{"id":"c9a0db38-d0ad-43d3-bc68-7d5ef043ca66","arxiv_id":"1908.04663","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Gamma-ray bursts with wider prompt-emission spectra tend to be longer, brighter, and more energetic, though the rest-frame correlation becomes significant only after removing outliers.","lead":"This paper measures the spectral width of gamma-ray burst spectra from Fermi GBM data and reports that wider spectra tend to come from longer, brighter, more energetic bursts. The finding could make spectral width a useful empirical luminosity indicator for GRBs, but the rest-frame correlation depends on post-hoc data cuts.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed W–Eiso/Liso correlation may be an artifact of brightness-dependent BEST model selection and shared k-correction, not an intrinsic spectral-width–energetics relation.","rationale":"The paper's strongest support is the set of observer-frame correlations in Table 4: W correlates with duration, fluence, and peak flux. Those are model-independent on the flux side, though W itself comes from a fitted model. The central claim, however, is the rest-frame statement about Eiso and Liso, and that is where the load-bearing risk sits. The reader's weakest_assumption points to the BEST catalog accuracy; my read agrees and sharpens it: the model-selection step itself correlates with brightness. The authors' Table 8 is an honest admission that BAND versus COMP discrimination is only about 46% confident, and Table 2 shows that the resulting W differs by model family. This creates a plausible non-physical path to the claimed correlations. I do not see an internal inconsistency in the paper; the concern is empirical robustness, and it is already reflected in the CONDITIONAL verdict. I therefore recommend no verdict change. A direct injection simulation with the published pipeline would settle whether the correlation is real or an artifact, and it is feasible with the GBM DRMs and RMFIT.","tokens_in":17105,"tokens_out":5389,"duration_ms":59807,"concrete_test":"Run an injection test with the actual GBM detector response matrices and the authors' fitting pipeline (RMFIT, ΔC-Stat crit = 11.83): simulate 10,000 spectra from a single Band model with fixed rest-frame width but with luminosities and redshifts drawn to match the observed Eiso/Liso sample. Fit each spectrum with BAND, COMP, and SBPL, assign BEST models as in the catalog, compute W and k-corrected Eiso/Liso exactly as in Section 3.3, and measure the Spearman rank correlation. If the pipeline reproduces the Table 5/7 correlations with no intrinsic W–Eiso relation, the central claim is a fitting artifact; if R drops to ~0, the observed correlation is real.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline rest-frame claim rests on spectral widths from the Fermi GBM BEST catalog, but the catalog's model choice is brightness-dependent. The authors' own simulations (Section 4, Table 8) show that when a true BAND spectrum is injected, the COMP model is preferred at only ~46% confidence, regardless of S/N; for fainter bursts the wrong, narrower COMP shape is often selected. Because COMP widths are systematically smaller than BAND/SBPL widths (Table 2: medians 0.93–1.02 for COMP versus 1.27–1.89 for BAND/SBPL), the observed W–fluence and W–peak-flux correlations in Section 3.2, and hence the rest-frame W–Eiso/Liso correlations in Section 3.3, can be produced by the selection of which fitting function is called 'BEST' rather than by a physical width–energy relation. This is compounded by Equations (4)–(6): Eiso and Liso are k-corrected with the same fitted spectral model that defines W, so any brightness-dependent bias enters both sides. The short-burst extension and the post-hoc switch to absolute width (Tables 6–7) do not remove this concern; they share the same fitted parameters.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper re-examines the spectral width W of gamma-ray burst (GRB) prompt-emission spectra, defined via the full-width at half-maximum in the EF(E) representation, using the BEST spectral fits (BAND, COMP, SBPL) from the Fermi GBM Burst Catalog. The authors compute W for both peak-flux (P) and time-integrated (F) spectra, compare the distributions for long and short bursts, and then test correlations of W with duration, fluence, peak flux, and, for a redshift-known subsample, with isotropic-equivalent energy E_iso and isotropic-equivalent peak luminosity L_iso. They report positive correlations in the observer frame, and positive rest-frame correlations for F spectra which become stronger after removing ~30% of the data as outliers and after switching from the relative width to an absolute width. The paper concludes that wider bursts have larger energy and luminosity and suggests possible use of W as a luminosity indicator.","tokens_in":17380,"tokens_out":5899,"duration_ms":55548,"significance":"If the rest-frame W–E_iso and W–L_iso correlations are genuine, they would add a new observable spectral-shape indicator to the known empirical relations (e.g., Amati, Yonetoku) and could potentially be calibrated as a distance indicator. A strength of the paper is that it explicitly performs simulations to test the reliability of the model selection, rather than taking the BEST catalog at face value. However, those same simulations identify a serious brightness-dependent model-selection bias that the paper does not correct, and the rest-frame claim rests on correlations that are modest and obtained after post-hoc data selections. The central claim is therefore plausible but not established to the standard implied by the abstract.","major_comments":[{"comment":"The simulation in Section 4 directly undercuts the main conclusion. Table 8 shows that the probability of preferring COMP over an injected BAND spectrum increases strongly with decreasing fluence: 0/10000 for GRB171010792 (fluence 6.33e-4), about 33% for GRB120711115, about 57% for GRB170705115, and about 80% for GRB081222204 (fluence 1.19e-5). Combined with Table 2, where COMP widths are systematically smaller (median 0.93–1.02) than BAND/SBPL widths (1.27–1.89), this brightness-dependent mis-selection can produce exactly the positive W–fluence and W–peak flux correlations reported in Section 3.2, and hence the rest-frame W–E_iso/L_iso correlations in Section 3.3. The problem is compounded by Equations (4)–(6), where E_iso and L_iso are k-corrected with the same fitted spectral model that defines W, so any brightness-dependent spectral bias enters both sides of the correlation. The authors acknowledge that 'the spectral parameters are affected by the spectral brightness' but do not correct for this effect or quantify how much of the measured correlation it can generate; this is load-bearing for the central claim.","section":null},{"comment":"The rest-frame correlations are not significant for P spectra in any of the parameter pairs: the Spearman coefficients range from 0.08 to 0.17 with all p-values greater than 0.15. The only significant correlations are in the F spectra, with R around 0.26–0.43. Despite this, the abstract and conclusion state without qualification that 'the spectral widths are correlated with energy and peak luminosity in GRBs with known redshifts.' The wording should be limited to the time-integrated F spectra and the modest strength of the correlations should be stated explicitly.","section":null},{"comment":"The large increase in correlation coefficients (from R≈0.3–0.4 to R>0.6) is obtained through two post-hoc procedures: first, removing the outer ~30% of the data as 'outliers' based on an iterative one-intrinsic-scatter cut, and second, replacing the relative width with the absolute width. Neither procedure is justified a priori, and the outlier removal is applied after inspecting the scatter plots and the weak correlations, which creates a serious risk of selection bias. The one-intrinsic-scatter criterion from Hyper-fit is not a standard robust-correlation estimator, and the paper does not report the final sample sizes after the cut, making the results difficult to reproduce or interpret.","section":null},{"comment":"The claim that short bursts 'extend' the long-burst correlation is based on only two (P spectra) and five (F spectra) short bursts with redshifts in the rest-frame sample. Such small numbers cannot support a claim of extension, especially given that the correlations within the long-burst subsample alone are weak (R≈0.24–0.38 for F spectra). This should be presented as a qualitative suggestion requiring a larger sample, not as a result of the present analysis.","section":null}],"minor_comments":[{"comment":"The definition of the absolute width is garbled as printed (the equation reads 'Wab = log. 2ab 21'); please provide a clear definition showing how the absolute width differs from the relative width of Equation (1).","section":null},{"comment":"The abstract states 'Our analysis results consist with the previous results'; this should be 'are consistent with.'","section":null},{"comment":"The text reports K-S probabilities of 5.14e-24 and 5.69e-43 for the long versus short width distributions, then states that the distributions are 'perfectly compatible when taking into account the variances.' This is contradictory and should be rephrased to indicate that while the medians differ significantly, there is substantial overlap.","section":null},{"comment":"For short P-spectra bursts, the W–fluence correlation is negative (R=-0.23, p=0.007), which is not discussed in the text; this apparent anti-correlation deserves comment, especially given the small number of short bursts.","section":null},{"comment":"The sentence 'The correlations of the long burst set are weaker than that of the entire burst set' has subject–verb disagreement; it should be 'those of the entire burst set.'","section":null}],"recommendation":"major_revision","confidential_remarks":"The paper was published in ApJ in 2019 and is being reviewed here as an arXiv submission; my comments should be read as a technical assessment of the manuscript as presented. The authors are clearly familiar with the catalog and have done a useful service in extending the width analysis to BEST models, but the central rest-frame correlation claim is not yet supported because of the brightness-dependent model-selection bias they themselves quantified. A revision that either corrects for this bias (e.g., by restricting to high-significance BAND fits or by modeling the selection effect) or substantially softens the claim would be appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Zhaoyang et al. report that the spectral width W of GRB prompt spectra, computed from Fermi GBM BEST fits, correlates with duration, fluence, and peak flux, and they push this to rest-frame correlations with Eiso and Liso. The observer-frame part is the real contribution. Those correlations are model-independent on one side, survive in long-burst subsamples, and the short-burst extension is a nice addition. The paper also does something rare: it tests its own assumption with simulations and reports the uncomfortable result that COMP is favored over BAND only ~46% of the time even when the input is a Band spectrum, regardless of S/N.\n\nWhere it soft: the headline rest-frame claim is not supported. First, the initial rest-frame correlations are weak (R~0.4 for F spectra, not significant for P). They become \"significant\" only after removing the outer 30% of the data—a post-hoc cut that will inflate correlation coefficients on truncated data—or after switching to absolute width, which changes the question. Second, the stress-test concern lands: the BEST model choice is brightness-dependent. Table 2 shows COMP widths are systematically ~0.9–1.0, while BAND/SBPL widths are ~1.3–1.9. Their own simulations show dimmer bursts are more likely to be misclassified as COMP. So the observed W–fluence and W–flux correlations can be generated by the catalog's model selection, and the k-corrected Eiso/Liso share the same fitted model that defines W. The paper acknowledges the uncertainty but does not quantify how much of the correlation survives when model choice is fixed.\n\nAnother soft spot: the claim that \"the wider bursts have larger energy and luminosity\" is stated more strongly than the evidence. The P spectra, which are more robust for peak flux, show no significant rest-frame correlation. The F spectra carry the signal, and F spectra are time-integrated, where model misclassification is worst.\n\nWho is this for? GRB observers interested in empirical luminosity indicators. It's an incremental addition to the Amati/Yonetoku program, and the observer-frame results are worth having. The rest-frame relation needs a cleaner sample or a fix for model selection before I'd use it.\n\nRecommendation: deserves a serious referee. The paper is honest, the data work is reproducible, and the selection-effect question is exactly what a good referee can push on. I'd send it to review with the expectation that the rest-frame section needs substantial revision or softening.","headline":"The observer-frame width correlations are solid; the rest-frame W–Eiso/Liso claim rests on a brightness-dependent model-selection effect and a post-hoc outlier cut that the paper itself cannot rule out.","tokens_in":17879,"tokens_out":2186,"would_cite":false,"duration_ms":22678,"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 paper claims that the spectral width of a gamma-ray burst's prompt-emission spectrum correlates positively with its isotropic-equivalent energy and peak luminosity, so wider bursts are more energetic and more luminous.","keywords":["gamma-ray bursts","spectral width","Fermi GBM","BEST spectral fits","isotropic-equivalent energy","peak luminosity","prompt emission"],"falsifier":"A decisive test is to simulate BAND spectra with known input luminosities and energies, fold them through the GBM detector response at a range of signal-to-noise ratios, fit them with the same COMP and BAND models used for the catalog, and recompute W from the recovered best-fit parameters; if the recovered widths no longer track the input energy and luminosity once model-selection ambiguity is included, the observed correlation is a fitting artifact rather than an intrinsic property.","tokens_in":16926,"feed_emoji":"💥","tokens_out":10669,"duration_ms":96992,"temperature":0.7,"pith_summary":"The paper sets out to establish that the width of a gamma-ray burst's prompt-emission spectrum is not a passive shape parameter but carries information about the burst's energetics. Working from the Fermi GBM catalog's best-fit ('BEST') spectra, the authors compute the relative spectral width $W=\\log(E_2/E_1)$ from the full width at half maximum of the $E F_E$ spectrum and show that $W$ correlates positively with duration, fluence, and peak flux in the observer frame. For the bursts with known redshift, they find that the correlation survives in the cosmological rest frame: wider bursts have larger isotropic-equivalent energy $E_{\\mathrm{iso}}$ and larger isotropic peak luminosity $L_{\\mathrm{iso}}$. The paper concludes that spectral shape is connected to energy and luminosity, and suggests that calibrated spectral widths could eventually serve as a distance indicator for gamma-ray bursts.","feed_headline":"Wider spectra mean brighter, more energetic gamma-ray bursts","feed_subtitle":"A Fermi GBM reanalysis ties spectral width to burst energy and peak luminosity, opening a new distance indicator.","key_machinery":"The machine carrying the argument is the spectral width $W=\\log(E_2/E_1)$, where $E_1$ and $E_2$ are the lower and upper energies at half maximum of the $E F_E$ spectrum. The paper computes $W$ from the Fermi GBM catalog's 'BEST' model fits, which include the Band function, a cutoff power law, and a smoothly broken power law; this matters because using only Band fits would overestimate the width. The authors also use an absolute width analogue and its rest-frame counterpart, showing that the correlations with $E_{\\mathrm{iso}}$ and $L_{\\mathrm{iso}}$ tighten when that quantity is used, so the width itself, not the fitted model, is doing the explanatory work.","core_discovery":"On its own terms, the paper's discovery is that the relative spectral width of GRB prompt emission, previously studied mainly as a diagnostic of radiation mechanisms, is also an energetics indicator. Using the Fermi GBM catalog's best-fitting spectral model rather than a Band-only assumption, the authors find that the median width of long bursts exceeds that of short bursts, with the short bursts extending the long-burst trend to lower widths. In the rest frame, they report Spearman correlations between $W$ and $E_{\\mathrm{iso}}$ and between $W$ and the 1024 ms peak luminosity for time-integrated F spectra, and tighter correlations when an absolute width measure is used. The authors are explicit that scatter is large and that model-selection effects contaminate the width, but they conclude that the correlations are intrinsic enough that spectral shape is tied to GRB energetics.","pith_inferences":["The paper does not test whether the correlation survives when the width and the energy or luminosity are derived from independent data; a natural extension would be to measure widths from time-resolved spectra alone and energies from bolometric light curves, then compare.","Because the correlations are stronger for time-integrated F spectra than for peak-flux P spectra, an implicit physical claim is that the emission episode as a whole carries the energetics signal; this could be tested by checking whether the relation steepens when spectra are summed over longer intervals.","A practical consequence left implicit is that a confirmed $W$–$L_{\\mathrm{iso}}$ relation gives a purely spectral distance indicator that does not require $E_p$ or a light-curve fit, potentially extending GRB cosmology to bursts without measured redshift.","The paper's own COMP-versus-BAND simulation could be repurposed as a falsifier: inject BAND spectra with known $E_{\\mathrm{iso}}$ and $L_{\\mathrm{iso}}$ through the GBM response, recover best-fit widths, and see whether the recovered $W$ still tracks the input luminosity despite the model-selection ambiguity."],"forward_implications":["If the correlation holds up, spectral width becomes a redshift-independent luminosity indicator: a measured width would place a burst on the $E_{\\mathrm{iso}}$ or $L_{\\mathrm{iso}}$ scale, and a calibrated relation could produce a Hubble diagram for GRBs.","Prompt-emission models would need to explain why more energetic bursts produce broader spectra, narrowing the space of viable radiation mechanisms beyond the blackbody and synchrotron cases the paper already disfavors.","Because short bursts extend the long-burst trend to smaller widths, the same width–energetics scaling may connect the two GRB classes despite their different progenitors.","The width–energy correlation sits alongside the established peak-energy–isotropic-energy correlation, suggesting that spectral shape encodes energetics through more than just the peak energy $E_p$."],"supporting_citations":[{"why":"Defines the relative spectral width and supplies the earlier Band-only width distributions that this paper recomputes with BEST models.","marker":"Paper I"},{"why":"Source of the Fermi GBM catalog's BEST peak-flux and time-integrated spectral fits used throughout.","marker":"Paper II"},{"why":"Defines the empirical Band function whose parameters set one family of spectral widths.","marker":"Band et al. 1993"},{"why":"Establishes that high signal-to-noise is needed to prefer BAND over COMP, motivating the simulation of model-selection bias.","marker":"Kaneko et al. 2006"},{"why":"Provides the peak-energy–isotropic-energy correlation that supports the interpretation that spectral shape is tied to energetics.","marker":"Amati et al. 2002"},{"why":"Shows that multicomponent fits shift both spectral indices and flux, which the authors use to argue the width–luminosity correlation is not an artifact of single-component fits.","marker":"Guiriec et al. 2011"},{"why":"Provides the regression approach used to estimate intrinsic scatter and identify outliers in the width–energy and width–luminosity relations.","marker":"Robotham & Obreschkow 2015"}],"fun_headline_variants":["GRB spectral width correlates with energy and peak luminosity","Wider gamma-ray bursts shine brighter and pack more energy","Short bursts extend width–luminosity link found in long GRBs","Spectral width is an energetics probe for gamma-ray bursts"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result depends on the Fermi catalog's best-fit spectral parameters being reliable: if those fits are biased for faint bursts, both the measured width and the derived energy and luminosity are biased in ways that could create or distort the correlation.","fun_headline_variants_meta":{"raw":{"variants":["GRB spectral width correlates with energy and peak luminosity","Wider gamma-ray bursts shine brighter and pack more energy","Short bursts extend width–luminosity link found in long GRBs","Spectral width is an energetics probe for gamma-ray bursts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000367,"raw_usage":{"total_tokens":1970,"prompt_tokens":942,"completion_tokens":1028,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":558,"completion_tokens_details":{"reasoning_tokens":958}},"tokens_in":558,"tokens_out":1028,"duration_ms":10457,"temperature":1.0,"reasoning_tokens":958,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:35:28.360307+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test is to simulate BAND spectra with known input luminosities and energies, fold them through the GBM detector response at a range of signal-to-noise ratios, fit them with the same COMP and BAND models used for the catalog, and recompute W from the recovered best-fit parameters; if the recovered widths no longer track the input energy and luminosity once model-selection ambiguity is included, the observed correlation is a fitting artifact rather than an intrinsic property.","supporting_citations":[{"cited_title":"L., Matteson, J., Ford, L., et al","cited_arxiv_id":null,"evidence_quote":"Defines the empirical Band function whose parameters set one family of spectral widths."},{"cited_title":"D., Briggs, M","cited_arxiv_id":null,"evidence_quote":"Establishes that high signal-to-noise is needed to prefer BAND over COMP, motivating the simulation of model-selection bias."},{"cited_title":"2002, A&A, 390, 81","cited_arxiv_id":null,"evidence_quote":"Provides the peak-energy–isotropic-energy correlation that supports the interpretation that spectral shape is tied to energetics."},{"cited_title":"2011, ApJL, 727, L33","cited_arxiv_id":null,"evidence_quote":"Shows that multicomponent fits shift both spectral indices and flux, which the authors use to argue the width–luminosity correlation is not an artifact of single-component fits."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the regression approach used to estimate intrinsic scatter and identify outliers in the width–energy and width–luminosity relations."}],"review_version":1}