{"id":"006d3b0d-733d-4c50-a142-3d6b4d898673","arxiv_id":"2501.19327","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"FLAMINGO, the largest of three simulations tested, reproduces observed submillimeter galaxy counts and redshifts with a standard IMF, and predicts TolTEC will find about 80,000 sources at 1.1 mm.","lead":"This paper applies three published recipes that turn simulated galaxy star formation and dust content into submillimeter brightness, then checks them against observations in three large cosmological simulations. The largest simulation, FLAMINGO, matches the observed population of bright dusty galaxies and predicts that the upcoming TolTEC survey will detect about 80,000 such sources in its deep field.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"H13 flux relation chosen over L21/C23 conflicts with EAGLE RT test by factor ~5; all headline numbers depend on this choice, so the central claim is not yet robust.","rationale":"Good-faith reading: the paper is a transparent population-synthesis study. It tests three parametric relations against EAGLE RT and against observed SMGs, and it shows results for both H13 and L21 throughout, with appendices covering resolution, feedback, and cosmology. The central claims, however, are stated as facts (e.g., 'FLAMINGO reproduces the observed redshift distribution and source number counts...' in the abstract) and are only true for H13. The reader's CONDITIONAL verdict already captures this. My stress-test sharpens the condition: the H13 choice passes an observation-driven flux test (Fig. 2) while failing a physically grounded RT test (Fig. 1) by a factor of ~5. The paper never resolves this contradiction. Since the SFRD contribution and TolTEC source counts are directly proportional to the flux normalization, an unresolved factor-of-5 ambiguity in the underlying relation means the quantitative headline numbers should be treated as conditional on H13, not as robust predictions. The proposed RT check on FLAMINGO-like galaxies would settle whether H13 is applicable at FLAMINGO's resolution or whether the apparent match is a coincidence. I therefore keep the reader's CONDITIONAL verdict: the paper is valuable and publishable with the caveat that the H13-based numbers are not yet uniquely supported.","tokens_in":30267,"tokens_out":6370,"duration_ms":57345,"concrete_test":"Run 3D dust radiative transfer (e.g., SKIRT) on ~20 FLAMINGO L1_m8 galaxies at z=2–3 with SFR>100 Msun/yr, using high-resolution zoom re-simulations that resolve the ISM, and compare the resulting S850 to the H13, L21, and C23 predictions for the same galaxies. If the RT fluxes are consistent with L21/C23 rather than H13, the central claims must be revised downward by the corresponding normalization factor. As a quicker complementary check, recompute Fig. 5 and the TolTEC 1.1mm forecasts for FLAMINGO using L21 and C23, and using H13 with DTM=0.2 and 0.6; if the observed-count agreement and the 80,000-source forecast shift by more than a factor of 2 under these alternatives, the results are not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claims—FLAMINGO reproduces observed SMG counts without a top-heavy IMF, SMGs contribute up to 27% of the SFRD at z=2.6, and TolTEC UDS will detect ~80k sources—are all computed with the H13 parametric relation (Eq. 1, Table 2). The paper's own validation against EAGLE RT fluxes (Fig. 1a) shows H13 overpredicts the number density of S850>1 mJy galaxies by a factor of ~5 at z≈2, whereas L21 and C23 agree with RT. H13 is adopted instead because it matches the observed fluxes of 892 SMGs (Fig. 2), but that test uses SFR and Mdust from MAGPHYS SED fitting, which share model assumptions with the H13 dust templates and are not an independent calibration. No physical argument explains why H13 should be trusted for low-resolution FLAMINGO galaxies when it fails the EAGLE RT test. Because all downstream predictions scale with the flux normalization, the agreement in Fig. 5 and the survey forecasts may be a compensation between H13's higher normalization and FLAMINGO's resolution-limited SFR/dust masses. If L21 or C23 is the correct relation, the predictions change by factors of a few (e.g., an order-of-magnitude deficit in 850um counts at 1 mJy), and the headline conclusions would not hold.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper models 850 µm fluxes of simulated galaxies by applying three published parametric S850(SFR, Mdust) relations (H13, L21, C23; Eq. 1 and Table 2) to EAGLE, IllustrisTNG, and FLAMINGO. After comparing against EAGLE radiative-transfer fluxes and against observed SFRs/dust masses of 892 SMGs, the authors adopt H13 as the preferred relation while also showing L21 throughout. With the FLAMINGO L1_m8 box they derive redshift distributions, cumulative source counts, the SMG contribution to the cosmic star formation rate density, flux density functions, and TolTEC UDS/LSS forecasts. The headline results are that FLAMINGO with a Chabrier IMF reproduces the observed SMG counts and redshift distribution; SMGs with S850>1 mJy contribute up to about 27% of the cosmic SFRD at z≈2.6; and the TolTEC UDS will detect about 80,000 sources at 1.1 mm, tracing about 50% of the SFRD at z≈2.5. Appendices test box size, subgrid prescriptions, and cosmology.","tokens_in":30599,"tokens_out":9297,"duration_ms":87128,"significance":"The large statistical volume of FLAMINGO is a genuine strength: it avoids the small-volume fluctuations that affect EAGLE and TNG100 at the bright end, and the appendix tests of box size, feedback variations, and cosmology provide a useful map of systematic sensitivity. The paper is also honest in showing both H13 and L21 results. If the H13 calibration is accepted, the conclusions are important: they suggest that no top-heavy IMF is required, give a quantitative SMG contribution to the cosmic SFRD, and provide concrete TolTEC survey forecasts. The use of public simulation data and explicit sample definitions additionally aids reproducibility. However, the central claim rests entirely on the choice of H13, and the manuscript's own EAGLE radiative-transfer test contradicts that choice; until this contradiction is resolved, the quantitative conclusions (27%, 80,000 sources, 50% SFRD) are not robust.","major_comments":[{"comment":"The H13 relation, which underlies the paper's central claims, overproduces the comoving number density of S850>1 mJy galaxies in EAGLE by a factor of about 5 at z≈2 relative to the radiative-transfer calculation, while L21 and C23 are consistent with the RT results. The motivation for adopting H13 in Section 3.2.2 is its agreement with observed fluxes derived from MAGPHYS SFRs and dust masses, but no physical argument is given for why H13 should be preferred for FLAMINGO's lower-resolution galaxies when it fails the direct RT test. Because all of the headline quantities (source counts, 27% SFRD contribution, TolTEC source numbers) are computed with H13, the paper needs to resolve this contradiction, for example by showing that resolution or differences between EAGLE and FLAMINGO galaxy properties can account for the factor of 5, or by demonstrating that the main conclusions are preserved with L21/C23 applied to the same FLAMINGO sample.","section":"Section 3.2.1 / Fig. 1(a)"},{"comment":"The model selection and the subsequent validation are not fully independent. The 892-galaxy sample used to prefer H13 (da Cunha et al. 2015; Dudzeviciute et al. 2020; Hyun et al. 2023) is drawn from the same submillimeter-selected population that enters the redshift-distribution and number-count comparisons in Figs. 3 and 5; in particular, Dudzeviciute et al. (2020) contributes both to Fig. 2 and to the observed redshift distribution and counts. The manuscript should state explicitly which observational points are independent of the selection sample and quantify how the agreement in Figs. 3 and 5 depends on the Section 3.2.2 choice. Without that, the agreement is partly a restatement of the selection criterion.","section":"Section 3.2.2 vs Sections 4.1-4.2"},{"comment":"The default choices DTM=0.4 and β=1.8 are treated as fixed, but the headline forecasts depend on them. The DTM uncertainty is not negligible: the EAGLE RT comparison is run with DTM=0.3 while the FLAMINGO population synthesis uses DTM=0.4, which through the c exponent in Eq. (1) changes S850 by about 17%; and the S850 to 1.1/1.4/2.0 mm conversions in Eq. (6) assume a single modified-blackbody spectrum, so dust-temperature variations are ignored. The TolTEC number counts (80,000 at 1.1 mm) and SFRD fractions should be quoted with a systematic band obtained by varying DTM, β, and the observed flux conversion, or the authors should argue that these variations do not affect the ranking of models.","section":"Sections 3.1 and 5, Eqs. (1) and (6)"}],"minor_comments":[{"comment":"The text describes the total SFRD curve in the bottom-left panel as yellow dashed, while the figure caption calls it green dotted; please reconcile the color and line-style description.","section":"Fig. 10 / Section 5"},{"comment":"The sentence 'following equation (6) in Section 4.2' refers to a relation defined later in the paper; reorder the presentation or give the conversion factor at first use.","section":"Section 3.2.2"},{"comment":"The phrase 'observational estimates form Dwek (1998)' contains a typo ('form' should be 'from').","section":"Section 3.1"},{"comment":"Because the simulated redshift distributions are rescaled to the height of the Dudzeviciute et al. (2020) histogram, the reader should be reminded that Fig. 3 tests only the shape; the absolute normalization is tested in Fig. 5. Adding the scaling factors to the caption would help avoid misinterpretation.","section":"Fig. 3 / Section 4.1"},{"comment":"The identifier 'L1_M9' is used in Table B.1, while the main text refers to 'L1_m9' for the same simulation; unify the naming convention.","section":"Table B.1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope for A&A and the large FLAMINGO volume gives it a useful niche. The main risk is not that H13 is outside current consensus, but that the internal RT test in Fig. 1 directly contradicts the adopted model while all headline numbers are derived from that model. I see this as fixable with a careful revision: the authors should quantify whether resolution and galaxy property differences can explain the RT discrepancy, or should demote the H13-based numbers to conditional forecasts with systematic error bars. The overlap between the model-selection sample and the later validation datasets should also be made explicit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new content here is the side-by-side test of three published RT-calibrated flux relations (H13, L21, C23) against both EAGLE radiative transfer and a compiled sample of 892 observed SMGs, plus the first SMG population synthesis in FLAMINGO and the TolTEC forecasts. The paper is transparent and the tests are clearly presented. The community will use these forecasts.\n\nThe main weakness is the model-selection circularity. H13 is preferred because it matches observed fluxes from the same surveys (da Cunha, Dudzeviciute, Hyun) that later appear as validation targets for number counts. Meanwhile, against EAGLE RT, H13 overpredicts by roughly a factor of five at z~2, while L21 and C23 agree with RT. The paper doesn't resolve this tension; it simply chooses observations over RT. That is a legitimate choice, but it means the 27% SFRD contribution and the 80,000 TolTEC sources are not robust predictions—they scale with the H13 normalization. If L21 is closer to correct, those numbers drop by factors of a few.\n\nThe redshift distribution comparison is shape-only after normalization, so it is a weaker test than it appears. The TolTEC band conversion also relies on a single power-law SED with beta=1.8, another layer of assumption.\n\nI still think the paper is worth publishing. The side-by-side comparison is new and useful; the FLAMINGO galaxy population does have the right SFR/dust combination to be SMGs when H13 is used; the forecasts are actionable for TolTEC. But the authors should clearly state that the model is calibrated to the same observations, so the agreement is consistency, not prediction, and they should discuss the unresolved RT conflict more explicitly.\n\nI would send it to a serious referee. It is not a desk reject. I would probably not cite the 27% number without caveats, but I would cite the forecasts and the model comparison.","headline":"Useful population-synthesis forecasts for SMGs from FLAMINGO, but the headline numbers hinge on a flux relation chosen against the simulation's own RT benchmark, so treat the quantitative claims as conditional.","tokens_in":31211,"tokens_out":3447,"would_cite":true,"duration_ms":31621,"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":"This paper claims that FLAMINGO, post-processed with a power-law flux recipe, reproduces observed submillimeter galaxy counts and redshift distributions, and forecasts that the TolTEC ultra-deep survey will detect about 80,000 sources at…","keywords":["submillimeter galaxies","cosmological simulations","FLAMINGO","star formation rate density","dust-to-metal ratio","TolTEC","source number counts","redshift distribution"],"falsifier":"Compare the predicted 1.1-mm cumulative number counts from FLAMINGO+H13 against the first TolTEC Ultra Deep Survey catalog across $S_{1.1\\,\\mathrm{mm}} \\approx 0.1$ to 10 mJy; a deviation larger than the 1$\\sigma$ realization scatter shown in the paper's Fig. 10 would rule out the calibration. A more direct test measures star formation rates and dust masses of a flux-limited SMG sample and checks whether the observed pairs satisfy the H13 relation with the adopted dust-to-metal ratio.","tokens_in":30091,"feed_emoji":"🌌","tokens_out":12616,"duration_ms":97454,"temperature":0.7,"pith_summary":"Bright submillimeter galaxies are the most intensely star-forming objects at high redshift, but cosmological models have struggled to reproduce their observed number counts and redshift distributions. This paper shows that the large-volume FLAMINGO simulation, when each galaxy's 850-micron flux is assigned from a simple power-law in star formation rate and dust mass, reproduces the observed abundance and redshift distribution of submillimeter galaxies without invoking a top-heavy stellar initial mass function. On that basis the paper estimates that galaxies brighter than 1 mJy at 850 microns contributed up to about 27% of the cosmic star formation rate density at $z\\approx 2.6$, and forecasts that the TolTEC Ultra Deep Survey at 1.1 mm will detect roughly 80,000 sources in 0.8 square degrees, capturing about half of the star formation at $z\\approx 2.5$. A reader should care because this turns the FLAMINGO volume into a quantitative bridge between the unresolved dusty star-forming population and upcoming wide-area submillimeter surveys.","feed_headline":"FLAMINGO matches submillimeter galaxies; TolTEC to find 80,000","feed_subtitle":"If right, the 0.8 square-degree TolTEC UDS will capture about half the cosmic star formation at z≈2.5.","key_machinery":"The central machinery is the power-law flux recipe of Hayward et al. (2013): $S_{850} = a\\,(\\mathrm{SFR}/100\\,M_\\odot\\,\\mathrm{yr}^{-1})^b\\,(M_{\\mathrm{dust}}/10^8\\,M_\\odot)^c$ with $(a,b,c)=(0.81,0.43,0.54)$. It converts each simulated galaxy's instantaneous star formation rate and dust mass, the latter taken from cold star-forming gas metal mass times a constant dust-to-metal ratio of 0.4, into an 850-micron flux density. This makes population synthesis possible over the 1 Gpc/h FLAMINGO volume where full dust radiative transfer is computationally unfeasible. The paper benchmarks the H13, L21, and C23 recipes against EAGLE radiative-transfer fluxes and against 892 observed SMGs, then selects H13 for its forecasts.","core_discovery":"The central claim is that FLAMINGO simultaneously reproduces the observed cumulative number counts and redshift distribution of submillimeter galaxies (SMGs) when the 850-micron flux density is computed with the parametric relation of Hayward et al. (2013), $S_{850} = a\\,(\\mathrm{SFR}/100\\,M_\\odot\\,\\mathrm{yr}^{-1})^b\\,(M_{\\mathrm{dust}}/10^8\\,M_\\odot)^c$ with $(a,b,c)=(0.81,0.43,0.54)$, and dust mass derived from cold-gas metal mass at a fixed dust-to-metal ratio of 0.4. The same post-processing applied to EAGLE and IllustrisTNG underproduces bright SMGs, which the paper attributes to the smaller volumes of those simulations. The discovery is that a sufficiently large cosmological volume combined with this flux calibration removes the historical need for a top-heavy initial mass function in modeling the bright submillimeter population, and it identifies galaxies with $S_{850} > 1$ mJy as carrying up to 27% of the cosmic star formation rate density at $z=2.6$.","pith_inferences":["If the H13 calibration is correct, the TolTEC number counts at 1.1 mm provide a direct independent test of the adopted dust-to-metal ratio and flux recipe; the measured counts will either confirm or rule out the assumption behind the 80,000-source forecast.","The validation hints at an internal tension: the L21 recipe matches EAGLE's radiative-transfer fluxes while the H13 recipe matches the observed fluxes of SMGs at given SFR and dust mass, suggesting the discrepancy lies in the simulations' dust properties or sample selection rather than in the recipe choice alone.","The forecast that about half of the cosmic SFRD at $z\\approx2.5$ appears in the TolTEC 1.1-mm sample implies that the submillimeter-selected population is the dominant obscured channel at that epoch, and that combining TolTEC with the ODIN Ly$\\alpha$ survey could trace the same large-scale structures in both obscured and unobscured tracers.","A testable extension would apply the same recipe to higher-resolution volumes, such as TNG50, to separate the volume effect from a resolution effect in the bright-end deficit of EAGLE and IllustrisTNG."],"forward_implications":["The FLAMINGO simulation reproduces the observed SMG redshift distribution and source number counts above $S_{850} > 1$ mJy without a top-heavy initial mass function, removing a long-standing tension in galaxy formation modeling.","Bright SMGs ($S_{850} > 1$ mJy) carry up to 27% of the cosmic star formation rate density at $z \\approx 2.6$, making them a major and quantifiable channel of obscured star formation.","The TolTEC Ultra Deep Survey will detect roughly 80,000 sources over 0.8 deg$^2$ at 1.1 mm at the 4$\\sigma$ limit, and these sources trace about 50% of the cosmic star formation rate density at $z \\approx 2.5$.","EAGLE and IllustrisTNG underproduce bright SMGs because their volumes are too small to sample the bright end, demonstrating that large-volume simulations are required for SMG statistics.","The flux density function grows from $z=6$ to $z=2.5$ and then declines sharply at the bright end, with Schechter-function parameters tracking this evolution."],"supporting_citations":[{"why":"Supplies the power-law flux-density recipe (H13) that the paper selects for its SMG population synthesis.","marker":"Hayward et al. (2013)"},{"why":"Supplies the competing L21 parametric relation derived from SIMBA; the paper shows L21 under-predicts observed fluxes.","marker":"Lovell et al. (2021)"},{"why":"Provides EAGLE radiative-transfer SCUBA-2 850-micron fluxes used to benchmark the parametric recipes.","marker":"Camps et al. (2018)"},{"why":"Provides the ALMA 870-micron observed redshift distribution and number counts that FLAMINGO is compared against.","marker":"Dudzevičiūtė et al. (2020)"},{"why":"Provides an independent ALMA 870-micron sample used for the redshift-distribution comparison.","marker":"Simpson et al. (2020)"},{"why":"Provides the observational estimate of the SMG contribution to the cosmic SFRD that the paper's 27% result is compared to.","marker":"Swinbank et al. (2014)"},{"why":"Defines the TolTEC camera and the UDS/LSS survey depths used for the forecasts.","marker":"Wilson et al. (2020)"},{"why":"Describes the FLAMINGO simulations whose large volume is the basis for statistically robust SMG modeling.","marker":"Schaye et al. (2023)"}],"fun_headline_variants":["FLAMINGO matches SMG counts; TolTEC forecast: 80k sources","FLAMINGO reproduces submillimeter galaxies without top-heavy IMF","TolTEC UDS to detect 80k SMGs, 50% of cosmic SFR at z=2.5","FLAMINGO fits SMG counts; EAGLE and TNG fall short"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the assumption that a single power-law relation between star formation rate, dust mass, and 850-micron flux, with a constant dust-to-metal ratio of 0.4, holds for every galaxy in FLAMINGO; if that relation drifts with redshift or galaxy type, the counts, the SFRD contribution, and the survey forecasts all shift.","fun_headline_variants_meta":{"raw":{"variants":["FLAMINGO matches SMG counts; TolTEC forecast: 80k sources","FLAMINGO reproduces submillimeter galaxies without top-heavy IMF","TolTEC UDS to detect 80k SMGs, 50% of cosmic SFR at z=2.5","FLAMINGO fits SMG counts; EAGLE and TNG fall short"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000689,"raw_usage":{"total_tokens":3250,"prompt_tokens":1202,"completion_tokens":2048,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":818,"completion_tokens_details":{"reasoning_tokens":1949}},"tokens_in":818,"tokens_out":2048,"duration_ms":13788,"temperature":1.0,"reasoning_tokens":1949,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T20:33:48.005173+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the predicted 1.1-mm cumulative number counts from FLAMINGO+H13 against the first TolTEC Ultra Deep Survey catalog across $S_{1.1\\,\\mathrm{mm}} \\approx 0.1$ to 10 mJy; a deviation larger than the 1$\\sigma$ realization scatter shown in the paper's Fig. 10 would rule out the calibration. A more direct test measures star formation rates and dust masses of a flux-limited SMG sample and checks whether the observed pairs satisfy the H13 relation with the adopted dust-to-metal ratio.","supporting_citations":[{"cited_title":"M., Smail , I., Dudzevi c i \\= u t \\","cited_arxiv_id":null,"evidence_quote":"Provides an independent ALMA 870-micron sample used for the redshift-distribution comparison."},{"cited_title":"M., Simpson , J","cited_arxiv_id":null,"evidence_quote":"Provides the observational estimate of the SMG contribution to the cosmic SFRD that the paper's 27% result is compared to."},{"cited_title":"W., Abi-Saad , S., Ade , P., et al","cited_arxiv_id":null,"evidence_quote":"Defines the TolTEC camera and the UDS/LSS survey depths used for the forecasts."}],"review_version":1}