Pith. sign in

REVIEW 4 major objections 7 minor 85 references

Direct physical stellar spectrum models are now fast enough to sit inside survey-scale fitting without any label-to-flux emulator.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-31 22:40 UTC pith:5D2Y45AG

load-bearing objection Solid engineering paper: real second-scale Kurucz synthesis/atmospheres without a flux emulator, public code, and measured parity; the APOGEE science claim is thinner than the compute claim. the 4 major comments →

arxiv 2607.24141 v1 pith:5D2Y45AG submitted 2026-07-27 astro-ph.SR astro-ph.IM

The Payne Zero Project I: Stellar Spectra from Physical Models in Seconds

Yuan-Sen Ting , Elliot M. Kim This is my paper
classification astro-ph.SR astro-ph.IM
keywords stellar atmospheresstellar spectral linesspectral synthesisastronomy softwarecomputational methodsabundance fittingline-list calibrationGPU computing
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Large stellar surveys measure millions of spectra, but a classical self-consistent atmosphere plus spectrum can still take tens of minutes per star. That cost drove the field toward grids, spectral emulators, and data-driven flux models that stand between stellar labels and observed light. This paper reorganizes one-dimensional LTE Kurucz atmosphere and synthesis calculations so broad wavelength work runs as GPU-native batches and atmosphere iterations run on multicore CPUs, with learned starts that cut how many physical passes are needed. A 300–1000 nm solar spectrum at high sampling takes about 14 seconds on one H100 GPU, the APOGEE near-infrared window about one second, and final spectra stay in practical parity with the original Fortran codes on tested dwarfs and giants. Those speeds let the physical forward model itself drive many-element abundance fits and joint calibration of more than 100,000 line-strength and damping corrections, bringing direct physical fitting and atomic-data updates to survey scale.

Core claim

Payne Zero shows that the supported one-dimensional LTE Kurucz physics, rearranged for GPU-native wavelength-parallel synthesis and multicore atmosphere iteration and validated against the original programs, is fast enough that direct synthesis can live inside an optimizer. No label-to-flux emulator is required: multi-element fits to reduced APOGEE spectra take under a minute per star on an H100 with independent CPU atmosphere checks, and the same computational graph jointly calibrates over 10^5 oscillator-strength and damping corrections on the Sun and Arcturus in about one minute, while spectra remain in practical parity across the tested regimes.

What carries the argument

Wavelength-parallel GPU synthesis: once the depth structure is fixed, independent wavelength columns share batched continuum, line-opacity, and transfer kernels on resident catalogs, with instrument shift, broadening, line-spread function, and sampling kept on-device at negligible extra cost. Multicore CPU atmosphere iteration plus a learned six-profile initializer only proposes the start; the physical solver still converges the accepted atmosphere. Automatic differentiation through that graph updates huge coupled sets of atomic corrections at once.

Load-bearing premise

That practical flux agreement with the original Kurucz codes on tested stars, plus always re-converging a learned starting atmosphere, is enough for the accepted fixed point to be the right physical structure across the labels and stars used in fitting—including regimes where cool dense models can lose local stability.

What would settle it

On a held-out benchmark sample with independently known parameters and abundances, run the full direct-synthesis fit and line calibration end to end; the claim fails if recovered labels or calibrated spectra disagree with those benchmarks beyond the paper’s reported residual levels, or if cool high-gravity cases diverge or mis-converge relative to the original codes on matched starts.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Survey analyses can evaluate self-consistent physical spectra inside the optimizer instead of rebuilding emulators when labels or physics change.
  • Oscillator strengths and damping can be recalibrated jointly at scale through radiative transfer rather than transition by transition.
  • New abundance coordinates or atomic parameters can enter the fit without regenerating synthetic libraries.
  • Atmosphere convergence and GPU synthesis can run on separate resources, so catalog cost is set mainly by the synthesis search.
  • Model–data residuals become tests of named physical inputs—line data, structure, missing opacity—rather than absorbed latent flux corrections.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Once synthesis is cheap, the binding limit on abundance zero points shifts from runtime to which benchmark standards, weights, and laboratory priors define the shared line catalog.
  • The large abundance shifts when temperature and gravity are freed imply that spectrum-only fits will need external temperature and gravity anchors before claiming a new absolute chemical scale.
  • The cool-dwarf stability edge means production pipelines will need regime-aware solver routes that keep the same physical acceptance tests, not one fixed-point loop for every star.
  • The same differentiable graph could co-optimize instrument and continuum nuisance parameters with labels without ever inserting a cached flux emulator.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The manuscript presents Payne Zero, a GPU- and multicore-CPU reorganization of one-dimensional LTE Kurucz atmosphere and synthesis calculations. It reports a 300–1000 nm solar synthesis at R_grid=300,000 in about 14 s on one H100 versus about 763 s for the original Fortran chain, APOGEE-band synthesis in about 1 s, 6–7-fold faster atmosphere iterations, and learned atmosphere initializers that reduce iteration counts. Parity with the original codes is demonstrated on four stellar types, with pooled normalized-flux rms near 9×10^-4 and 99th-percentile residuals of a few ×10^-3. The paper then demonstrates a controlled five-label direct fit, joint calibration of 101,124 oscillator-strength and damping corrections against the Sun and Arcturus, and a 12-abundance fit to 1,600 APOGEE giants in about 40 s of GPU search per star. The code is public.

Significance. If the inference claims are established at the level of the timing claims, this would be a substantial advance: direct synthesis could replace label-to-flux emulators in the demonstrated LTE regime, while coupled abundances, instrument effects, and atomic-data corrections remain explicit. The manuscript gives unusually concrete computational evidence: measured head-to-head wall times, residual maps against the original Fortran implementation, GPU-resident instrument operators, a frozen-sample APOGEE transfer test, and public code. The engineering core is therefore credible and potentially broadly useful. The weaker part is validation of the fitting and calibration layer—particularly the unconverged-atmosphere search, physical convergence across the survey sample, and identifiability of the enormous line-correction vector.

major comments (4)
  1. [§5, Algorithm 1 and Fig. 7] The emulator-free fitting claim depends on evaluating f_fast on an initialized but unconverged atmosphere and then holding Δf=f_phys−f_fast fixed for one bounded correction. The paper supports the required local behavior with a single K-giant mock at S/N=100, R_grid=20,000, five labels, and no instrument model. Please map the amplitude and label derivatives of Δf across representative trust regions and stellar regimes, and run a mock ensemble with multiple starts, APOGEE-like abundances/instrument effects, and difficult initializer regimes. Report label bias, correction acceptance, and cases requiring iteration. Otherwise the survey-scale inference claim should be restricted to the demonstrated case.
  2. [§3.1 Eq. (16), §4.3, and §7] The claim that every accepted model uses a physically converged atmosphere needs aggregate evidence. Equation (16) tests only deep-layer temperature change and is explicitly described as local; flux balance is said to be checked separately, but no ε_F distribution is shown. The restart evidence uses 16 cases for the five/eight-label initializers and another 16 for the direct-abundance family, while §8.3 reports unstable cool high-gravity controls. For the APOGEE sample and withheld atmospheres, please report iteration counts, ε_T and Eq. (14) flux residuals, failures/fallbacks, and convergence from independent starts, and identify which initializer was used in §7.
  3. [§6, Table 5 and Figs. 8–10] The 101,124-correction fit demonstrates optimization capacity, but not yet the physical identifiability implied by calling the released product a calibrated line list. Two fixed standard-star atmospheres constrain strongly degenerate oscillator-strength and damping corrections, while line centers and lower excitation energies are fixed. The frozen APOGEE control improvement is valuable, but please add held-out lines/regions or standards, report correction uncertainties or correlations and adopted priors/bounds, and test sensitivity to the standard-star labels and atmospheres. Until then, describe the product as an effective two-standard correction rather than a generally calibrated atomic-data set.
  4. [§7 and Appendix A, Figs. 12–15] The main APOGEE result is conditional on frozen DR14 T_eff, log g, and ξ. When those parameters are fitted, the median scale moves by −200 K and +0.17 dex, with element zero-point changes as large as about 0.22 dex. Thus Figs. 12–13 establish preservation of chemical morphology, not an independent absolute abundance scale. This limitation should be stated prominently in the abstract/main conclusions, with per-star uncertainties and convergence/selection cuts. Ideally the free-label fit, or one externally anchored by photometry/parallax and benchmark stars, should be presented as a coequal demonstration.
minor comments (7)
  1. [§2.3 and §3.3] Please give the full timing protocol in one place: number of repetitions, warm-up and one-time preparation treatment, floating-point precision, PyTorch/CUDA/Numba versions, compiler and flags for the Fortran reference, memory limits, and repository commit hashes. This would make the otherwise well-documented wall-time comparisons easier to reproduce.
  2. [§2.3, Fig. 1] “Practical parity” would benefit from a quantitative definition. The red-giant 99th-percentile residual of 5.8×10^-3 may be relevant for weak features, so please show whether residuals are wavelength- or feature-dependent and consider expanding the parity sample beyond four stars.
  3. [§4.2] The choices K=160 and λ_∇, λ_τ, and λ_hse are largely asserted. A short sensitivity test or reference to validation results would clarify how much restart performance depends on these hyperparameters.
  4. [§6] Please summarize the distributions and bounds of the fitted δq,gf and damping corrections, including the largest changes and their wavelength/species locations. This would help readers distinguish plausible effective corrections from extreme compensating values.
  5. [§7] The instrument demonstration averages the LSF over six representative fibers and uses one Gaussian broadening parameter. Please quantify the residual sensitivity to fiber/exposure-dependent LSF variation and discuss possible covariance between v_b and abundances.
  6. [Figures 11–13] The captions should state how many selected stars converged, how many were excluded, and whether any objective-quality cuts were applied before plotting. Per-star abundance uncertainties would also make the population comparison easier to interpret.
  7. [Throughout] Some notation is dense even with Table 1; in particular, the fast/physical spectra, Q objective, and atmosphere states could be defined once in a compact fitting subsection and then used consistently. A brief repository guide linking scripts to each figure and table would also help reproduction.

Circularity Check

0 steps flagged

No significant circularity: speed, parity, and survey morphology are checked against external Fortran and ASPCAP benchmarks; line corrections are openly fitted calibrations, not predictions.

full rationale

Payne Zero is a methods/software paper whose load-bearing claims are empirical wall-clock measurements and residual comparisons against the original serial Kurucz Fortran chain, plus morphological comparison of multi-element APOGEE fits to the independent ASPCAP/DR14 catalog. Synthesis and atmosphere parity (Fig. 1; 99th-percentile residuals of order a few×10^{-3}) and speedups (Figs. 2–4) are defined relative to that external reference implementation, not by construction from the authors’ own outputs. The learned atmosphere initializer is trained on physically converged atmospheres and is explicitly not accepted as the final state; the physical fixed-point iteration must still pass convergence checks. Line-list work (§6) jointly fits >10^5 oscillator-strength and damping corrections to Sun and Arcturus FTS atlases and reports mean-square residual reductions—this is labeled calibration and transfer testing (e.g., median Q drop on a matched APOGEE control), not an independent first-principles prediction. The controlled mock fit (§5) recovers labels when the generating and fitted models are identical, which is a consistency check. Self-citations (Kim & Ting 2026 lineage; prior Payne papers) supply software ancestry and motivation, not a uniqueness theorem or a forced ansatz that substitutes for the external benchmarks. No step reduces a claimed prediction to its fitted inputs by definition. Concerns about the Δf-local-constancy assumption in the fast optimizer and cool-star fixed-point stability are correctness/robustness issues, not circularity.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 2 invented entities

The work inherits the full 1D plane-parallel LTE Kurucz physical stack and treats computational reorganization, learned starts, and differentiable line calibration as the contribution. Load-bearing modeling choices (LTE, mixing-length, fixed selection fraction, frozen stellar parameters in the main survey demo, standard-star line corrections with fixed atmospheres) are domain assumptions or fitted effective parameters, not new physics entities.

free parameters (6)
  • 101124 line-parameter corrections (δq,gf and three damping families over 25281 groups) = shared Sun–Arcturus solution after 40 L-BFGS steps (~64 s on H100)
    Jointly optimized against Sun and Arcturus FTS atlases; effective astrophysical calibrations, not lab constants.
  • atmosphere initializer network weights (five-, eight-, and direct-abundance families) = K=160 PCA coefficients; family-specific λ∇, λτ, λhse
    Trained on corpora of 5e4–8e4 converged atmospheres; only used to propose x(0), but affect iteration count and basin.
  • line selection fraction η_sel = 10^{-3}
    Fixed continuum-referenced keep threshold controlling which lines enter opacity sampling.
  • deep-layer temperature convergence threshold ε_T = 5×10^{-4}
    Production stopping test on layers 40–75; defines when an atmosphere is accepted.
  • fast-fit correction threshold δQ_min and continuum Legendre order = δQ_min=0.005 (controlled diagnostic); 4th-order Legendre per APOGEE detector
    Controls when a second physical atmosphere is launched and how survey continuum is profiled per detector.
  • adopted Sun/Arcturus labels, Gaussian dispersions, and equal star/epoch calibration weights = e.g. Sun 5777 K, 4.44, 0.7 km/s; Arcturus 4286 K, 1.66, [M/H]=-0.52; equal total weight
    Fixed during inner line calibration; set the effective gf/damping solution transferred to APOGEE.
axioms (5)
  • domain assumption One-dimensional plane-parallel LTE Kurucz atmosphere and synthesis physics (no NLTE or turbulent-pressure branches) adequately represent the tested dwarf/giant regimes for the claimed parity and fitting demos.
    Stated scope in §§1–3 and Discussion §8; all speed and science claims are inside this model class.
  • domain assumption Opacity-sampling radiative transfer with mixing-length convection and the stated EOS/molecular equilibrium closes a unique usable fixed point for the production 80-layer Rosseland grid when the deep-layer ε_T test passes.
    Atmosphere iteration structure Eqs. 10–16; cool-star counterexamples in §8.3 show the assumption can fail.
  • standard math Wavelengths are independent given fixed depth structure, so GPU batching of the synthesis equations does not change the mathematical solution relative to serial Kurucz.
    §2.1–2.2; validated empirically by synthesis-only residuals in Fig. 1.
  • ad hoc to paper Freezing DR14 Teff, log g, and microturbulence isolates a meaningful many-element abundance and instrument problem for the main APOGEE demonstration.
    §7 explicitly retains published stellar parameters; Appendix A shows median −200 K / +0.17 dex shifts and large [X/Fe] zero-point moves when freed.
  • domain assumption A shared Sun+Arcturus effective line calibration transfers usefully to APOGEE giants without per-survey zero-point abundance recalibration.
    §6–7; authors note remaining zero-point differences vs ASPCAP and call for broader standard-star sets.
invented entities (2)
  • Payne Zero computational graph (GPU synthesis + multicore atmosphere + instrument operators) independent evidence
    purpose: Re-express existing Kurucz physics for resident, parallel execution and differentiable line calibration.
    Software/method entity, not a new physical field or particle; validated by parity with Fortran.
  • Learned six-field PCA atmosphere initializers (label→K=160 coefficients→decoded profiles) independent evidence
    purpose: Place the physical fixed-point iteration inside a smaller convergence basin to cut iteration count.
    New ML component; accepted atmospheres still require physical convergence checks, so not a replacement physics model.

pith-pipeline@v1.2.0-grok45-kimik3 · 40031 in / 4431 out tokens · 96380 ms · 2026-07-31T22:40:43.437262+00:00 · methodology

0 comments
read the original abstract

Modern stellar surveys measure millions of spectra, yet one self-consistent atmosphere and spectrum can require tens of minutes. This cost has motivated grids, spectral emulators, and data-driven models. We present Payne Zero, which reorganizes one-dimensional LTE Kurucz calculations for GPU-native synthesis and multicore atmosphere iteration, and validate it against the original Fortran programs. A 300--1000 nm solar spectrum sampled at $R_{\rm grid}=300{,}000$ takes about 14 s on an NVIDIA H100 GPU, while the APOGEE 1500--1700 nm interval takes about 1 s. Physical atmosphere iterations take 2--5 s on 16 AMD CPU threads, and learned initializers reduce the iterations required for convergence. Final spectra remain in practical parity across the tested dwarf and giant regimes. These speeds place direct synthesis inside an optimizer without a label-to-flux spectral emulator. We demonstrate direct many-element fitting of reduced APOGEE spectra and recover multi-element abundance trends broadly consistent with the survey catalog. GPU-resident velocity shifts, broadening, line-spread-function convolution, and detector sampling add negligible cost relative to synthesis. The direct-synthesis search takes less than one minute per star on an H100, while atmosphere verification runs independently on multicore CPUs. The same computational graph calibrates more than $10^5$ oscillator-strength and damping corrections jointly against the Sun and Arcturus in about one minute on an H100. Payne Zero therefore brings direct physical fitting and atomic-data calibration to survey scale. The code is available at https://github.com/tingyuansen/payne-zero.

Figures

Figures reproduced from arXiv: 2607.24141 by Elliot M. Kim, Yuan-Sen Ting.

Figure 1
Figure 1. Figure 1: Agreement with the original Kurucz calculation over 300–1000 nm at Rgrid = 300,000. Rows show a hot dwarf, the Sun, a red giant, and a K dwarf. Columns show the full normalized spectrum, a representative feature, and ∆fnorm = f Payne Zero norm − f Kurucz norm . Black and orange compare the complete calculations. In the residual panels, blue holds the atmosphere fixed to isolate synthesis, while orange incl… view at source ↗
Figure 2
Figure 2. Figure 2: Synthesis runtime per spectrum at Rgrid = 300,000. The left panel compares 300–1000 nm calculations for four stellar types with the original serial Kurucz programs and Payne Zero on one NVIDIA V100, A100, or H100 GPU. The upper middle and right panels decompose the GPU wall for the Sun and K dwarf. The lower panel gives the total solar wall over 1500–1700 nm. Atomic and molecular line opacity dominate the … view at source ↗
Figure 3
Figure 3. Figure 3: Synthesis scaling with wavelength span at Rgrid = 300,000. All windows begin at 300 nm and end at 400, 500, 700, or 1000 nm. Colored curves show the four stellar types on one H100. The dashed curve and shaded band give the median and range of the original Kurucz times for the same stars. The shallower H100 scaling increases the speedup as the spectral interval and retained transition count grow. total and … view at source ↗
Figure 4
Figure 4. Figure 4: Atmosphere performance on an AMD EPYC 9B45 CPU. The left panel compares the mean wall per physical iteration from matched three-iteration runs for four stellar types. Payne Zero uses 16 threads. The middle panel shows its iteration wall from 1 to 32 threads. The right panel decomposes the red-giant iteration into physical stages for original Kurucz and 16-thread Payne Zero. The latter is 6–7 times faster a… view at source ↗
Figure 5
Figure 5. Figure 5: Accuracy and iteration cost of the five- and eight-label atmosphere initializers. The left panel compares the K = 160 PCA compression floor (dashed, open) with the complete held-out label-to-atmosphere error (solid, filled). Points are median standardized depth RMS values, with 16th–84th-percentile ranges. The middle panel gives a conservative spectral error bound after three physical-solver passes for six… view at source ↗
Figure 6
Figure 6. Figure 6: Physical atmosphere structure and learned initialization. Columns vary Teff , log g, [M/H], [α/M], or microturbulence as encoded by their color scales. Rows show six transformed depth profiles on a common Rosseland optical-depth grid. Dashed curves are converged physical atmospheres, and solid curves are the eight-label predictions used to initialize the physical solver. rors are 1.5 × 10−2 and 1.1 × 10−2 … view at source ↗
Figure 7
Figure 7. Figure 7: Controlled five-label fit to a normalized mock spectrum with S/N = 100 and known truth. The upper panels trace accepted fast-search candidates (circles), the optional correction informed by the first converged atmosphere (diamond), and the final solution from a converged atmosphere and synthesis (square) in the Teff –log g and [M/H]–[α/M] planes. Color gives cumulative model time, and the hollow star marks… view at source ↗
Figure 8
Figure 8. Figure 8: Full-band calibration of the solar and Arcturus FTS atlases in six contiguous wavelength strips. For each strip, the upper panel shows normalized flux and the lower panel shows atlas minus model. Black is the atlas, grey dashed uses the original line data, and orange dashed uses the independent four-parameter line fit for that standard. Residual panels share a fixed range within each star. The Arcturus col… view at source ↗
Figure 9
Figure 9. Figure 9: Absolute normalized-flux residual distributions for the Sun and Arcturus. At each horizontal value, the vertical axis is the fraction of retained atlas pixels with a larger absolute atlas–model residual. Grey dashed uses the original line data, blue dash-dotted fits oscillator strength alone, orange solid fits all four line parameters independently for each standard, and magenta dotted uses the shared Sun–… view at source ↗
Figure 10
Figure 10. Figure 10 [PITH_FULL_IMAGE:figures/full_fig_p020_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Representative APOGEE DR14 fit after the converged atmosphere calculation. Black is the observed normalized spectrum, orange dashed is the profiled Payne Zero model after velocity shift, broadening, line-spread function convolution, and pixel sampling, and grey bands show the adopted 1σ uncertainty. The upper spectrum and residual panels cover all three detectors. The three lower pairs expand one line-ric… view at source ↗
Figure 12
Figure 12. Figure 12: Chemical structure of the balanced 1,600-target APOGEE sample. The first two panels show [Mg/Fe]– [Fe/H] from the calibrated DR14 catalog and the corresponding Payne Zero fits. Orange and blue mark the APOGEE high- and low-α samples. The third panel compares Payne Zero and APOGEE DR14 iron abundances. The dashed line denotes equality. Contours enclose 50, 80, and 95 percent of the inverse-selection-weight… view at source ↗
Figure 13
Figure 13. Figure 13: Elemental-abundance distributions from the balanced 1,600-target Payne Zero fit. Each panel shows one [X/Fe] ratio against [Fe/H]. The element is named above the panel. Orange and blue points and contours denote stars drawn from the APOGEE high- and low-α samples, respectively. Colored contours enclose 50, 80, and 95 percent of the inverse-selection￾weighted fitted density. Grey dashed contours show the c… view at source ↗
Figure 14
Figure 14. Figure 14: Sensitivity to fitting Teff , log g, and ξ together with the 12 abundance coordinates. Left, inverse-selection-weighted contours for the combined sample. Gray dashed curves retain the DR14 stellar parameters and dark solid curves show the final physically converged fit. Right, fixed-parameter iron abundances are gray open circles and the final values are colored by the APOGEE high- and low-α samples. −1.0… view at source ↗
Figure 15
Figure 15. Figure 15: Elemental-abundance distributions from the final physically converged fit of Teff , log g, ξ, and the abundances. The layout and scales match [PITH_FULL_IMAGE:figures/full_fig_p029_15.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

85 extracted references · 6 canonical work pages

  1. [1]

    S., Aguilar, G., et al

    Abolfathi, B., Aguado, D. S., Aguilar, G., et al. 2018, ApJS, 235, 42, doi: 10.3847/1538-4365/aa9e8a

    doi:10.3847/1538-4365/aa9e8a 2018

  2. [2]

    2023, FERRE: Match physical models to measurements, Astrophysics Source Code Library, record ascl:2301.016

    Allende-Prieto, C., & Apogee Team. 2023, FERRE: Match physical models to measurements, Astrophysics Source Code Library, record ascl:2301.016. http://ascl.net/2301.016

    2023

  3. [3]

    M., Nissen, P

    Amarsi, A. M., Nissen, P. E., & Sk´ ulad´ ottir,´A. 2019, A&A, 630, A104, doi: 10.1051/0004-6361/201936265

    doi:10.1051/0004-6361/201936265 2019

  4. [4]

    2005, ARA&A, 43, 481, doi: 10.1146/annurev.astro.42.053102.134001

    Asplund, M. 2005, ARA&A, 43, 481, doi: 10.1146/annurev.astro.42.053102.134001

    arXiv 2005

  5. [5]

    S., & Collet, R

    Barklem, P. S., & Collet, R. 2016, A&A, 588, A96, doi: 10.1051/0004-6361/201526961

    doi:10.1051/0004-6361/201526961 2016

  6. [6]

    S., Piskunov, N., & O’Mara, B

    Barklem, P. S., Piskunov, N., & O’Mara, B. J. 2000, A&AS, 142, 467, doi: 10.1051/aas:2000167

    doi:10.1051/aas:2000167 2000

  7. [7]

    2014, in Determination of Atmospheric Parameters of B-, A-, F- and G-Type Stars, ed

    Bergemann, M., & Nordlander, T. 2014, in Determination of Atmospheric Parameters of B-, A-, F- and G-Type Stars, ed. E. Niemczura, B. Smalley, & W. Pych, 169–185, doi: 10.1007/978-3-319-06956-2 16

    doi:10.1007/978-3-319-06956-2 2014

  8. [8]

    2014, A&A, 569, A111, doi: 10.1051/0004-6361/201423945 B¨ ohm-Vitense, E

    Blanco-Cuaresma, S., Soubiran, C., Heiter, U., & Jofr´ e, P. 2014, A&A, 569, A111, doi: 10.1051/0004-6361/201423945 B¨ ohm-Vitense, E. 1958, ZA, 46, 108

    doi:10.1051/0004-6361/201423945 2014

  9. [9]

    A., Conroy, C., Johnson, B

    Cargile, P. A., Conroy, C., Johnson, B. D., et al. 2020, ApJ, 900, 28, doi: 10.3847/1538-4357/aba43b

    doi:10.3847/1538-4357/aba43b 2020

  10. [10]

    R., Hogg, D

    Casey, A. R., Hogg, D. W., Ness, M., et al. 2016, arXiv e-prints, arXiv:1603.03040, doi: 10.48550/arXiv.1603.03040

    Pith/arXiv arXiv doi:10.48550/arxiv.1603.03040 2016

  11. [11]

    Castelli, F., & Kurucz, R. L. 2003, in IAU Symposium, Vol. 210, Modelling of Stellar Atmospheres, ed. N. Piskunov, W. W. Weiss, & D. F. Gray, A20, doi: 10.48550/arXiv.astro-ph/0405087

    Pith/arXiv arXiv doi:10.48550/arxiv.astro-ph/0405087 2003

  12. [12]

    2010, A&A, 522, A10, doi: 10.1051/0004-6361/201014432

    Charbonnel, C., & Lagarde, N. 2010, A&A, 522, A10, doi: 10.1051/0004-6361/201014432

    doi:10.1051/0004-6361/201014432 2010

  13. [13]

    2019, The Messenger, 175, 30, doi: 10.18727/0722-6691/5122

    Chiappini, C., Minchev, I., Starkenburg, E., et al. 2019, The Messenger, 175, 30, doi: 10.18727/0722-6691/5122

    doi:10.18727/0722-6691/5122 2019

  14. [14]

    P., Koposov, S

    Cooper, A. P., Koposov, S. E., Allende Prieto, C., et al. 2023, ApJ, 947, 37, doi: 10.3847/1538-4357/acb3c0

    doi:10.3847/1538-4357/acb3c0 2023

  15. [15]

    L., Sordo, R., Pailler, F., et al

    Creevey, O. L., Sordo, R., Pailler, F., et al. 2023, A&A, 674, A26, doi: 10.1051/0004-6361/202243688

    doi:10.1051/0004-6361/202243688 2023

  16. [16]

    2012, Research in Astronomy and Astrophysics, 12, 1197, doi: 10.1088/1674-4527/12/9/003

    Cui, X.-Q., Zhao, Y.-H., Chu, Y.-Q., et al. 2012, Research in Astronomy and Astrophysics, 12, 1197, doi: 10.1088/1674-4527/12/9/003

    doi:10.1088/1674-4527/12/9/003 2012

  17. [17]

    Green, G. M. 2015, ApJ, 812, 128, doi: 10.1088/0004-637X/812/2/128

    doi:10.1088/0004-637x/812/2/128 2015

  18. [18]

    C., Abrams, D

    Dalton, G., Trager, S. C., Abrams, D. C., et al. 2012, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 8446, Ground-based and Airborne Instrumentation for Astronomy IV, ed. I. S

    2012

  19. [19]

    McLean, S. K. Ramsay, & H. Takami, 84460P, doi: 10.1117/12.925950 De Silva, G. M., Freeman, K. C., Bland-Hawthorn, J., et al. 2015, MNRAS, 449, 2604, doi: 10.1093/mnras/stv327

    doi:10.1117/12.925950 2015

  20. [20]

    A., O’Briain, T., et al

    Fabbro, S., Venn, K. A., O’Briain, T., et al. 2018, MNRAS, 475, 2978, doi: 10.1093/mnras/stx3298

    doi:10.1093/mnras/stx3298 2018

  21. [21]

    1961, Physical Review, 124, 1866, doi: 10.1103/PhysRev.124.1866 Garc ´ ıa P´ erez, A

    Fano, U. 1961, Physical Review, 124, 1866, doi: 10.1103/PhysRev.124.1866 Garc ´ ıa P´ erez, A. E., Allende Prieto, C., Holtzman, J. A., et al. 2016, AJ, 151, 144, doi: 10.3847/0004-6256/151/6/144

    doi:10.1103/physrev.124.1866 1961

  22. [22]

    2012, The Messenger, 147, 25

    Gilmore, G., Randich, S., Asplund, M., et al. 2012, The Messenger, 147, 25

    2012

  23. [23]

    2003, Inverse Problems, 19, R1, doi: 10.1088/0266-5611/19/2/201

    Golub, G., & Pereyra, V. 2003, Inverse Problems, 19, R1, doi: 10.1088/0266-5611/19/2/201

    doi:10.1088/0266-5611/19/2/201 2003

  24. [24]

    Gray, D. F. 2005, The Observation and Analysis of Stellar Photospheres, doi: 10.1017/CBO9781316036570 Gunes Baydin, A., Pearlmutter, B. A., Andreyevich Radul, A., & Siskind, J. M. 2015, arXiv e-prints, arXiv:1502.05767, doi: 10.48550/arXiv.1502.05767

    Pith/arXiv arXiv doi:10.48550/arxiv.1502.05767 2005

  25. [25]

    2008, A&A, 486, 951, doi: 10.1051/0004-6361:200809724

    Gustafsson, B., Edvardsson, B., Eriksson, K., et al. 2008, A&A, 486, 951, doi: 10.1051/0004-6361:200809724

    doi:10.1051/0004-6361:200809724 2008

  26. [26]

    R., Bovy, J., Holtzman, J

    Hayden, M. R., Bovy, J., Holtzman, J. A., et al. 2015, ApJ, 808, 132, doi: 10.1088/0004-637X/808/2/132

    doi:10.1088/0004-637x/808/2/132 2015

  27. [27]

    2015, A&A, 582, A49, doi: 10.1051/0004-6361/201526319

    Heiter, U., Jofr´ e, P., Gustafsson, B., et al. 2015, A&A, 582, A49, doi: 10.1051/0004-6361/201526319

    doi:10.1051/0004-6361/201526319 2015

  28. [28]

    2021, A&A, 645, A106, doi: 10.1051/0004-6361/201936291

    Heiter, U., Lind, K., Bergemann, M., et al. 2021, A&A, 645, A106, doi: 10.1051/0004-6361/201936291

    doi:10.1051/0004-6361/201936291 2021

  29. [29]

    1995, PASP, 107, 1042, doi: 10.1086/133660

    Hinkle, K., Wallace, L., & Livingston, W. 1995, PASP, 107, 1042, doi: 10.1086/133660

    doi:10.1086/133660 1995

  30. [30]

    Ho, A. Y. Q., Rix, H.-W., Ness, M. K., et al. 2017, ApJ, 841, 40, doi: 10.3847/1538-4357/aa6db3

    doi:10.3847/1538-4357/aa6db3 2017

  31. [31]

    A., Hasselquist, S., Shetrone, M., et al

    Holtzman, J. A., Hasselquist, S., Shetrone, M., et al. 2018, AJ, 156, 125, doi: 10.3847/1538-3881/aad4f9

    doi:10.3847/1538-3881/aad4f9 2018

  32. [32]

    1995, ApJ, 439, 875, doi: 10.1086/175226

    Hubeny, I., & Lanz, T. 1995, ApJ, 439, 875, doi: 10.1086/175226

    doi:10.1086/175226 1995

  33. [33]

    2014, Theory of Stellar Atmospheres

    Hubeny, I., & Mihalas, D. 2014, Theory of Stellar Atmospheres

    2014

  34. [34]

    G., & Mihalas, D

    Hummer, D. G., & Mihalas, D. 1988, ApJ, 331, 794, doi: 10.1086/166600

    doi:10.1086/166600 1988

  35. [35]

    2013, A&A, 553, A6, doi: 10.1051/0004-6361/201219058 Jofr´ e, P., Heiter, U., Soubiran, C., et al

    Husser, T.-O., Wende-von Berg, S., Dreizler, S., et al. 2013, A&A, 553, A6, doi: 10.1051/0004-6361/201219058 Jofr´ e, P., Heiter, U., Soubiran, C., et al. 2014, A&A, 564, A133, doi: 10.1051/0004-6361/201322440 Stellar Spectra from Physical Models in Seconds31 —. 2015, A&A, 582, A81, doi: 10.1051/0004-6361/201526604

    doi:10.1051/0004-6361/201219058 2013

  36. [36]

    John, T. L. 1988, A&A, 193, 189

    1988

  37. [37]

    R., & Krupp, B

    Johnson, H. R., & Krupp, B. M. 1976, ApJ, 206, 201, doi: 10.1086/154371

    work page doi:10.1086/154371 1976

  38. [38]

    T., & Cadima, J

    Jolliffe, I. T., & Cadima, J. 2016, Philosophical Transactions of the Royal Society of London Series A, 374, 20150202, doi: 10.1098/rsta.2015.0202 J¨ onsson, H., Allende Prieto, C., Holtzman, J. A., et al. 2018, AJ, 156, 126, doi: 10.3847/1538-3881/aad4f5

    arXiv 2016

  39. [39]

    2022, ApJS, 258, 31, doi: 10.3847/1538-4365/ac3b4d

    Kawahara, H., Kawashima, Y., Masuda, K., et al. 2022, ApJS, 258, 31, doi: 10.3847/1538-4365/ac3b4d

    doi:10.3847/1538-4365/ac3b4d 2022

  40. [40]

    M., & Ting, Y.-S

    Kim, E. M., & Ting, Y.-S. 2026, arXiv e-prints, arXiv:2603.11693, doi: 10.48550/arXiv.2603.11693

    Pith/arXiv arXiv doi:10.48550/arxiv.2603.11693 2026

  41. [41]

    A., Zasowski, G., Rix, H.-W., et al

    Kollmeier, J. A., Zasowski, G., Rix, H.-W., et al. 2017, arXiv e-prints, arXiv:1711.03234, doi: 10.48550/arXiv.1711.03234

    Pith/arXiv arXiv doi:10.48550/arxiv.1711.03234 2017

  42. [42]

    Kurucz, R. L. 1996, in IAU Symposium, Vol. 176, Stellar Surface Structure, ed. K. G. Strassmeier & J. L. Linsky, 523

    1996

  43. [43]

    Kurucz, R. L. 2005, Memorie della Societa Astronomica Italiana Supplementi, 8, 14 —. 2013, ATLAS12: Opacity sampling model atmosphere program, Astrophysics Source Code Library, record ascl:1303.024. http://ascl.net/1303.024

    2005

  44. [44]

    L., & Avrett, E

    Kurucz, R. L., & Avrett, E. H. 1981, SAO Special Report, 391

    1981

  45. [45]

    2003, ApJS, 146, 417, doi: 10.1086/374373 —

    Lanz, T., & Hubeny, I. 2003, ApJS, 146, 417, doi: 10.1086/374373 —. 2007, ApJS, 169, 83, doi: 10.1086/511270

    doi:10.1086/374373 2003

  46. [46]

    2019, A&A, 624, A60, doi: 10.1051/0004-6361/201833553

    Laverick, M., Lobel, A., Royer, P., et al. 2019, A&A, 624, A60, doi: 10.1051/0004-6361/201833553

    work page doi:10.1051/0004-6361/201833553 2019

  47. [47]

    W., & Bovy, J

    Leung, H. W., & Bovy, J. 2019, MNRAS, 483, 3255, doi: 10.1093/mnras/sty3217

    doi:10.1093/mnras/sty3217 2019

  48. [48]

    Li, J., Jian, M., Ting, Y.-S., & Green, G. M. 2025, arXiv e-prints, arXiv:2507.06357, doi: 10.48550/arXiv.2507.06357

    Pith/arXiv arXiv doi:10.48550/arxiv.2507.06357 2025

  49. [49]

    1991, An atlas of the solar spectrum in the infrared from 1850 to 9000 cm-1 (1.1 to 5.4 micrometer)

    Livingston, W., & Wallace, L. 1991, An atlas of the solar spectrum in the infrared from 1850 to 9000 cm-1 (1.1 to 5.4 micrometer)

    1991

  50. [50]

    2011, Canadian Journal of Physics, 89, 395, doi: 10.1139/p11-030

    Lobel, A. 2011, Canadian Journal of Physics, 89, 395, doi: 10.1139/p11-030

    work page doi:10.1139/p11-030 2011

  51. [51]

    2013, A&A, 557, A26, doi: 10.1051/0004-6361/201321274

    Magic, Z., Collet, R., Asplund, M., et al. 2013, A&A, 557, A26, doi: 10.1051/0004-6361/201321274

    doi:10.1051/0004-6361/201321274 2013

  52. [52]

    R., Schiavon, R

    Majewski, S. R., Schiavon, R. P., Frinchaboy, P. M., et al. 2017, AJ, 154, 94, doi: 10.3847/1538-3881/aa784d

    doi:10.3847/1538-3881/aa784d 2017

  53. [53]

    2016, MNRAS, 456, 3655, doi: 10.1093/mnras/stv2830

    Martig, M., Fouesneau, M., Rix, H.-W., et al. 2016, MNRAS, 456, 3655, doi: 10.1093/mnras/stv2830

    doi:10.1093/mnras/stv2830 2016

  54. [54]

    2015, MNRAS, 453, 1855, doi: 10.1093/mnras/stv1731

    Masseron, T., & Gilmore, G. 2015, MNRAS, 453, 1855, doi: 10.1093/mnras/stv1731

    doi:10.1093/mnras/stv1731 2015

  55. [55]

    M., Carlos, M., & Nissen, P

    Matsuno, T., Amarsi, A. M., Carlos, M., & Nissen, P. E. 2024, A&A, 688, A72, doi: 10.1051/0004-6361/202450057

    doi:10.1051/0004-6361/202450057 2024

  56. [56]

    D., Beckman, R

    McKay, M. D., Beckman, R. J., & Conover, W. J. 1979, Technometrics, 21, 239, doi: 10.1080/00401706.1979.10489755 Mel´ endez, J., & Barbuy, B. 1999, ApJS, 124, 527, doi: 10.1086/313261 M´ esz´ aros, S., & Allende Prieto, C. 2013, MNRAS, 430, 3285, doi: 10.1093/mnras/stt130 M´ esz´ aros, S., Allende Prieto, C., Edvardsson, B., et al. 2012, AJ, 144, 120, doi...

    arXiv 1979

  57. [57]

    1978, Stellar atmospheres

    Mihalas, D. 1978, Stellar atmospheres

    1978

  58. [58]

    2015, ApJ, 808, 16, doi: 10.1088/0004-637X/808/1/16

    Zasowski, G. 2015, ApJ, 808, 16, doi: 10.1088/0004-637X/808/1/16

    doi:10.1088/0004-637x/808/1/16 2015

  59. [59]

    L., Bovy, J., Bird, J

    Nidever, D. L., Bovy, J., Bird, J. C., et al. 2014, ApJ, 796, 38, doi: 10.1088/0004-637X/796/1/38

    doi:10.1088/0004-637x/796/1/38 2014

  60. [60]

    L., Holtzman, J

    Nidever, D. L., Holtzman, J. A., Allende Prieto, C., et al. 2015, AJ, 150, 173, doi: 10.1088/0004-6256/150/6/173

    doi:10.1088/0004-6256/150/6/173 2015

  61. [61]

    Nocedal, J., & Wright, S. J. 2006, Numerical Optimization, 2nd edn., Springer Series in Operations Research and Financial Engineering (Springer New York), doi: 10.1007/978-0-387-40065-5 Nordlund, ˚A., Stein, R. F., & Asplund, M. 2009, Living Reviews in Solar Physics, 6, 2, doi: 10.12942/lrsp-2009-2 O’Briain, T., Ting, Y.-S., Fabbro, S., et al. 2021, ApJ, ...

    doi:10.1007/978-0-387-40065-5 2006

  62. [62]

    Piskunov, N., & Valenti, J. A. 2017, A&A, 597, A16, doi: 10.1051/0004-6361/201629124

    doi:10.1051/0004-6361/201629124 2017

  63. [63]

    E., Kupka, F., Ryabchikova, T

    Piskunov, N. E., Kupka, F., Ryabchikova, T. A., Weiss, W. W., & Jeffery, C. S. 1995, A&AS, 112, 525

    1995

  64. [64]

    2012, Turbospectrum: Code for spectral synthesis, Astrophysics Source Code Library, record ascl:1205.004

    Plez, B. 2012, Turbospectrum: Code for spectral synthesis, Astrophysics Source Code Library, record ascl:1205.004. http://ascl.net/1205.004 Ram ´ ırez, I., & Allende Prieto, C. 2011, ApJ, 743, 135, doi: 10.1088/0004-637X/743/2/135 R´ o˙ za´ nski, T., Ting, Y.-S., & Jab lo´ nska, M. 2025, ApJ, 980, 66, doi: 10.3847/1538-4357/ad9b99

    doi:10.1088/0004-637x/743/2/135 2012

  65. [65]

    L., et al

    Ryabchikova, T., Piskunov, N., Kurucz, R. L., et al. 2015, PhyS, 90, 054005, doi: 10.1088/0031-8949/90/5/054005

    doi:10.1088/0031-8949/90/5/054005 2015

  66. [66]

    M., & Cassisi, S

    Salaris, M., Pietrinferni, A., Piersimoni, A. M., & Cassisi, S. 2015, A&A, 583, A87, doi: 10.1051/0004-6361/201526951

    doi:10.1051/0004-6361/201526951 2015

  67. [67]

    E., et al

    Shetrone, M., Bizyaev, D., Lawler, J. E., et al. 2015, ApJS, 221, 24, doi: 10.1088/0067-0049/221/2/24

    work page doi:10.1088/0067-0049/221/2/24 2015

  68. [68]

    Shore, B. W. 1967, Reviews of Modern Physics, 39, 439, doi: 10.1103/RevModPhys.39.439

    work page doi:10.1103/revmodphys.39.439 1967

  69. [69]

    V., Bizyaev, D., Cunha, K., et al

    Smith, V. V., Bizyaev, D., Cunha, K., et al. 2021, AJ, 161, 254, doi: 10.3847/1538-3881/abefdc 32Ting & Kim

    doi:10.3847/1538-3881/abefdc 2021

  70. [70]

    2012, MOOG: LTE line analysis and spectrum synthesis, Astrophysics Source Code Library, record ascl:1202.009

    Sneden, C., Bean, J., Ivans, I., Lucatello, S., & Sobeck, J. 2012, MOOG: LTE line analysis and spectrum synthesis, Astrophysics Source Code Library, record ascl:1202.009. http://ascl.net/1202.009

    2012

  71. [71]

    S., Kraft, R

    Sobeck, J. S., Kraft, R. P., Sneden, C., et al. 2011, AJ, 141, 175, doi: 10.1088/0004-6256/141/6/175

    doi:10.1088/0004-6256/141/6/175 2011

  72. [72]

    S., Chiba, M., et al

    Takada, M., Ellis, R. S., Chiba, M., et al. 2014, PASJ, 66, R1, doi: 10.1093/pasj/pst019

    doi:10.1093/pasj/pst019 2014

  73. [73]

    2016, ApJ, 826, 83, doi: 10.3847/0004-637X/826/1/83

    Ting, Y.-S., Conroy, C., & Rix, H.-W. 2016, ApJ, 826, 83, doi: 10.3847/0004-637X/826/1/83

    work page doi:10.3847/0004-637x/826/1/83 2016

  74. [74]

    2018, ApJ, 860, 159, doi: 10.3847/1538-4357/aac6c9

    Ting, Y.-S., Conroy, C., Rix, H.-W., & Asplund, M. 2018, ApJ, 860, 159, doi: 10.3847/1538-4357/aac6c9

    doi:10.3847/1538-4357/aac6c9 2018

  75. [75]

    2017a, ApJ, 843, 32, doi: 10.3847/1538-4357/aa7688 —

    Ting, Y.-S., Conroy, C., Rix, H.-W., & Cargile, P. 2017a, ApJ, 843, 32, doi: 10.3847/1538-4357/aa7688 —. 2019, ApJ, 879, 69, doi: 10.3847/1538-4357/ab2331

    doi:10.3847/1538-4357/aa7688 2019

  76. [76]

    Ting, Y.-S., Rix, H.-W., Conroy, C., Ho, A. Y. Q., & Lin, J. 2017b, ApJL, 849, L9, doi: 10.3847/2041-8213/aa921c

    doi:10.3847/2041-8213/aa921c 2041

  77. [77]

    A., & Fischer, D

    Valenti, J. A., & Fischer, D. A. 2005, ApJS, 159, 141, doi: 10.1086/430500

    doi:10.1086/430500 2005

  78. [78]

    A., & Piskunov, N

    Valenti, J. A., & Piskunov, N. 1996, A&AS, 118, 595

    1996

  79. [79]

    R., Cooper, J., & Smith, E

    Vidal, C. R., Cooper, J., & Smith, E. W. 1973, ApJS, 25, 37, doi: 10.1086/190264

    doi:10.1086/190264 1973

  80. [80]

    2023, A&A, 671, A171, doi: 10.1051/0004-6361/202244482

    Wehrhahn, A., Piskunov, N., & Ryabchikova, T. 2023, A&A, 671, A171, doi: 10.1051/0004-6361/202244482

    doi:10.1051/0004-6361/202244482 2023

Showing first 80 references.