{"id":"8e4cfbad-233f-4e48-99a6-91085f4013ae","arxiv_id":"1908.01438","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Cosmological simulations predict that high-redshift galaxies transition between UV-bright and dusty IR-bright phases on roughly 100 Myr timescales, with ionizing escape fractions and metal-line luminosities fluctuating accordingly.","lead":"Using two cosmological zoom-in simulations with radiative transfer, this study predicts that the first galaxies rapidly switch between ultraviolet-bright and infrared/sub-millimeter-bright phases as supernova feedback ejects and then re-accretes dusty gas. The predictions connect observable high-redshift galaxy SEDs, [O III] and [C II] line luminosities, and ionizing escape fractions to the starburst-outflow cycle.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"UV/IR phase-cycling claim rests on two extreme zoom-in halos with no sample-variance test; population-level statements inherit selection bias.","rationale":"The reader's conditional verdict is appropriate; my stress test does not move it. The strongest claim requires population-level generality, but the evidence is two halos chosen as extreme (most massive) objects, so the inference to 'first galaxies' is under-supported. The paper itself points to 'more details' in Arata et al. (2019), and the observed comparisons are qualitative; nothing here is internally contradictory in a way that would force rejection. I therefore recommend the verdict stay conditional pending a sample-variance/convergence check.","tokens_in":5258,"tokens_out":7213,"duration_ms":74088,"concrete_test":"Run 10–20 additional zoom-in halos spanning M_h ~ 10^10–10^12 h^-1 M_sun at z=6 drawn from the same parent boxes, and compute the waiting-time distribution between UV- and IR-bright phases from the autocorrelation of log10(L_UV/L_IR) in the RT post-processing. If the ~100 Myr period is not a robust peak across the sample, the rapid-transition claim fails as a population statement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that high-z galaxies cycle between UV-bright and IR-bright phases on ~100 Myr with fesc varying from 1% to 40%—holds only if these behaviors are typical of the z=6–15 galaxy population. The paper's entire support is two zoom-in halos, Halo-11 and Halo-12, selected as the most massive halo at z=6 in 20 and 100 h^-1 cMpc boxes (Table 1). This selection targets rare massive systems; at z>10 these are high-sigma peaks, not the typical first galaxies probed by current surveys. No sample-variance test, no convergence study (the two runs differ by roughly an order of magnitude in particle mass), and no independent subgrid implementation are shown. The 100 Myr timescale in §3 is read off Figure 2 and equated to the halo free-fall time; with n=2 it cannot be distinguished from stochastic burst intervals produced by the adopted SN-feedback prescription. The population-level statements (satellite observability >50%, sub-mm source density 10^-2 cMpc^-3, line-ratio trend) are all derived from the same two halos and inherit this selection bias. A secondary inconsistency: the abstract quotes fesc=1–40% at z>10 for ionizing photons, whereas §3 reports fesc=0.2–0.8 at z<10 for UV photons; the paper does not reconcile these, which complicates verification of the central quantitative claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses cosmological SPH zoom-in simulations of two massive halos (Halo-11 and Halo-12) together with multi-wavelength radiative transfer (Art2) to predict UV/IR SEDs, escape fractions, sub-mm fluxes, and [O III]/[C II] line luminosities for z=6-15 galaxies. The central claim is that supernova feedback and gas accretion drive intermittent star formation that cycles galaxies between UV-bright and IR-bright phases on a ~100 Myr timescale, with the escape fraction of ionizing photons varying between 1 and 40% at z>10. The paper presents maps and light curves showing spatial offsets between UV and FIR peaks, compares predicted line luminosities to ALMA observations, and makes forecast-type statements including a >50% sub-mm observability for massive satellites, a sub-mm source number density of 10^-2 cMpc^-3, and extended [C II] emission over ~20 kpc.","tokens_in":5581,"tokens_out":6844,"duration_ms":67076,"significance":"If the phase-cycling picture is correct, it would substantially change the interpretation of high-z galaxies: rest-UV-selected and sub-mm-selected samples would trace different evolutionary phases of the same population, and the escape fraction would be a strongly time-dependent quantity for reionization modeling. The paper's strengths are that it uses established simulation machinery (Gadget-3 with OWLS/FiBY subgrid models and the Art2 radiative transfer code) and that its predictions are falsifiable in form: the 100 Myr SED shift, the 1-40% ionizing escape fraction, the sub-mm source density, and the decreasing L[O III]/L[C II] ratio with bolometric luminosity can all be checked against future JWST/ALMA data. However, the evidence base is narrow: all population-level results rest on only two zoom-in halos, model details are deferred to companion papers, and the comparison with observations is qualitative. The paper would be much more convincing with a resolution-convergence test, a sample-variance estimate, and a reconciliation of the abstract's ionizing escape-fraction range with the UV escape-fraction range stated in Section 3.","major_comments":[{"comment":"The population-level statements in §3 (satellite observability exceeding 50%, sub-mm source density 10^-2 cMpc^-3, the ~100 Myr UV/IR cycling timescale, and the escape-fraction range) are all derived from only two zoom-in halos, Halo-11 and Halo-12, which are selected as the most massive halos in their respective boxes and whose dark-matter particle masses differ by a factor of about 17 (m_DM = 6.6e4 versus 1.1e6 h^-1 M_sun). No convergence test and no sample-variance estimate are presented, so the reader cannot assess whether these halos are representative of the z=6-15 galaxy population or are high-sigma outliers. The authors should add a resolution study and at least a small ensemble of independent halos, or explicitly reframe the claims as case studies rather than population predictions.","section":"Table 1; §3"},{"comment":"The abstract's central quantitative claim is that the escape fraction of ionizing photons changes between 1 and 40% at z>10, while §3 reports f_esc = 0.2-0.8 at z<10 for ultraviolet photons. The paper does not state whether these are the same quantity evaluated at different redshifts or whether they refer to different definitions, and this ambiguity prevents verification of the headline number. A plot of the ionizing-photon escape fraction versus redshift for both halos, with starburst and outflow phases marked, should be added, and the abstract should be reconciled with the body of the paper.","section":"Abstract vs §3"},{"comment":"The essential modeling choices that control the SED and line predictions are not specified in this manuscript: the subgrid star formation and feedback parameters are referred to Yajima et al. (2017), the dust and RT details to Arata et al. (2019), and the [O III]/[C II] line calculation to Arata et al. (in prep.). In particular, the dust mass is set proportional to gas metallicity with an assumed dust temperature, and the metal-line calculation assumes photoionization equilibrium under the stellar radiation field; these choices directly set the IR/sub-mm fluxes and line luminosities. The authors should either summarize the key parameters (dust-to-metal ratio, assumed dust temperature, treatment of the radiation field) in the text or provide a stable reference to a published methods paper, since the comparison with observations in §4 cannot otherwise be evaluated.","section":"§2"},{"comment":"The 100 Myr timescale for the UV/IR phase transition is read off Figure 2 and equated to the halo free-fall time, but the SFR and escape-fraction light curves are visibly bursty and only two realizations are shown. With n=2, the apparent period could be a stochastic property of the adopted SN-feedback prescription rather than a robust duty cycle. The period should be quantified (for example, by computing the autocorrelation function of the SFR or f_esc time series) and checked for convergence with numerical resolution, or the claim should be softened to a description of the two simulated cases.","section":"§3, Figure 2"}],"minor_comments":[{"comment":"There are typographical errors in names and references: 'Riechars et al. 2013' should be 'Riechers et al. 2013', 'Marron et al. 2018' should be 'Marrone et al. 2018', and 'Decalri' in the reference list should be 'Decarli'.","section":"§1"},{"comment":"The author name 'Yuexing Li' appears as 'Y uexing Li' in the article header; this should be corrected.","section":"Author list"},{"comment":"The observational points in Figure 3 are shown as open symbols without error bars, and the text notes that they 'may shift to higher SFR because actual dust temperature might be higher than the assumed one.' A quantitative estimate of this shift would make the claimed agreement more informative.","section":"Figure 3"},{"comment":"The sentence 'We focus on how the RT results depend on the dust distribution' is not followed by a test of different dust distributions; the paper presents only one dust model. The authors should either add such a comparison or revise the sentence to describe what is actually shown.","section":"§2"}],"recommendation":"major_revision","confidential_remarks":"This is a proceedings-style contribution whose central claims are more ambitious than the two-halo evidence base. I believe the paper is fixable within its scope, but the authors should be asked for the concrete additions identified in the report: a resolution/sample-variance statement, a f_esc versus redshift figure for ionizing photons, and a self-contained description of the dust and line-emission assumptions. The heavy reliance on companion papers is understandable for a proceedings, but the current text is difficult to referee as a stand-alone claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a short proceedings paper that condenses the authors' earlier work (Yajima+17, Arata+19) into a set of predictions for ALMA and JWST. The genuinely new piece is the [O III]/[C II] versus bolometric luminosity trend, driven by metal enrichment suppressing [O III] while boosting Lbol via cooling; that is physical and testable. The paper also does a decent job of honestly comparing to observations without claiming precision. The radiative transfer is standard but careful, and the qualitative comparisons are not circular: no model parameters are fitted to the [O III] or [C II] data.\n\nThat said, the load-bearing claim—that galaxies cycle between UV-bright and IR-bright phases on a 100 Myr timescale with fesc varying from a few to tens of percent—rests entirely on two zoom-in halos, Halo-11 and Halo-12, which are the most massive halos in their respective boxes. Those are rare, high-sigma peaks at z>10, not typical first galaxies. No sample-variance test, no convergence test (the two runs differ by ~an order of magnitude in particle mass), and no released data are provided. The 100 Myr timescale is read off a single figure and equated to the free-fall time; with n=2 you can't distinguish that from stochastic burst intervals produced by the adopted feedback prescription. The population-level statements—satellite observability >50%, sub-mm source density 10^-2 cMpc^-3—inherit this selection bias and should be taken as illustrative.\n\nThere is also a minor inconsistency that will trip up readers: the abstract quotes fesc = 1–40% at z>10, while §3 reports fesc = 0.2–0.8 at z<10 for UV photons. These refer to different photon energies (ionizing vs. UV continuum) and may be perfectly consistent, but the paper doesn't say so, which complicates verification.\n\nOverall, this is a reasonable summary for a proceedings. It does not oversell its evidence within the text, though the abstract generalizes a bit beyond the two-halo sample. For a main-journal submission, I'd want convergence and sample-variance tests plus a self-contained description of the metal-line model (currently deferred to Arata in prep.). For this venue, it is acceptable as a condensed progress report.\n\nRecommendation: Engage with it, and point any reader to the underlying Arata 2019 paper for the actual support. It deserves a serious referee only in the sense that a proceedings editor should check whether the claims are appropriately qualified; they mostly are.","headline":"A competent proceedings summary of the authors' own simulation program, with one genuinely new line-ratio trend but the central phase-cycling picture resting on two rare halos and no sample-variance analysis.","tokens_in":6121,"tokens_out":1975,"would_cite":false,"duration_ms":21596,"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 argues that the first galaxies cycle between UV-bright and dusty infrared-bright phases on roughly 100-million-year timescales, so each ALMA or JWST snapshot catches a different stage of the same starburst–outflow cycle.","keywords":["high-redshift galaxies","reionization","radiative transfer","cosmological simulations","dust emission","escape fraction","intermittent star formation","submillimeter galaxies"],"falsifier":"Measure rest-UV and sub-mm fluxes for a sample of $z>7$ galaxies with JWST and ALMA: the model predicts an anti-correlation between UV brightness and sub-mm brightness across the population, because galaxies cycle between dust-cleared UV-bright phases and dusty sub-mm-bright phases on ~100 Myr timescales; a population that is largely bright in both bands simultaneously would falsify the phase-cycling picture.","tokens_in":5092,"feed_emoji":"🔭","tokens_out":10143,"duration_ms":94152,"temperature":0.7,"pith_summary":"This paper tries to establish that the multi-wavelength appearance of the first galaxies is not set by a fixed dust content but by a rapid cycle of star formation and supernova feedback. Using two zoom-in cosmological simulations at $z=6$–$15$ with radiative transfer, it shows the peak of a galaxy's spectral energy distribution can shift between UV and infrared wavelengths on a $\\sim 100\\,{\\rm Myr}$ timescale, and that the escape fraction of ionizing photons swings between roughly 1 and 40 percent at $z>10$. If true, high-redshift galaxies observed by ALMA and JWST are snapshots of a cycling population: some caught in a UV-bright, outflowing phase, others in a dusty, sub-millimeter-bright starburst. The predicted behavior of [O III] 88 $\\mu$m and [C II] 158 $\\mu$m lines gives observers a way to read which phase a galaxy is in.","feed_headline":"First galaxies flip between UV-bright and dusty phases in 100 Myr","feed_subtitle":"Simulations show supernova feedback cycles move the SED peak from UV to infrared, shaping ALMA and JWST views.","key_machinery":"The central mechanism is the star formation–supernova feedback cycle acting on dust: a starburst is quenched when supernovae eject gas and dust, clearing escape routes for UV and ionizing photons, after which gas re-accretes on roughly a free-fall timescale and the next burst is shrouded in dusty gas. The calculations combine the Gadget-3 SPH code with OWLS and FiBY sub-grid models, and the Art2 radiative-transfer code, which propagates photon packets on adaptive refinement grids that follow the simulated gas down to physical scales of a few parsecs. Dust mass in each cell is set proportional to gas metallicity, and the hydrogen ionization structure from the radiative transfer drives [O III] 88 $\\mu$m and [C II] 158 $\\mu$m luminosities through ionization equilibrium and level-population rate equations.","core_discovery":"The central claim is that intermittent star formation, regulated by supernova feedback, controls the radiative properties of the first galaxies. In a starbursting phase, dusty gas covers the star-forming regions, absorbing UV photons and re-emitting them in the infrared, so the galaxy appears bright in observed-frame sub-millimeter wavelengths; when supernova feedback ejects the gas and dust, UV photons escape and the galaxy turns UV-bright. The simulations find the SED peak shifts between these two states on a timescale of about 100 Myr, comparable to the halo free-fall time, and that the escape fraction of ionizing photons varies between 1 and 40 percent at $z>10$. Metal-line luminosities follow the same cycle: [O III] 88 $\\mu$m is bright only in starbursting phases, reaching $10^{42}$–$10^{43}\\,{\\rm erg\\,s^{-1}}$ for halos of $\\sim 10^{11}$–$10^{12}\\,{\\rm M}_\\odot$, while [C II] 158 $\\mu$m persists in neutral gas, and the ratio $L_{\\rm [O\\,III]}/L_{\\rm [C\\,II]}$ drops by about an order of magnitude as metal enrichment proceeds. The mass fraction of H II regions changes with star formation history, so metal-line and Lyman-$\\alpha$ luminosities fluctuate as well.","pith_inferences":["The phase-cycling picture implies that much of the observed scatter in UV-to-IR flux ratios at fixed stellar mass among high-z galaxies could be temporal phase diversity rather than galaxy-to-galaxy variation in dust content; this is an extension the paper does not state.","If escape fractions swing between 1 and 40 percent on 100 Myr timescales, global reionization models that assume a constant escape fraction per galaxy may need to use a time-averaged value, which could change the inferred ionizing photon budget; this is an editorial inference.","The predicted ~1.4 arcsec spatial offset between UV and FIR brightness peaks within one simulated clumpy galaxy suggests that high-resolution ALMA/JWST imaging of individual $z\\sim6$ sources could catch the two phases spatially separated; the paper presents the offset as an image property but does not develop it as a test.","The metal-enrichment explanation of the $L_{\\rm [O\\,III]}/L_{\\rm [C\\,II]}$ decline could be turned into a rough metallicity indicator for $z>7$ galaxies, since the simulated ratio drops about an order of magnitude from sub-solar to solar metallicity; the paper stops short of proposing this application."],"forward_implications":["If the SED does cycle on $\\sim 100\\,{\\rm Myr}$, then a galaxy observed at two epochs separated by about 100 Myr should flip between UV-bright and sub-mm-bright appearances, and surveys should see an anti-correlation between UV and sub-mm brightness in a coeval sample.","[O III] 88 $\\mu$m acts as a starburst-phase indicator, while [C II] 158 $\\mu$m traces neutral gas even during outflow phases, so the line ratio $L_{\\rm [O\\,III]}/L_{\\rm [C\\,II]}$ reveals both the evolutionary phase and the metal content of a $z>7$ galaxy.","Deep ALMA observations reaching $\\sim 10^{-4}\\,{\\rm mJy\\,arcsec^{-2}}$ should detect extended [C II] emission tracing neutral gas out to roughly 20 physical kpc around luminous high-z halos.","A future sub-mm survey with sensitivity $\\gtrsim 10^{-2}\\,{\\rm mJy}$ should find a number density of about $10^{-2}\\,{\\rm cMpc^{-3}}$ for these sources at $z\\sim6$–$7$.","Combined ALMA and JWST observations can separate starbursting from outflowing phases, making the multi-phase ISM structure of reionization-era galaxies observable."],"supporting_citations":[{"why":"Supplies the zoom-in simulations and sub-grid physics, and established the intermittent star formation that drives the claimed cycle.","marker":"Yajima et al. (2017)"},{"why":"Provides the Art2 radiative-transfer code used for the UV/IR SED, dust emission, and ionization structure.","marker":"Li et al. (2008)"},{"why":"Provides the OWLS sub-grid models incorporated into the SPH runs.","marker":"Schaye et al. (2010)"},{"why":"Provides the FiBY sub-grid models used alongside OWLS in the hydrodynamic simulations.","marker":"Johnson et al. (2013)"},{"why":"Demonstrated that supernova feedback and gas accretion make high-z star formation intermittent, the premise of the cycle.","marker":"Kimm & Cen (2014)"},{"why":"Offers the $z\\approx7.5$ dust continuum detection used to calibrate the predicted sub-mm fluxes and dust masses.","marker":"Watson et al. (2015)"},{"why":"Reported the observed $L_{\\rm [O\\,III]}/L_{\\rm [C\\,II]}$ versus bolometric luminosity anti-correlation that the paper explains by metal enrichment.","marker":"Hashimoto et al. (2019)"},{"why":"Supports the claim that deep [C II] observations can trace neutral gas extending roughly 20 kpc around high-z galaxies.","marker":"Fujimoto et al. (2019)"}],"fun_headline_variants":["Supernova feedback flips early galaxies between UV and dusty phases","First galaxies cycle UV and sub-mm brightness every 100 Myr","Simulations show 100-Myr switch between UV-bright and dusty states in first galaxies","Supernovae eject dust to unveil UV, then dusty gas re-covers stars: a 100-Myr cycle","ALMA and JWST will see first galaxies flicker as supernovae cycle dust"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The quantitative predictions rest on two zoom-in halos, the most massive halos in a 20 and a 100 comoving Mpc/h box at $z=6$, and the paper does not test how representative these two systems are of the full $z=6$–$15$ galaxy population.","fun_headline_variants_meta":{"raw":{"variants":["Supernova feedback flips early galaxies between UV and dusty phases","First galaxies cycle UV and sub-mm brightness every 100 Myr","Simulations show 100-Myr switch between UV-bright and dusty states in first galaxies","Supernovae eject dust to unveil UV, then dusty gas re-covers stars: a 100-Myr cycle","ALMA and JWST will see first galaxies flicker as supernovae cycle dust"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000734,"raw_usage":{"total_tokens":3421,"prompt_tokens":1220,"completion_tokens":2201,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":836,"completion_tokens_details":{"reasoning_tokens":2091}},"tokens_in":836,"tokens_out":2201,"duration_ms":16572,"temperature":1.0,"reasoning_tokens":2091,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:12:54.972051+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure rest-UV and sub-mm fluxes for a sample of $z>7$ galaxies with JWST and ALMA: the model predicts an anti-correlation between UV brightness and sub-mm brightness across the population, because galaxies cycle between dust-cleared UV-bright phases and dusty sub-mm-bright phases on ~100 Myr timescales; a population that is largely bright in both bands simultaneously would falsify the phase-cycling picture.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the zoom-in simulations and sub-grid physics, and established the intermittent star formation that drives the claimed cycle."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the OWLS sub-grid models incorporated into the SPH runs."},{"cited_title":"L., Dalla Vecchia C., & Khochfar S","cited_arxiv_id":null,"evidence_quote":"Provides the FiBY sub-grid models used alongside OWLS in the hydrodynamic simulations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrated that supernova feedback and gas accretion make high-z star formation intermittent, the premise of the cycle."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Offers the $z\\approx7.5$ dust continuum detection used to calibrate the predicted sub-mm fluxes and dust masses."},{"cited_title":"First Identification of 10-kpc Scale [CII] 158um Halos around Star-Forming Galaxies at z=5-7","cited_arxiv_id":"1902.06760","evidence_quote":"Supports the claim that deep [C II] observations can trace neutral gas extending roughly 20 kpc around high-z galaxies."}],"review_version":1}