{"id":"3689448d-e988-472f-be80-b0689f1552d4","arxiv_id":"2504.17597","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A new LTE model atmosphere grid, NewEra, provides 37,438 spherical stellar models and synthetic spectra with updated atomic and molecular line data, publicly released in HDF5 format.","lead":"The PHOENIX team has released NewEra, a new grid of 37,438 computer models of stars and their spectra, built with updated atomic and molecular opacity data. It is designed as a community resource for analyzing stellar spectra, stellar populations, and exoplanet transit light curves.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The molecular-line reduction is tested only at 25 Å for a single model; the released high-sampling spectra are not protected by that check.","rationale":"The strongest part of the paper is the completeness of the release: 37,438 models, HDF5 files with restart info, MD5 checksums, and access scripts, plus the 10^-6 cross-platform reproducibility tests in Section 4.1. These support the existence claim. The scientific use of the spectra, however, depends on the line-list reduction. The paper's own convergence check is the only evidence that the 20.5B-line list reproduces the full opacity, and it is a single 25 Å flux test. The HSR product is explicitly claimed to have λ/Δλ > 10^6, so users will degrade it to high resolutions; a 25 Å flux match does not guarantee that omitted lines are irrelevant at high resolution. Observational validation would also be welcome, but it is a separate step and the LTE caveat is stated; the line-list reduction is more internal to the claim that the released spectra are what the full input physics would produce. The concrete test would either confirm the reduction or reveal a gap that should be documented. This leaves the verdict at CONDITIONAL rather than ACCEPT, because the release itself is sound but the accuracy claim for the highest-resolution products needs a stronger check.","tokens_in":18311,"tokens_out":9023,"duration_ms":93878,"concrete_test":"For at least one cool model (Teff=2300 K, log g=4.5, solar) and one warmer molecular-rich model (e.g., Teff=4000 K, log g=4.0, solar), recompute the spectrum at the native HSR sampling of Section 4.6 with both the 823.8B-line full list and the released 20.5B-line list, and report the maximum |ΔF|/F in 0.8-1.4 μm and 3-5 μm before any 25 Å convolution. If the maximum exceeds 0.4% or shows line-like residuals, the claim that the reduced list preserves opacity needs qualification.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.2 reduces the molecular line list from 823.8 billion to 20.5 billion lines using an opacity-estimate cut and then by omitting molecules judged absent for Teff >= 2300 K. The only validation is a flux comparison at 25 Å resolution for one model (Teff=2300 K, log g=4.5, [M/H]=0), with max difference below 0.4%. The published HSR product has sampling rate at least 10^6 (Section 4.6, Table 3). A 25 Å flux match does not establish that the omitted ~91% of the intermediate list (or, in the full-to-reduced comparison, the omitted weak lines) are negligible at high spectral resolution, where individual weak lines can still be visible, nor does it test other grid corners where different molecules dominate (e.g., TiO/VO bands near 4000-5000 K, or metal-poor models where line-to-continuum ratios differ). The criterion for omitting 'molecules not present' is not quantified, so the reduction is not independently reproducible. Hence the load-bearing premise that the reduced list preserves the opacity of the full database is not established for the highest-resolution data products, even though it is adequate for 25 Å comparisons.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents the NewEra LTE grid of 37,438 spherically symmetric PHOENIX/1D model atmospheres and synthetic spectra, covering Teff 2300–12000 K, log g 0.0–6.0, [M/H] −4.0 to +0.5, and alpha-element variations. The input physics uses the ACES equation of state, about 851 million Kurucz atomic lines, and about 823.8 billion molecular lines, mostly from Exomol, reduced to roughly 20.5 billion lines. The paper documents the atmospheric structures, spectral energy distributions, limb darkening, and comparisons to the NextGen, GAIA DR1, and ACES grids. The main scientific claims are that the grid is computed, publicly released in HDF5 format with high- and low-sampling spectra, and that the spectra differ substantially from previous grids, mostly because of updated molecular line lists.","tokens_in":18603,"tokens_out":5082,"duration_ms":50981,"significance":"If the grid is accurate, it will be a valuable community resource for stellar spectroscopy, stellar populations, and transit light-curve analysis. The paper has clear strengths: the grid construction is carefully documented, cross-platform reproducibility is checked to 10^-6 relative accuracy, the line-selection procedure is described in detail, the data release includes checksums and access software, and the 37,438-model grid is a substantial extension of the ACES parameter space. The main weakness is that the central claim of improved spectra is not validated against observed stellar spectra or flux-calibrated observations; the comparisons are exclusively against older synthetic grids. The paper is transparent about the LTE approximation and about the hydrogen line dissolution calibration being in preparation, but these caveats do not by themselves replace external validation. As a data-release paper the construction is largely sound; the missing observational anchor is what prevents me from recommending acceptance without revision.","major_comments":[{"comment":"The molecular-line reduction is validated only at 25 Å resolution for a single model (Teff=2300 K, log g=4.5, [M/H]=0), while the primary delivered product is the HSR spectrum with sampling rate at least 10^6 (Table 3) and the supplied GAIA/JWST convolutions use Gaussian filters with 0.1–2 Å scales. A flux agreement below 0.4% at 25 Å does not establish that the omitted weak lines are negligible in the high-resolution data, where individual weak lines remain visible, nor does it test other grid corners where different molecules dominate, such as TiO/VO bands near 4000–5000 K or metal-poor models with different line-to-continuum ratios. Additionally, the thresholds ('10^-3 to 10^-4 depending on species') and the criterion for 'molecules not present' are not itemized, so the reduction is not independently reproducible. Please provide high-resolution flux comparisons for several grid corners and quantify the per-species thresholds and molecule-omission criterion.","section":"Section 3.2, Table 3"},{"comment":"The abstract concludes that the spectra show significant differences 'mostly due to the updates of the molecular line lists,' and Section 4.4 presents these differences as improvements, but no NewEra spectrum is compared with observed stellar spectra or observed flux-calibrated SEDs. The only external anchor mentioned, the hydrogen line dissolution calibration against Sirius A, is cited as 'in preparation.' As a result, the paper supports the statement that the grid differs from older grids, but not that it is more accurate. Please add comparisons to observed spectra for a small set of representative stars spanning the grid, or explicitly reword the conclusions to present the grid as an internally consistent update whose observational validation remains to be demonstrated.","section":"Section 4.4, Abstract"},{"comment":"The LTE assumption up to Teff=12000 K is flagged as an approximation with an NLTE grid in preparation, but the caveat is not quantified. For effective temperatures above roughly 8000–9000 K, non-LTE effects on Balmer lines and ionization equilibria are known to be important, and users of the released HSR spectra cannot judge where the LTE assumption degrades. I request either quantitative statements (for example, comparisons of selected models with existing NLTE calculations) or a clear statement in the abstract and data documentation restricting the reliable temperature range for spectroscopic applications.","section":"Sections 1 and 5"}],"minor_comments":[{"comment":"The phrase 'intervals based sampling rate' is awkward and should be rephrased, for example as 'The HSR spectra have a sampling rate lambda/delta-lambda of at least one million.'","section":"Section 4.6"},{"comment":"The statement that NewEra and ACES spectra differ little in the near-IR because of very similar water line data sits in some tension with the broad claim that molecular line updates drive the large differences; please clarify which wavelength regions are dominated by which molecular species and why the near-IR agreement does not contradict the overall conclusion.","section":"Section 4.4"},{"comment":"The LSR wavelength range is described as extending from the soft X-ray to the radio region, but Table 4 ends at 1000 microns (sub-millimeter); please adjust the wording to match the actual coverage.","section":"Section 4.6"},{"comment":"Given that limb darkening is a stated motivation for the grid, making limb darkening data available only 'upon request' is a limitation for users of the release; consider providing at least a representative set of center-to-limb intensity profiles in the HDF5 files.","section":"Section 6"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of Astronomy & Astrophysics and the construction is careful, but the 'improved' claim needs observational grounding before community adoption. I do not see a circularity problem or an internal inconsistency; the concern is external validation of the reduced line list and of the model spectra. A revision that adds a focused comparison to observed spectra and strengthens the reduction-validation section would make the paper suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"If you work with stellar model grids, this is a paper you should know about. It releases a new 37,438-model PHOENIX grid, NewEra, with updated Kurucz atomic and Exomol molecular line lists, computed in spherical symmetry, with spectra available in HDF5. The central claim is that the spectra differ meaningfully from previous grids, mostly from the molecular data, and that claim is backed by direct comparisons.\n\nWhat is actually new is the data product itself. The parameter coverage is the same as the ACES grid (Husser et al. 2013); the novelty is the input physics and the public release. They document the line selection, the EOS, cross-platform reproducibility to 1e-6, and the file format in enough detail that users can recreate or at least understand the models. That is real, reproducible work—code, data, and checksums are shipped.\n\nWhat the paper does well: it is careful about code consistency across HPC systems, it separates structure differences from spectral differences (structures close to ACES, spectra different), and it is explicit about its own limitations—LTE up to 12000 K, hydrogen line dissolution calibration still in preparation, limb-darkening data on request rather than in the release. That honesty matters.\n\nThe biggest soft spot is validation. The paper compares NewEra to older grids, but not to any observed stellar spectrum, so \"improved\" is not actually demonstrated—only \"different.\" For a community resource that people will use to interpret GAIA and transit data, that is a real gap. The stress-test concern is related but narrower: the molecular line-list reduction from 823 billion to 20.5 billion lines is validated only by a flux comparison at 25 Å resolution for one model (2300 K, log g 4.5, solar). The high-sampling-rate spectra have sampling rates up to 10^6, and a 25 Å match does not guarantee that weak lines omitted at high resolution are negligible, especially in other grid corners (TiO/VO bands, metal-poor models). That is a legitimate technical concern, and it would be good to see the reduction checked at higher resolution or at more grid points. It is not fatal—the test is an honest internal check and the authors do not overclaim it—but it should be flagged as a limitation and, ideally, addressed before final submission.\n\nWho this is for: anyone using model grids for SED fitting, stellar parameter inference, population synthesis, or transit light curves. It is a community resource paper, not a conceptual advance.\n\nRecommendation: send it to peer review, yes. It deserves serious refereeing. A minor-to-moderate revision should ask for (a) observational validation, even on a small sample, and (b) a higher-resolution check of the line-list reduction, or an explicit caveat about the HSR product.","headline":"A well-documented, genuinely useful new PHOENIX model grid with updated opacities; the main soft spot is that validation is internal (line-list reduction tested only at 25 Å for one model) rather than against observed spectra.","tokens_in":19115,"tokens_out":2356,"would_cite":true,"duration_ms":22658,"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 NewEra LTE grid—37,438 spherical stellar atmosphere models—is now public, with spectra substantially changed by updated molecular line lists.","keywords":["stellar atmosphere models","synthetic spectra","LTE model grid","PHOENIX","ACES equation of state","molecular line lists","limb darkening","HDF5"],"falsifier":"Take a well-observed benchmark star with known parameters, such as the Sun or a nearby M dwarf, and compare high-resolution observed spectra in molecular bands (for example water or FeH in the optical or near-infrared) to the corresponding NewEra model after proper convolution; wavelength-correlated residuals beyond the 0.4% flux level quoted for the 25 Å convergence test would show the reduced line list misses opacity that matters at high resolution.","tokens_in":18158,"feed_emoji":"🌟","tokens_out":9591,"duration_ms":77066,"temperature":0.7,"pith_summary":"This paper presents and publicly releases the NewEra LTE grid: 37,438 spherical stellar atmosphere models and their synthetic spectra covering effective temperatures 2300 K to 12000 K, surface gravities $\\log g$ 0.0 to 6.0, metallicities [M/H] −4.0 to +0.5, plus $\\alpha$-element variations. The grid is built on about 851 million atomic lines and 834 billion molecular lines, reduced per model to a tractable working list, and it uses a self-consistent chemical-equilibrium equation of state. The models' internal structures differ only slightly from the previous grid generation, but the predicted spectra differ significantly, especially in optical cool-star regions, because the molecular line data are much improved. The HDF5 release, with high- and low-sampling spectra, is intended to replace earlier grids in stellar spectrum analysis, stellar population synthesis, and transit light-curve fitting.","feed_headline":"37,438 spherical stellar atmosphere models go public","feed_subtitle":"Updated molecular opacities shift cool-star optical spectra; the release replaces older grids for transit and population work.","key_machinery":"The load-bearing machinery is the PHOENIX/1D stellar atmosphere code in LTE and spherical symmetry, driven by three input blocks: the ACES chemical-equilibrium solver for the equation of state, the atomic line database with 851 million lines, and a molecular database, dominated by Exomol, with 834 billion lines. A line-selection step compares each candidate line's central opacity to the local continuous opacity at reference layers and keeps lines whose line-to-continuum opacity ratio exceeds $10^{-4}$; strong lines get Voigt profiles while weaker lines get Gauss profiles. This reduces the list actually used per model to roughly 800,000 to 2.5 million atomic lines and about 20.5 billion molecular lines. The HDF5 delivery format is the final part of the machinery: one file per model contains the restartable atmosphere structure, the input namelist, and high-sampling spectra from 900 Å to 30 µm at sampling rates above one million except in parts of the 5.8–30 µm range.","core_discovery":"On its own terms, the paper's central claim is that the new NewEra LTE grid delivers substantially more realistic synthetic spectra than earlier PHOENIX grids while keeping the parameter coverage and structure of the well-used ACES grid. Most of the spectral change comes from new molecular line opacities, not from changes in the atmospheric structure. A second, quantitative claim is that the adopted line-selection reduction, from 823.8 billion molecular lines down to about 20.5 billion actually used lines, changes the emitted flux at 25 Å resolution by less than 0.4%, so the grid is considered converged relative to the full line list. In addition, all models use spherical symmetry, which changes center-to-limb variation in a way that matters for both giants and, to a lesser degree, dwarfs; this affects limb-darkening modeling in transits. The paper does not compare any NewEra spectrum to an observed stellar spectrum; calibration of hydrogen line dissolution is cited as in preparation.","pith_inferences":["A testable consequence of better molecular lists is that elemental abundances derived from M-dwarf spectra will shift systematically relative to older grids; a benchmark sample with independent abundance constraints would reveal whether the shifts are improvements.","The 25 Å convergence test does not by itself guarantee convergence at the highest sampling rates delivered; users working near the sampling limit should verify that omitted weak lines do not affect narrow-band features such as exoplanet transmission windows.","The stated plan for an NLTE version implies that LTE residuals will soon be quantified for hotter stars; until then, the LTE grid is the natural baseline against which those NLTE corrections will be measured.","Because limb-darkening data are only available upon request, a natural follow-up is precomputed limb-darkening coefficients; wavelength-dependent spherical limb darkening should measurably improve transit depth precision for small planets around cool stars."],"forward_implications":["Analyses currently using ACES, GAIA DR1, or NextGen spectra can be rerun on NewEra, with the largest changes in optical cool-star spectra where molecular opacities dominate.","Transit light-curve fits relying on limb-darkening coefficients need updated values from spherical models, especially for giants and for dwarfs where the wavelength-dependent apparent radius matters.","The HDF5 files make it possible to reconstruct the exact model structure and input line lists, so derived abundances or C/O ratios can be traced back to specific line data.","The GAIA-compatible and JWST-compatible low-sampling spectra allow direct use in population synthesis and JWST archive tools without re-gridding.","The newly calibrated hydrogen line dissolution will shift Balmer-line predictions and hotter-star analyses relative to earlier grids."],"supporting_citations":[{"why":"supplies the predecessor ACES grid: the parameter range, the Teff-log(g)-mass relation, and the baseline for spectrum comparison","marker":"Husser et al. 2013"},{"why":"introduces the ACES chemical-equilibrium solver used for the equation of state","marker":"Barman et al. 2011"},{"why":"provides the Exomol molecular line database that dominates the molecular opacity","marker":"Tennyson et al. 2016"},{"why":"supplies the atomic line database used for all atomic opacity","marker":"Kurucz 2017"},{"why":"establishes the line-selection procedure the grid evolves, selecting lines by line-to-continuum opacity ratio","marker":"Allard & Hauschildt 1995"},{"why":"defines the NextGen grid, one of the previous generations against which NewEra spectra are compared","marker":"Hauschildt et al. 1999b"},{"why":"defines the GAIA DR1 grid, another previous generation used for comparison at low effective temperatures","marker":"Kučinskas et al. 2005"},{"why":"used to argue that PHOENIX/1D limb darkening matches observed transit light curves","marker":"Kreidberg et al. 2014"},{"why":"provides the data-access software and DOI to download individual models","marker":"Hauschildt et al. 2025"}],"fun_headline_variants":["New stellar atmosphere grid with 37,438 spherical models","Spherical models and new opacities: NewEra grid released","37K spherical stellar models, updated molecular opacities","NewEra grid: 37,438 models cover cool to hot stars"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the line opacities—the reduced 20.5-billion-line molecular list, the atomic database, and the LTE approximation up to 12000 K—preserve the real opacity of stellar atmospheres closely enough that the released spectra and limb darkening match actual stars; the paper checks convergence of the reduction at 25 Å resolution and code consistency, but does not compare any model to an observed spectrum.","fun_headline_variants_meta":{"raw":{"variants":["New stellar atmosphere grid with 37,438 spherical models","Spherical models and new opacities: NewEra grid released","37K spherical stellar models, updated molecular opacities","NewEra grid: 37,438 models cover cool to hot stars"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000507,"raw_usage":{"total_tokens":2543,"prompt_tokens":1085,"completion_tokens":1458,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":701,"completion_tokens_details":{"reasoning_tokens":1393}},"tokens_in":701,"tokens_out":1458,"duration_ms":11472,"temperature":1.0,"reasoning_tokens":1393,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T10:35:38.434558+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a well-observed benchmark star with known parameters, such as the Sun or a nearby M dwarf, and compare high-resolution observed spectra in molecular bands (for example water or FeH in the optical or near-infrared) to the corresponding NewEra model after proper convolution; wavelength-correlated residuals beyond the 0.4% flux level quoted for the 25 Å convergence test would show the reduced line list misses opacity that matters at high resolution.","supporting_citations":[{"cited_title":"O., Wende-von Berg, S., Dreizler, S., et al","cited_arxiv_id":null,"evidence_quote":"supplies the predecessor ACES grid: the parameter range, the Teff-log(g)-mass relation, and the baseline for spectrum comparison"},{"cited_title":"N., Al-Refaie, A","cited_arxiv_id":null,"evidence_quote":"provides the Exomol molecular line database that dominates the molecular opacity"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the atomic line database used for all atomic opacity"},{"cited_title":"L., Désert, J.-M., et al","cited_arxiv_id":null,"evidence_quote":"used to argue that PHOENIX/1D limb darkening matches observed transit light curves"}],"review_version":1}