{"id":"676b40d9-c651-4928-bc4c-f96c07f8203f","arxiv_id":"2506.20422","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Galaxies with double-peaked Lyman-alpha lines are younger and drive weaker outflows than single-peaked ones, linking feedback onset to the delayed appearance of core-collapse supernovae.","lead":"Using 338 gravitationally lensed galaxies seen by MUSE, this study ties the shape of Lyman-alpha emission lines to the age of the starburst and the strength of its outflows. It concludes that double-peaked Lyman-alpha profiles come from very young starbursts whose outflows have not yet strengthened because core-collapse supernovae take about 4 million years to begin.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"BEAGLE age separation between DP and SP rests on untested constant-SFH and 300 Msun IMF choices; a broader model grid is needed to confirm the 4 Myr/10 Myr SNe-onset interpretation.","rationale":"I agree with the reader's identification of the BEAGLE model assumptions as the weakest link. The empirical correlations are carefully built: systemic redshifts from optically-thin lines, stacked down-the-barrel absorption profiles with robustness tests, and clear EW differences between SP and DP sources all support a real physical distinction. The zELDA finding that homogeneous expanding shell models overestimate systemic redshifts is an honest and useful negative result. However, the central, novel claim is the temporal one: that DP sources are younger than ~4 Myr and that the feedback transition coincides with the onset of core-collapse SNe. That claim is only as strong as the BEAGLE age posteriors, which assume a constant SFH and a 300 Msun upper-mass cutoff. These choices are not just nuisance parameters; they directly set the age scale. The paper's own statement that a single-burst SFH fails to reproduce the data does not establish that constant SFH is correct, and the choice of IMF cutoff is data-driven rather than independently justified. Because the same diagnostics (He II, C IV, N V P-Cygni) are also the most sensitive to binaries, rotation, and IMF shape, the inference is fragile. I therefore recommend keeping the reader's CONDITIONAL verdict: the paper should be accepted only after the proposed model-grid test is run. If the age separation survives, the paper is a significant step forward; if not, the interpretation should be substantially softened.","tokens_in":32486,"tokens_out":8548,"duration_ms":107367,"concrete_test":"Refit the same stacked SP and DP spectral constraints (the emission-line EWs and the 1220-1260 Angstrom N V P-Cygni window) with BEAGLE or an equivalent code (BPASS/CIGALE) across a grid that includes (a) constant SFH, (b) exponentially declining SFH with e-folding times of 1, 3, 10, 30, and 100 Myr, (c) a two-component SFH with an old underlying population plus a young burst, and (d) IMF slopes from Salpeter to Chabrier with upper cutoffs of 100 and 300 Msun. If the posterior age difference between DP and SP does not robustly remain below ~5 Myr for DP and above ~10 Myr for SP across this grid, the delayed-SNe interpretation is not supported by the current data and the paper should be revised to present the age result as model-dependent.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that DP Ly-alpha sources are extremely young (<~4 Myr) and therefore show weaker feedback because core-collapse SNe have not yet begun, whereas SP sources are older (>~10 Myr) and show fully developed SNe-driven outflows. The quantitative age separation is obtained exclusively from BEAGLE fits to the stacked SP/DP spectra under a constant SFH and a Chabrier IMF with a 300 Msun upper-mass cutoff (Sec. 3.2.3, Table 2). The 300 Msun cutoff is itself adopted because it alone reproduces the DP He II 1640 EW; the alternative 100 Msun cutoff was rejected. This is not a peripheral detail: the He II, C IV, and N V P-Cygni diagnostics that drive the age estimate are precisely the features most sensitive to the massive-star IMF and to the assumed SFH. Under an IMF with a lower cutoff or a steeper slope, the same He II EW would require a considerably younger age (or a different SFH). Under an exponentially declining or two-burst SFH, the parameter that BEAGLE calls 'age since onset of star formation' no longer maps cleanly to the time when core-collapse SNe begin. The paper tests only single-burst versus constant SFH and only 100 versus 300 Msun cutoffs; it does not vary the IMF slope, use binary-rich stellar population models, or explore declining/multiple-burst SFHs. Thus the landmarks of 4 Myr and 10 Myr, and the claim that the DP-to-SP transition aligns with the onset of core-collapse SNe, are supported only within a specific, untested model grid. If a broader grid shifts the DP age above ~6 Myr or the SP age below ~8 Myr, the temporal alignment with SNe onset weakens and the 'delayed feedback' picture loses its primary quantitative support. The empirical stacking results (stronger, broader low-ionization absorption in SP; higher metal EWs in DP) are robust, but the interpretation attached to them is not.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes 338 gravitationally lensed Lyman-alpha emitters at z ~ 2.9-6.6 from the MUSE cluster fields of Richard et al. (2021). The authors fit Lyman-alpha profiles, measure systemic redshifts with optically thin metal emission lines, and stack metal absorption and emission lines for sources with single-peaked (SP) and double-peaked (DP) Lyman-alpha profiles. They report that SP sources show stronger, broader, more blueshifted low-ionization absorption than DP sources, while DP sources show higher emission-line equivalent widths and younger BEAGLE stellar ages (about 4 Myr versus about 10 Myr or more). They interpret this as delayed feedback: DP sources are extremely young starbursts in which core-collapse supernovae have not yet begun, whereas SP sources are older and host SNe-driven outflows. They also fit expanding-shell radiative transfer models with zELDA and find that, despite good chi-square values, the models yield unphysical parameters and systematically overestimate systemic redshifts, cautioning against the use of such models for outflow inference.","tokens_in":32859,"tokens_out":3909,"duration_ms":45512,"significance":"If the age-feedback interpretation holds, the paper provides a rare empirical constraint on the onset of supernova feedback in high-redshift starbursts, linking Lyman-alpha profile morphology to stellar population age. The empirical correlations themselves--stronger and wider low-ionization absorption in SP than DP stacks, and higher metal emission EWs in DP sources--are supported by careful stacking, bootstrap uncertainties, and several robustness tests. The paper also makes a valuable methodological point by showing that homogeneous expanding-shell models can fit Lyman-alpha profiles while returning unphysical parameters. These strengths make the central idea appealing and worth publishing if the model-dependent age inference can be hardened.","major_comments":[{"comment":"The quantitative age separation between DP and SP sources rests entirely on BEAGLE fits that assume a constant star formation history, a Chabrier IMF with a 300 Msun upper mass cutoff, and a gas density of 10^3 cm^-3. The paper tests only 100 versus 300 Msun cutoffs and single-burst versus constant SFH, adopting the 300 Msun cutoff because the DP He II 1640 EW requires it. Since the He II EW, C IV emission, and N V P-Cygni profile are the features most sensitive to the massive-star IMF and SFH, the inferred ages of about 4 Myr and about 10 Myr are conditional on these choices. I request a broader model grid--varying the IMF slope and upper cutoff, using declining or two-burst SFHs, and ideally including binary-rich stellar population models--to demonstrate that the age separation and its alignment with the onset of core-collapse supernovae persist.","section":"Section 3.2.3, Table 2 and Figure 13"},{"comment":"The DP/SP classification is acknowledged to be sensitive to SNR and IGM attenuation, yet the individual-source EW comparison in Figure 9 does not apply the redshift cut used in the stacked analysis (SP at z > 4 excluded). This creates a possible selection bias: intrinsically DP sources at high redshift whose blue peaks are attenuated by the IGM would be classified as SP, and the KS test p-value of 0.0025 could then reflect IGM/SNR selection rather than an intrinsic age difference. Please apply the same redshift and SNR cuts to the individual-EW comparison, or otherwise show that the emission-line EW difference is not driven by these selection effects.","section":"Sections 3.2.1 and 3.2.3, Figures 9 and 12"},{"comment":"The claim that SP outflows are 'slower' or 'less turbulent' than DP outflows is weaker than the abstract suggests. The stacked absorption centroid offset is -60 km/s with a 95% confidence interval of [-105, +6] km/s, which includes zero, and the width difference loses formal significance when the three strongest SP absorbers are excluded (as the paper itself notes). The absorption strength difference is robust, but the 'narrower' and especially 'slower' language should be softened or supported by additional analysis, because it is part of the central feedback-timescale interpretation.","section":"Section 3.2.2, Figure 10 and Abstract"}],"minor_comments":[{"comment":"The sentence 'we excluded any SP sources with z < 4, which may be intrinsically DP sources affected by IGM attenuation' appears to be a typo: the comparison with Section 3.2.2 indicates that the intended cut is z > 4.","section":"Section 3.2.3"},{"comment":"The joint KS test is reported with p = 0.0025, but the text does not specify whether the test treats each line detection as an independent entry or combines lines per source; this should be clarified because multiple lines from the same source are not independent.","section":"Section 3.2.1, Figure 9"},{"comment":"The line is referred to as Si II 1527, whereas the standard air wavelength is 1526.72 Å; please use the conventional notation.","section":"Section 3.1.3"},{"comment":"The toy models in Table 4 and Figure 17 are useful illustrations, but the text should more explicitly state that the two-component mixtures neglect radiative transfer between components and therefore provide only a qualitative demonstration, not a quantitative calibration, of the HTS model biases.","section":"Section 3.3"}],"recommendation":"major_revision","confidential_remarks":"The paper's central physical claim is attractive and timely, but the load-bearing age inference depends on a narrow BEAGLE model grid. The authors should be encouraged to run the requested model variations; if the age separation persists across a broader grid, the paper would be a strong contribution to the feedback-timescale literature. The zELDA cautionary result is a nice secondary contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The empirical part of this paper is the real contribution. Working from the R21 lensed MUSE sample, they measure systemic redshifts from optically-thin UV lines, stack down-the-barrel absorption separately for single- and double-peaked Ly-alpha emitters, and show that SP sources have roughly 2-3x stronger and wider low-ionization absorption while DP sources have higher metal emission EWs and a more pronounced N V P-Cygni feature. Those contrasts hold up under the robustness tests they run. The zELDA result is a useful caution: simple expanding-shell fits reproduce the profiles but systematically overestimate systemic redshifts and give unphysical parameters, especially when the gas is clumpy or multi-velocity as their absorption-line ratios suggest.\n\nThe soft spot is the BEAGLE age separation. The claim that DP sources are under ~4 Myr and SP sources over ~10 Myr comes from fits that assume a constant SFH, a Chabrier IMF with a 300 Msun upper cutoff, and a single gas density. The 300 Msun cutoff is itself chosen because it alone reproduces the DP He II 1640 EW - exactly the feature that is most sensitive to IMF and SFH. They test single-burst vs constant SFH and 100 vs 300 Msun cutoffs, but not IMF slope, binary populations, or declining/bursty SFHs. If those shift the DP age above ~6 Myr or the SP age below ~8 Myr, the neat alignment with core-collapse SNe onset weakens. They do acknowledge the mass degeneracy and show no F160W luminosity difference, which helps, but the model grid is still narrow.\n\nI would not call this circular - the ages are fitted, not predicted from the input, and the empirical stacking results stand on their own. The paper is also honest about its limitations, including the binary-evolution caveat and the alternative mass-driven scenario.\n\nWho this is for: observers working on high-z outflows and Ly-alpha RT modelers. The empirical correlations and the zELDA caution are worth citing now; the delayed-SNe interpretation is a good hypothesis to test with JWST masses and broader stellar population models. I would send it to a serious referee, with the request that the age analysis either be expanded to a broader grid or explicitly framed as conditional on the adopted model.","headline":"Solid empirical core, plausible but model-dependent age interpretation; worth refereeing with a request for a broader BEAGLE grid.","tokens_in":33479,"tokens_out":2788,"would_cite":true,"duration_ms":30717,"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 double-peaked Lyman-$\\alpha$ profiles are the signature of starbursts younger than about 4 million years, before core-collapse supernovae have turned on, and that feedback, outflow speed, and outflow mass strengthen…","keywords":["Lyman-alpha emitters","high-redshift galaxies","stellar feedback","galactic outflows","core-collapse supernovae","gravitational lensing","metal absorption lines","spectral fitting"],"falsifier":"Measure the stellar age of individual double-peaked Lyman-alpha galaxies from Balmer-line or continuum-break fitting, while simultaneously measuring the width and centroid of their Si II absorption. If a galaxy with an independently measured age below roughly 4 Myr shows broad (more than about 200 km/s) strongly blueshifted absorption, the claim that feedback remains weak before core-collapse supernovae would fail; if no such young strong-outflow galaxy exists, the delayed-feedback picture is supported.","tokens_in":32231,"feed_emoji":"🔭","tokens_out":10933,"duration_ms":106412,"temperature":0.7,"pith_summary":"The paper analyzes 338 gravitationally lensed Lyman-$\\alpha$-emitting galaxies at redshifts 2.9 to 6.6, combining the Lyman-$\\alpha$ profile with metal emission and absorption lines. It tries to establish that the shape of the Lyman-$\\alpha$ line encodes the age of the starburst and the strength of its stellar feedback: double-peaked profiles come from extremely young systems ($\\lesssim 4$ Myr) with weak, slow outflows, while single-peaked profiles come from older systems ($\\gtrsim 10$ Myr) with stronger, faster outflows. The proposed mechanism is delayed feedback, because core-collapse supernovae can only begin after the most massive stars evolve for a few million years. A sympathetic reader would care because, if true, Lyman-$\\alpha$ morphology becomes a direct probe of feedback timing in the early universe, and simple expanding-shell fits to Lyman-$\\alpha$ profiles would be shown to give misleading outflow parameters.","feed_headline":"Double-peaked Lyman-alpha marks starbursts younger than 4 Myr","feed_subtitle":"Metal absorption stacks show feedback strengthens only after core-collapse supernovae switch on","key_machinery":"The machinery is a two-sided comparison of single- versus double-peaked Lyman-$\\alpha$ profiles. On the outflow side, stacked spectra of low-ionization (Si II, C II) and high-ionization (Si IV) absorption lines measure the column density, covering fraction, and velocity width of gas along the line of sight, providing a direct, model-independent readout of outflow strength. On the source side, stacked UV emission line equivalent widths (C III], C IV, He II, O III], Si III]) and the N V $\\lambda\\lambda 1238,1243$ P-Cygni profile, a stellar-wind feature that decays within a few million years, are interpreted through a Bayesian spectral-fitting code built on stellar-population and photoionization models; the P-Cygni feature is the key age indicator because it fades on exactly the timescale of interest. The homogeneous thin-shell (HTS) expanding-shell radiative transfer model, fitted by Markov-chain Monte Carlo, is the object whose failure is diagnosed: it fits profiles well but cannot recover the outflow parameters seen in absorption.","core_discovery":"The central discovery is a systematic, empirically grounded link between Lyman-$\\alpha$ line morphology and the evolutionary state of stellar feedback in high-redshift starbursts. Using systemic redshifts from optically thin metal emission lines, the authors confirm that single Lyman-$\\alpha$ peaks are redshifted relative to the systemic velocity while double peaks straddle it. In stacked down-the-barrel (line-of-sight) absorption spectra, single-peaked sources show roughly three times stronger and wider low-ionization absorption than double-peaked sources, indicating higher neutral column, larger covering fraction, and faster or more turbulent outflowing gas; double-peaked sources also lack significant high-ionization absorption. Conversely, double-peaked sources show much stronger UV metal emission lines and a deeper N V P-Cygni wind feature, which, when fitted with stellar population and photoionization models, imply ages of $\\lesssim 4$ Myr versus $\\gtrsim 10$ Myr for single-peaked sources. The paper reads this as feedback being weak in the first few million years and strengthening dramatically at the onset of core-collapse supernovae. It also finds that homogeneous thin-shell radiative transfer models reproduce the observed Lyman-$\\alpha$ profiles while returning unphysical parameters and systematically overestimating systemic redshifts, so those models should not be trusted for outflow inference.","pith_inferences":["If the delayed-feedback reading is right, the same double-to-single transition should be observable across the lifetime of a single long-lived starburst: a galaxy observed first at $\\sim 3$ Myr and later at $\\sim 15$ Myr should shift from double- to single-peaked Lyman-$\\alpha$ while its metal absorption broadens and strengthens.","A testable extension would compare outflow momentum flux per unit star-formation rate against stellar age; the paper's picture predicts a sharp jump near 4 Myr rather than a smooth rise.","The documented overestimate of systemic redshift in shell models implies that published outflow velocities from Lyman-$\\alpha$ alone may be systematically low, and estimates of Lyman-continuum escape based on systemic Lyman-$\\alpha$ flux may be biased high.","The clumpy, multiphase absorption geometry seen here suggests that next-generation Lyman-$\\alpha$ radiative transfer should move from homogeneous shells to porosity or multi-component velocity fields; the paper's toy models are a first step in that direction."],"forward_implications":["Lyman-$\\alpha$ profile shape becomes an empirical age indicator for high-redshift starbursts: double-peaked means $\\lesssim 4$ Myr, single-peaked means $\\gtrsim 10$ Myr.","Outflow properties change by roughly a factor of three in strength and width within the first 10 million years of a starburst, so simulations should not treat feedback as constant during this window.","Expanding-shell fits that look excellent can still return unphysical neutral columns, dust optical depths, and intrinsic line widths, and they systematically overestimate systemic redshifts; outflow velocities and Lyman-continuum-leakage estimates derived from them should be re-examined.","Metal emission line ratios secure the sources as H II regions rather than AGN, so the age-feedback sequence applies to normal star-forming galaxies, not a hidden active-galaxy subpopulation."],"supporting_citations":[{"why":"Supplies the MUSE observations and catalogs of 338 gravitationally lensed Lyman-alpha emitters on which the entire sample is built.","marker":"Richard et al. (2021)"},{"why":"Provides the stellar population templates used for continuum fitting and for the P-Cygni features in the age modeling.","marker":"Bruzual & Charlot (2003)"},{"why":"Provides the photoionization model grid used to interpret emission line ratios and to predict equivalent widths in the spectral fitting.","marker":"Gutkin et al. (2016)"},{"why":"Describes the Bayesian spectral-fitting code used to infer stellar ages and ISM parameters from the stacked spectra.","marker":"Chevallard & Charlot (2016)"},{"why":"Establishes that the N V P-Cygni profile weakens within a few million years, the age diagnostic that anchors the DP versus SP age difference.","marker":"Chisholm et al. (2019)"},{"why":"Provides the expanding shell radiative transfer models that link Lyman-alpha profile morphology to outflow velocity and column density.","marker":"Verhamme et al. (2006)"},{"why":"Supplies the expanding shell model grid used to fit the observed Lyman-alpha profiles.","marker":"Gurung-López et al. (2019)"},{"why":"Supplies the MCMC fitting routines that produce the fitted shell model parameters and systemic redshift estimates.","marker":"Gurung-López et al. (2022)"},{"why":"Provides the stellar evolution timescale for core-collapse supernovae, the approximately 4 Myr onset that the age interpretation hinges on.","marker":"Woosley et al. (2002)"},{"why":"Provides the massive-star evolutionary tracks used to locate the supernova onset timescale in the stellar population modeling.","marker":"Leitherer et al. (2014)"}],"fun_headline_variants":["Double-peaked Lyman-alpha tags starbursts under 4 Myr","Young starbursts betray weak feedback via Lyman-alpha","Lyman-alpha double peaks trace delayed supernovae","Starburst age read from Lyman-alpha profile shape","Metal lines link Lyman-alpha peaks to feedback onset"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The age split that carries the argument is inferred by fitting stacked spectra with a model that assumes steady star formation, a Chabrier initial mass function extending to 300 solar masses, no significant binary evolution, and a gas density of $10^{3}$ $cm^{-3}$; if real starbursts are bursty, binary-rich, or have a lighter IMF, the roughly 4-versus-10 Myr ages and the supernova-timing interpretation could shift.","fun_headline_variants_meta":{"raw":{"variants":["Double-peaked Lyman-alpha tags starbursts under 4 Myr","Young starbursts betray weak feedback via Lyman-alpha","Lyman-alpha double peaks trace delayed supernovae","Starburst age read from Lyman-alpha profile shape","Metal lines link Lyman-alpha peaks to feedback onset"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000578,"raw_usage":{"total_tokens":2839,"prompt_tokens":1172,"completion_tokens":1667,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":788,"completion_tokens_details":{"reasoning_tokens":1582}},"tokens_in":788,"tokens_out":1667,"duration_ms":12275,"temperature":1.0,"reasoning_tokens":1582,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:48:14.057796+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the stellar age of individual double-peaked Lyman-alpha galaxies from Balmer-line or continuum-break fitting, while simultaneously measuring the width and centroid of their Si II absorption. If a galaxy with an independently measured age below roughly 4 Myr shows broad (more than about 200 km/s) strongly blueshifted absorption, the claim that feedback remains weak before core-collapse supernovae would fail; if no such young strong-outflow galaxy exists, the delayed-feedback picture is supported.","supporting_citations":[],"review_version":1}