the FUV-to-FIR emission without invoking a varying IMF or a redshift-dependent dust-to-metals scaling, hinges on extrapolating a local empirical relation (Remy-Ruyer et al. 2014) to z=10. This is indeed the least secure link in the attenuation chain. The paper is commendably transparent about this limitation, and it tests several dust-scaling variants, but the high-redshift UV LF and UV slope agreement is obtained only with the RR14-steep variant, and no model with redshift-evolving dust is explored. The paper cites competing models that disagree by up to 1.5 dex at z=8-10, so the assumption is not merely a conservative choice but a potentially consequential one. A dedicated test with an explicit evolving dust-to-metal model would settle whether the simultaneous match is robust or an artifact of the assumed scaling. Since the paper already presents the assumption as a limitation and the verdict is CONDITIONAL, no change to the reader's verdict is needed; the concern strengthens the case for conditional acceptance rather than full acceptance.","tokens_in":35464,"tokens_out":5721,"duration_ms":61506,"concrete_test":"Re-run the Shark+ProSpect pipeline replacing the invariant Remy-Ruyer et al. (2014) dust-to-metal relation with an explicit redshift-dependent dust model, e.g., following Vijayan et al. (2019) or Popping et al. (2017), while keeping all other Shark parameters and the universal IMF fixed. Recompute the z=3-10 rest-frame UV luminosity functions (Fig. 14), the z<=8 UV slopes (Fig. 15), and the 850 micron number counts and redshift distribution (Figs. 16-17). If the simultaneous agreement with observations is preserved within the quoted uncertainties, the invariant-scaling assumption is not load-bearing; if the UV LF bright end or the 850 micron counts move systematically outside the observed error bars, the central claim is conditional on the local scaling holding to z=10.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that Shark+ProSpect reproduces the FUV-to-FIR galaxy population from z=0 to z=10 with a universal IMF, without invoking a varying IMF or a redshift-dependent dust-to-gas metallicity scaling. This claim rests on assuming the Remy-Ruyer et al. (2014) dust-to-metal ratio versus gas metallicity relation, including the RR14-steep variant, is invariant with redshift (Section 2.2; Section 6: 'we assume this relation to hold out to z=10'). The paper itself flags that competing dust models predict strong evolution: Vijayan et al. (2019) find the dust-to-metal ratio at z=8-10 to be about 1.5 dex lower than at z=0 at fixed stellar mass, while Popping et al. (2017) find little evolution. The high-redshift UV luminosity function and UV slope comparisons (Figs. 14 and 15) only succeed with the RR14-steep variant; the default RR14 model over-attenuates bright galaxies at z>3. Thus the headline agreement at z=3-10 depends on a specific, untested choice of both the shape and the redshift invariance of the dust scaling. If the true dust-to-metal ratio evolves with redshift, stellar population age, or environment, the simultaneous match could be an artifact of the assumed scaling rather than a success of the underlying Shark galaxy formation model. The statement in Section 5.1 and the conclusions that no varying IMF or redshift-dependent dust scaling is needed is therefore the least secure part of the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper combines the Shark semi-analytic model of galaxy formation with the ProSpect SED generation code to predict galaxy emission from the FUV to the FIR over 0 ≤ z ≤ 10. Dust attenuation is built from radiative-transfer-derived Charlot & Fall parameters calibrated on EAGLE galaxies, dust masses are computed from gas mass and metallicity using local empirical relations (Rémy-Ruyer et al. 2014), and absorbed light is re-emitted with Dale et al. (2014) templates under energy balance. The authors compare four attenuation variants against a wide set of observations: z=0 FUV-to-FIR luminosity functions, rest-frame K-band and FUV luminosity functions out to z=3 and z=10, UV slopes, number counts from the NUV to 850 µm, the redshift distribution of bright 850 µm galaxies, and the cosmic SED. The central claim is that Shark reproduces these observations simultaneously with a universal Chabrier IMF, without retuning to SED data and without invoking a varying IMF or a redshift-dependent dust-to-metal scaling.","tokens_in":35777,"tokens_out":9149,"duration_ms":106358,"significance":"If the central claim holds, this is a significant advance: it would show that a full cosmological galaxy formation model can simultaneously match UV-to-optical and FIR observations without a varying IMF, and it would provide a generative, panchromatic tool for interpreting upcoming surveys. The paper's strengths include the open-source software infrastructure (Shark, ProSpect, Viperfish), the systematic comparison in 27 bands, the explicit decomposition into disk, merger-driven bulge, and disk-instability-driven bulge contributions, and the unusually candid discussion of model limitations, including the dust-to-metal scaling and dust temperature assumptions. The claim of being 'unprecedented' is plausible but needs sharper qualification because some comparisons (K-band, near-IR) partly inherit the model's tuning to the stellar mass function, and because the headline high-redshift result is obtained with a specific steep variant of the local dust-to-metal relation.","major_comments":[{"comment":"The central conclusion that no redshift-dependent dust-to-metal scaling is needed is not fully established. The attenuation chain assumes the local Rémy-Ruyer et al. (2014) dust-to-metal ratio versus gas metallicity relation is invariant out to z=10, as the paper itself states in §6. The manuscript also cites Vijayan et al. (2019), who predict a dust-to-metal ratio about 1.5 dex lower at z=8–10 at fixed stellar mass, while Popping et al. (2017) find little evolution. Because the high-redshift UV LFs and UV slopes are the main probes of this regime, the claimed simultaneous match could be an artifact of the assumed redshift-invariant scaling rather than a success of Shark's baryon physics. Please add a sensitivity test in which the dust-to-metal ratio evolves with redshift within the published range and re-evaluate the UV LFs, the 850 µm counts, and the SMG redshift distribution; if the conclusions survive, state this explicitly, and otherwise soften the abstract and conclusion claims accordingly.","section":"§2.2, §6, Figs. 14–15"},{"comment":"The high-redshift agreement is obtained with the EAGLE-τ RR14-steep variant, not with the model labelled as the default (EAGLE-τ RR14). At z=3–6 the default RR14 and fdust-const models over-attenuate bright galaxies, with differences of up to ~1.5 mag at z=3 and ~2 mag at z=6 in the disk contribution to the UV LF. The paper is transparent about this in the text and captions, but the abstract and conclusions present the high-z match without saying that it depends on selecting one particular steep variant within the local scatter of the Rémy-Ruyer relation. This makes the 'without retuning' claim less clean: choosing among variants after inspecting the comparisons is a mild form of model selection even if no parameter is re-fit. Please clarify throughout that the headline high-redshift result refers to a specified steep variant, and discuss the extent to which that choice is motivated a priori by De Vis et al. (2019) versus selected after comparing with the observed LFs.","section":"§4.3, Fig. 14"},{"comment":"The FIR predictions that support the 'no varying IMF' conclusion assume a fixed Dale et al. (2014) dust SED with two constant effective dust temperatures (≈20–25 K for the diffuse ISM and ≈50–60 K for birth clouds) at all redshifts. The 850 µm number counts and the redshift distribution of bright 850 µm galaxies are temperature-sensitive, and the paper itself notes that GALFORM's self-consistently computed dust temperature evolves with redshift. Please quantify how much the predicted 850 µm counts and N(z) change if the dust temperature or the αSF parameters evolve with redshift within a plausible range, or explicitly restrict the claim of no varying IMF to the fixed-temperature model adopted here. As written, the FIR side of the headline claim contains an additional untested invariance assumption beyond the dust-to-metal scaling.","section":"§3.2, §5.1, Figs. 16–17"}],"minor_comments":[{"comment":"The lightcone area is printed as '107 deg2' and the caption says '107 deg2 deep lightcone'. If this is meant to be 10^7 deg2, it exceeds the whole sky (~4.1×10^4 deg2) and is unphysical; if it is meant to be 10^7 arcmin2 or another value, please state the correct units explicitly.","section":"§5.1, Fig. 16"},{"comment":"The near-infrared and K-band luminosity function agreement is partly inherited from tuning Shark to the z=0, 1, and 2 stellar mass functions and to the mass-size relations; the text acknowledges this ('not necessarily surprising'), but the abstract's broad 'without retuning' claim should be qualified so readers can see which comparisons are truly independent of the tuning.","section":"Abstract, §4.2, §4.3"},{"comment":"The comparisons are assessed visually, with qualitative statements such as 'excellent' and 'very good'. Adding a quantitative goodness-of-fit measure (including model cosmic variance and observed systematic uncertainties) would make the 'unprecedented agreement' claim easier to evaluate.","section":"Figs. 10–17"},{"comment":"The caption should define the 'Remy-Ruyer14 - XCO Z' and 'Remy-Ruyer14 - XCO MW' variants, since these labels are not self-explanatory without going to the Rémy-Ruyer et al. paper.","section":"Fig. 1"},{"comment":"The y-axis of the GALEX UV LFs is presented without normalization by bin size, while later figures normalize by bin size; this inconsistency should be flagged in the caption to avoid confusion.","section":"Fig. 8 caption and figure"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope and the modelling effort is substantial. The main issue is that several 'no additional physics' claims rest on untested redshift-invariance assumptions that the authors themselves flag. I would be willing to accept after the sensitivity tests and careful softening of the abstract/conclusion claims are carried out; I do not see a fatal internal error requiring rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this is a large, honest, and technically serious paper that claims a first in SAMs—simultaneously reproducing observed galaxy emission from FUV to FIR at z=0–10 with a universal IMF and no retuning to SED data. I think the claim is mostly earned, but the high-redshift part leans on a specific assumption that the authors themselves flag.\n\nWhat is new: the assembled pipeline—Shark SAM, ProSpect/Viperfish SED generation, EAGLE-based Charlot & Fall parameters dependent on dust surface density, and the Remy-Ruyer dust-to-metal relation—is new as a package. The comparison set is broad: 27-band z=0 LFs, rest-frame K-band to z=3, FUV to z=10, UV slopes, number counts through 850um, SMG redshift distributions, and CSEDs. The model is not tuned to these observables, and several key ingredients (EAGLE attenuation, local dust scaling) are independent. The component breakdown (disks vs bulges) is illuminating. The codes are open source; that helps reproducibility.\n\nSoft spots in proportion: the central fragility is real. The RR14 dust-to-metal relation is local and assumed invariant to z=10; the paper says this, and cited models disagree by ~1.5 dex at z=8–10. If the true relation evolves, the simultaneous fit could be partly an artifact. Also, the best high-z UV LF and UV slope agreement comes from the RR14-steep variant—default RR14 over-attenuates bright galaxies at z>3. That choice is presented transparently, though it qualifies the 'no retuning' framing. The caveats on cosmic variance and model uncertainty are acknowledged but not quantified.\n\nThe z=0 FUV-to-FIR LFs and number counts are a genuine, broad success, and the 850-um redshift distribution with a universal IMF is an important counterpoint to earlier GALFORM results. The paper is careful to note that the answer is model-dependent.\n\nWho would get value: SAM and galaxy-formation modellers, survey-forecast people, anyone building lightcones. It deserves a serious referee. My recommendation: engage with it; ask for a more explicit test of redshift-dependent dust scaling (or clearer presentation of the steep-variant dependence) and for uncertainty estimates on the model curves. This is not a desk reject.","headline":"A serious, broad panchromatic SAM validation that mostly delivers, with the high-z agreement resting on an unproven dust-to-metal scaling that the authors themselves flag.","tokens_in":36426,"tokens_out":2448,"would_cite":true,"duration_ms":27365,"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":"The Shark galaxy-formation model reproduces observed UV-to-FIR galaxy emission from z=0 to z=10 without changing the stellar initial mass function.","keywords":["galaxy formation","semi-analytic model","spectral energy distributions","dust attenuation","luminosity function","submillimetre galaxies","cosmic SED","far-infrared"],"falsifier":"Measure the dust-to-metal mass ratio of galaxies at z=4-8 using ALMA dust-continuum observations and JWST/NIRSpec gas metallicities; if the ratio falls roughly 1.5 dex below the local Remy-Ruyer relation at fixed metallicity, as some dust-formation models predict, the model's high-redshift UV slopes and 850-micron counts would not be reproduced with the assumed invariant scaling.","tokens_in":35250,"feed_emoji":"🌌","tokens_out":6933,"duration_ms":66345,"temperature":0.7,"pith_summary":"This paper claims that a state-of-the-art semi-analytic model of galaxy formation, Shark, can reproduce observed galaxy emission across the whole far-ultraviolet to far-infrared range and across cosmic time, from z=0 to z=10, in one go. The matching is done by attaching to each simulated galaxy a spectral energy distribution built from its star-formation and metal-enrichment history, then applying dust attenuation scaled by galaxy dust surface density and re-emitting absorbed light in the infrared. The authors report simultaneous agreement with z=0 luminosity functions in 27 bands, rest-frame K-band luminosity functions to z=3, rest-frame UV luminosity functions to z=10, UV slopes to z=8, number counts including the disputed 850-micron counts, the redshift distribution of bright submillimetre galaxies, and the z=0 to z=1 cosmic SED. The central point is that this is achieved with a universal Chabrier initial mass function and with no redshift-dependent adjustments to the dust-to-metal ratio. If right, it means earlier models that required a top-heavy IMF in starbursts to fit the far-infrared were resolving a tension specific to their assumptions, not a universal feature of galaxy formation.","feed_headline":"Shark model matches galaxy light from UV to 850 µm","feed_subtitle":"Without retuning or a varying IMF, one model fits luminosity functions, number counts, and the cosmic SED.","key_machinery":"The load-bearing object is the attenuation chain that maps a simulated Shark galaxy's physical state to an SED. Dust masses come from the local dust-to-metal vs metallicity scaling of Remy-Ruyer et al. (2014), assumed to hold to z=10. Diffuse ISM optical depth and power-law index, tau_ISM and eta_ISM, are sampled from the EAGLE radiative-transfer parametrisation of Trayford et al. (2019) as functions of dust surface density, separately for disks and bulges; birth-cloud optical depth scales with the dust surface density of molecular clouds following Lacey et al. (2016). The absorbed light is re-emitted using Dale et al. (2014) templates at fixed effective temperatures for diffuse ISM and birth clouds, ensuring energy balance. This machinery produces attenuation that grows with stellar mass, peaks at z~1-2, and declines at high redshift as galaxies become metal poor, which is what allows the UV and FIR to fit simultaneously.","core_discovery":"The paper's central claim is that Shark, a semi-analytic galaxy formation model with a universal Chabrier IMF, is capable of reproducing the observed panchromatic emission of galaxies from the FUV to the FIR across 0<=z<=10 without retuning or changing physical assumptions. The authors construct SEDs from simulated star formation and metallicity histories, compute dust masses from a local empirical relation between dust-to-metal ratio and gas metallicity (Remy-Ruyer et al. 2014), derive diffuse-ISM attenuation parameters from radiative transfer calculations of EAGLE galaxies (Trayford et al. 2019), and re-emit attenuated light in the infrared using energy-conserving Dale et al. (2014) templates. They find the model matches observed luminosity functions, number counts, UV slopes, the redshift distribution of bright 850-micron galaxies, and the cosmic SED. Their key discovery is that the long-standing tension between simultaneously reproducing UV-optical and FIR emission, which earlier models resolved by invoking a top-heavy IMF in starbursts, is not required here; the answer is model dependent. They attribute the difference to the combination of realistic gas metallicities, galaxy sizes, dust surface densities, attenuation curves, and dust temperatures in Shark.","pith_inferences":["If the local dust-to-metal scaling is truly invariant, then high-redshift UV-selected galaxies should be nearly dust-free at fixed stellar mass, a prediction that can be tested with ALMA dust-continuum observations of z~6-8 galaxies.","The model's success suggests the earlier need for a top-heavy IMF in other models was partly due to those models' gas metallicities and sizes, not an intrinsic constraint; applying the same attenuation prescription across different galaxy-formation models would isolate the cause.","The fixed dust temperatures adopted for re-emission mean the model's 850-micron counts are sensitive to the assumed dust SED, so allowing a redshift-dependent dust temperature would sharpen the claim of a simultaneous fit.","A direct extension is to use the same machinery to generate mock catalogues for JWST and Euclid to predict colour-selection biases and photometric-redshift systematics."],"forward_implications":["The model can be used to build panchromatic lightcones for upcoming surveys, since the SED machinery requires no retuning when new bands or redshifts are considered.","The 850-micron number counts and the redshift distribution of bright submillimetre galaxies are reproduced without a top-heavy IMF, so current and future submm surveys can be interpreted within a universal-IMF galaxy formation model.","Predicted cosmic SEDs at z>1 are genuine predictions awaiting observation; if JWST and ALMA confirm the steep UV slopes and FIR dominance by starbursts, the model's attenuation prescriptions will be validated.","The observed cosmic SEDs at z<=1 are matched within about 0.1-0.15 dex, providing a benchmark for interpreting the extragalactic background light.","The model demonstrates that complex, non-parametric star-formation histories from a semi-analytic model can be used directly for SED generation, rather than assuming simple analytic forms."],"supporting_citations":[{"why":"Supplies the local dust-to-metal versus gas-metallicity scaling used to assign dust masses to simulated galaxies; the model assumes it holds to z=10.","marker":"Remy-Ruyer et al. (2014)"},{"why":"Provides the EAGLE radiative-transfer parametrisation of Charlot & Fall attenuation parameters as functions of dust surface density, the key link between galaxy properties and extinction.","marker":"Trayford et al. (2019)"},{"why":"Supplies the two-phase dust attenuation model (birth clouds plus diffuse ISM) adopted for SED generation.","marker":"Charlot & Fall (2000)"},{"why":"Supplies the FIR dust emission templates used to re-emit attenuated light with energy balance.","marker":"Dale et al. (2014)"},{"why":"Provides the stellar population library used to generate intrinsic SEDs from star-formation and metallicity histories.","marker":"Bruzual & Charlot (2003)"},{"why":"Supplies the birth-cloud optical depth scaling with dust surface density and represents the GALFORM case requiring a varying IMF.","marker":"Lacey et al. (2016)"},{"why":"Established the original UV-to-FIR tension that motivated a top-heavy IMF in starbursts; this paper argues that tension is model dependent.","marker":"Baugh et al. (2005)"},{"why":"Provides observed z=0 luminosity functions across many bands used as the primary local comparison.","marker":"Driver et al. (2012)"},{"why":"Supplies the observed redshift distribution of bright 850-micron galaxies, a key test for the submillimetre population.","marker":"Wardlow et al. (2011)"},{"why":"Supplies observed 850-micron number counts used to test the model's far-infrared predictions.","marker":"Geach et al. (2017)"}],"fun_headline_variants":["Shark matches galaxy light from UV to 850 µm across z=0–10","No IMF tweaks: Shark fits UV to FIR galaxy emission","From far-UV to sub-mm: one model, no retuning, all redshifts","Shark reproduces galaxy SEDs z0–z10 without changing IMF","No IMF crisis: Shark fits UV–FIR galaxy emission at all z"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The model assumes that the relation between dust mass, gas metals, and gas metallicity measured in nearby galaxies holds unchanged out to redshift 10; if dust production or destruction changes that ratio with time, the claimed joint fit to ultraviolet and 850-micron observations could be an artifact of the assumed scaling.","fun_headline_variants_meta":{"raw":{"variants":["Shark matches galaxy light from UV to 850 µm across z=0–10","No IMF tweaks: Shark fits UV to FIR galaxy emission","From far-UV to sub-mm: one model, no retuning, all redshifts","Shark reproduces galaxy SEDs z0–z10 without changing IMF","No IMF crisis: Shark fits UV–FIR galaxy emission at all z"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000926,"raw_usage":{"total_tokens":4105,"prompt_tokens":1220,"completion_tokens":2885,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":836,"completion_tokens_details":{"reasoning_tokens":2783}},"tokens_in":836,"tokens_out":2885,"duration_ms":22256,"temperature":1.0,"reasoning_tokens":2783,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:12:36.073503+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the dust-to-metal mass ratio of galaxies at z=4-8 using ALMA dust-continuum observations and JWST/NIRSpec gas metallicities; if the ratio falls roughly 1.5 dex below the local Remy-Ruyer relation at fixed metallicity, as some dust-formation models predict, the model's high-redshift UV slopes and 850-micron counts would not be reproduced with the assumed invariant scaling.","supporting_citations":[{"cited_title":"G., Baugh C","cited_arxiv_id":null,"evidence_quote":"Supplies the birth-cloud optical depth scaling with dust surface density and represents the GALFORM case requiring a varying IMF."},{"cited_title":"P., Robotham A","cited_arxiv_id":null,"evidence_quote":"Provides observed z=0 luminosity functions across many bands used as the primary local comparison."},{"cited_title":"L., Smail I., Coppin K","cited_arxiv_id":null,"evidence_quote":"Supplies the observed redshift distribution of bright 850-micron galaxies, a key test for the submillimetre population."},{"cited_title":"E., Dunlop J","cited_arxiv_id":null,"evidence_quote":"Supplies observed 850-micron number counts used to test the model's far-infrared predictions."}],"review_version":1}