pith. sign in

arxiv: 2606.29552 · v1 · pith:WD3CHV7Unew · submitted 2026-06-28 · 🌌 astro-ph.EP · astro-ph.IM· astro-ph.SR

Improving the Precision of Line-by-Line Radial Velocities: A Data-Driven Iterative Algorithm for Spectral Line Selection

Kanishk Pandey , Eric A. Ford , Joseph M. Salzer , Jessi Cisewski-Kehe , Lily L. Zhao , Arvind F. Gupta , Michael L. Palumbo III , Ryan C. Terrien
show 12 more authors
Evan Fitzmaurice Brady Barth Jason T. Wright Cullen H. Blake Suvrath Mahadevan Arpita Roy Joe P. Ninan Gudmundur Stefansson Paul Robertson Christian Schwab Daniel M. Krolikowski Leonardo A. Paredes
This is my paper

Pith reviewed 2026-06-30 01:59 UTC · model grok-4.3

classification 🌌 astro-ph.EP astro-ph.IMastro-ph.SR
keywords radial velocityspectral line selectionexoplanet detectionline-by-line analysissolar observationsprecision radial velocitiesiterative algorithmNEID
0
0 comments X

The pith

An iterative algorithm selects 24 spectral lines to reach 1.122 m/s radial velocity precision on solar data.

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

The paper introduces FLARES as a data-driven iterative method for choosing which spectral lines to use when measuring radial velocities one line at a time. Starting from thousands of lines observed over hundreds of days, the algorithm keeps only those with extreme values in several properties and removes the ones that most increase scatter when dropped. The result is a short list that produces lower scatter than the complete set or random subsets of the same size. A reader would care because radial velocity precision directly limits the smallest planets that can be found, and fewer, better lines may reduce the effect of stellar activity on the measurements.

Core claim

FLARES selects candidate spectral lines with extreme values of multiple line metrics such as depth, signal-to-noise ratio, and detector position, then iteratively rejects lines whose removal produces the largest decrease in weighted RV scatter. On 383 days of NEID solar observations this yields an RV RMS of 1.122 m/s with only 24 lines, which is lower than the 2.012 m/s obtained from the full line-by-line list and also lower than several benchmark selection methods. Monte Carlo tests confirm the selections are reproducible, and comparisons with alternative lists matched on similar properties show that FLARES is identifying the combination of line traits that actually drive the improvement.

What carries the argument

FLARES, an iterative line-selection algorithm that ranks and prunes lines using multiple metrics including depth, SNR, and position to minimize weighted RV scatter.

If this is right

  • Subsets chosen by line metrics such as depth and intrinsic RV scatter produce lower RV RMS than the full line list or random subsets of equal size.
  • FLARES reaches 1.122 m/s RMS with 24 lines and outperforms the benchmark selection methods tested.
  • Monte Carlo simulations show the FLARES selections are robust and reproducible.
  • Alternative lists matched on similar properties to the FLARES lines do not achieve the same RV performance, confirming the algorithm identifies effective combinations of line traits.

Where Pith is reading between the lines

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

  • If the solar-derived metrics transfer, FLARES could be run once on calibration data and then used on science targets without per-star adjustment.
  • Focusing on a small number of lines may lower the computational cost of line-by-line analysis and reduce the influence of instrument artifacts that affect only certain detector regions.
  • The same selection logic could be tested on data from other spectrographs to see whether the retained lines remain optimal across instruments.

Load-bearing premise

Line metrics and rejection rules derived from solar observations will identify similarly high-performing lines when applied to other stars without star-specific re-tuning.

What would settle it

Apply the FLARES line list derived from solar data to a different star and check whether its RV RMS remains lower than that of the full line list or random subsets of equal size on the same observations.

Figures

Figures reproduced from arXiv: 2606.29552 by Arpita Roy, Arvind F. Gupta, Brady Barth, Christian Schwab, Cullen H. Blake, Daniel M. Krolikowski, Eric A. Ford, Evan Fitzmaurice, Gudmundur Stefansson, Jason T. Wright, Jessi Cisewski-Kehe, Joe P. Ninan, Joseph M. Salzer, Kanishk Pandey, Leonardo A. Paredes, Lily L. Zhao, Michael L. Palumbo III, Paul Robertson, Ryan C. Terrien, Suvrath Mahadevan.

Figure 1
Figure 1. Figure 1: The filtered (red) and remaining (black) lines on the detector after removing outlier lines that deviate from the weighted time average, described in Section 2.5 and fur￾ther in Appendix F. The gray shade shows an approximate projection of the NEID FSR onto the extracted orders. All lines redward of physical ´echelle order 85 are removed which is likely due to telluric contamination, and a significant frac… view at source ↗
Figure 2
Figure 2. Figure 2: Evolution of RV RMS as a function of the number of spectral lines retained for Random (dashed gray; see Section 3.2), BLARES (solid dark-red; see Section 3.3), FLARESdepth (solid pink; see Section 3.5), FLARES (solid dark-blue; see Section 4), and FLARES MC NN (solid dark-green; see Section 5.5). Also shown is the expected photon-noise limit (dashed red; see Section 3.1). Shaded regions for Random, FLARES,… view at source ↗
Figure 3
Figure 3. Figure 3: The dependence of RV properties on spectral line depth. All 2, 809 lines are sorted by depth and divided into 10 bins such that each bin contains ≈ 281 lines (see Section 3.4). For comparison, each panel shows the median (dashed black) and the 1σ (dark gray) and 2σ (light gray) dispersion across 1,000 bootstrap simulations of randomly generated line lists with the same size as each bin to illustrate the ex… view at source ↗
Figure 4
Figure 4. Figure 4: The evolution of RV properties as the shallowest lines are iteratively removed using FLARESdepth (see Sec￾tion 3.5). For comparison, each panel includes the 1σ (gray) and 2σ (light gray) dispersion across 10,000 bootstrap sim￾ulations of randomly generated line lists, with sizes drawn from a log-uniform prior spanning 10 to 2,809 lines, to illus￾trate the expected behavior of typical random clusters. Top: … view at source ↗
Figure 5
Figure 5. Figure 5: The dependence of RV RMS on various spectral line metrics described in Section 4.1. For each metric, all 2, 809 lines are divided into 10 equal bins, such that each bin contains approximately 281 lines. For comparison, each panel includes the 1σ (dark gray) and 2σ (light gray) confidence intervals derived from 1,000 randomly generated line lists with the same size as each bin to illustrate the expected beh… view at source ↗
Figure 6
Figure 6. Figure 6: Evolution of the RV RMS as the worst lines are iteratively removed, based on a single run of the FLARES algorithm described in Section 4. Top: the RV RMS as a function of the number of lines remaining. Points are colored by the metric category used for the corresponding removal step. The metric categories are the same as those defined in Section 4.1 and are activity (pink), atomic (orange), damping (green)… view at source ↗
Figure 7
Figure 7. Figure 7: Locations of all lines on the detector, colored by the step they were removed by FLARES. Lines colored a darker red are removed later in FLARES, suggesting these lines yield RVs less affected by stellar and instrumental vari￾ability. The gray shade shows an approximate projection of the NEID FSR onto the extracted physical ´echelle orders. Lines near the edge of the orders are removed in early itera￾tions … view at source ↗
Figure 9
Figure 9. Figure 9: Evolution of the RV RMS as the worst lines are iteratively removed when running FLARES on only the metrics within each metric category (see Section 5.2). Black points show the evolution of the algorithm using all metrics (same as the top panel of [PITH_FULL_IMAGE:figures/full_fig_p017_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Left: the distributions of the depth for all lines (black), for the best 1000 lines (brown), and the best 100 lines (teal) selected by FLARES. Middle: same as the left panel, but for the distribution of the pixels of each line. Right: same as the left panel, but for the distribution of the SNR of each line. The 100 best lines selected by FLARES tend to be deeper, located near the center of the detector, a… view at source ↗
Figure 11
Figure 11. Figure 11: The robustness of FLARES across 100 MC sim￾ulations (FLARES MC; see Section 5.4). Top: the mean RV RMS of FLARES MC as a function of the number of lines remaining. The dark and light blue shaded regions indicate the 1σ and 2σ dispersion across FLARES MC, respectively. The small dispersion in RV RMS even with only 20 − 30 lines remaining suggests that FLARES is robust across mul￾tiple MC simulations. Botto… view at source ↗
Figure 12
Figure 12. Figure 12: Distributions of the RV RMS of various methods in this work, evaluated for 25 lines. FLARES MC NN (dark-green; see Section 5.5) shows the distribution of RV RMS for 100 line lists chosen to have lines that have similar properties to the best lines ranked by FLARES MC. Also shown is the photon-noise limit (dashed red; see Section 3.1), the distribution of RV RMS for Random (gray; see Section 3.2), the RV R… view at source ↗
Figure 13
Figure 13. Figure 13: Comparison of the RV RMS across NEID physical ´echelle orders for DRP v1.3 and v1.4. Top: the RV RMS for every physical ´echelle order for v1.3 data (teal circles) and v1.4 data (orange stars). The RV RMS was calculated using the PSRP and a Gaussian fit on CCFs across the best days where v1.3 and v1.4 data were both available. Physical ´echelle orders 113−118 and 155 have a much lower RV RMS in v1.4 data … view at source ↗
Figure 14
Figure 14. Figure 14: The distribution of the mean pyrheliometer flux (left), mean exposure meter flux (middle), and airmass (right) for all exposures. The distributions are colored blue, yellow, and green for runs 1, 2, and 3, respectively. Vertical colored lines indicate the threshold for each run, and the black line indicates the airmass threshold for the right panel. We then impose a minimum flux (averaged over the observa… view at source ↗
Figure 15
Figure 15. Figure 15: Segment of the daily-averaged NEID solar spectrum from 2021-02-06 (black points) overlaid with spectral lines from the line list (red vertical lines) along with their corresponding 15-pixel window used in this work (orange shade). We note that this window may not be the best window for every line and may be susceptible to blends for narrow lines. An alternative approach could be to explore windows that va… view at source ↗
Figure 16
Figure 16. Figure 16: The definitions of line-fitting parameters used throughout this work. A synthetic observed line (solid blue) is shown, and the corresponding Gaussian parameters that describe the observed line are shown as dashed lines and include the continuum level (a, green), the continuum slope (b, red), the central wavelength (λc, purple), the depth (d, brown), and the width (σ, pink). E. GAUSSIAN MODEL FOR LINE FITT… view at source ↗
Figure 17
Figure 17. Figure 17: The procedure to remove lines that significantly deviate from the behavior of all lines and cannot be explained by photon noise, based on Section 2.5. Top left: the initial heatmap of RVs for every line across time. Vertical solid black lines in each heatmap indicate boundaries between observing runs (see [PITH_FULL_IMAGE:figures/full_fig_p026_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Comparison of RVs derived from the CCF method based on Ford et al. (2024) and the LBL method in this work. Top: the RV time series for the CCF (teal circles) and LBL (orange stars) methods. The RV RMS of the LBL approach is slightly less (≈ 2.012 m s−1 ) than the CCF approach (≈ 2.129 m s−1 ), and also shows a lower periodicity at the solar rotation period and at 365 days. Bottom: the RV residuals between… view at source ↗
Figure 19
Figure 19. Figure 19: The CCF vs LBL RVs from runs 1 (brown circles), run 2 (gray stars), and run 3 (black X markers). The CCF and LBL RVs show a strong correlation of ρ = 0.99. Solid colored lines depict a best-fit linear slope to CCF vs LBL RVs within each run, and the red dashed line shows unity. ≈ −0.08, which is similar to ACF365 ≈ −0.116 for the CCF RVs. The reduced power of ACF30 and ACF365 for the LBL method as compare… view at source ↗
read the original abstract

Independent analysis of individual spectral lines, or line-by-line (LBL) analyses, can improve upon standard cross-correlation function (CCF) methods for measuring radial velocities (RVs) because they preserve critical information about individual line shape changes that can be caused by stellar activity. In this work, we measure LBL RVs of 3,830 spectral lines across 383 days of NEID solar observations. Our LBL approach achieves an RV RMS of $2.012~\mathrm{m\,s^{-1}}$, which is slightly lower than the $2.129~\mathrm{m\,s^{-1}}$ achieved by a CCF approach using a shared line list. Then, we describe and benchmark several methods for selecting line lists based on line properties such as depth and intrinsic RV scatter. We find that these subsets have a lower RV RMS compared to either the full line list or random subsets of equal size. Motivated by these results, we present FLARES (Filtering Lines for Accurate Radial-velocity Exoplanet Search), an iterative line-selection algorithm. FLARES selects candidate spectral lines with extreme values of multiple line metrics and properties such as depth, signal-to-noise ratio, and detector position, and preferentially rejects lines whose removal produces the largest decrease in the weighted RV scatter. FLARES achieves an RV RMS of $1.122~\mathrm{m\,s^{-1}}$ using just 24 lines and performs better than the benchmark methods. We perform Monte Carlo simulations and show FLARES is robust and reproducible. Comparisons to alternative line lists chosen to have properties similar to the best FLARES-selected lines demonstrate that FLARES is successfully identifying line properties that lead to effective line lists for future extreme-precision RV measurements.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript introduces FLARES, an iterative data-driven algorithm for selecting optimal spectral lines in line-by-line (LBL) radial velocity analysis. On 383 days of NEID solar observations measuring 3830 lines, the full LBL approach yields 2.012 m/s RMS (vs. 2.129 m/s for CCF), while FLARES selects 24 lines to reach 1.122 m/s RMS, outperforming random and property-matched subsets; Monte Carlo simulations are cited to show robustness and reproducibility.

Significance. If the selected line properties generalize, FLARES could meaningfully advance extreme-precision RV work by providing a reproducible method to minimize scatter from activity or instrumental effects. The concrete RMS values, direct comparisons to benchmarks, and use of a large homogeneous solar time series are strengths for internal validation on solar data.

major comments (3)
  1. [Abstract] Abstract: the claim that FLARES produces line lists 'suitable for future extreme-precision RV measurements' rests on an untested assumption that solar-derived metrics (depth, SNR, detector position, and iterative scatter thresholds) will identify high-performing lines on other stars; no independent stellar dataset or non-solar test is described.
  2. [FLARES algorithm] FLARES algorithm description: the target line count (24) and extreme-value thresholds for candidate selection are treated as fixed inputs without a sensitivity study or pre-specification, leaving open whether they were adjusted after inspecting the final RMS on the same 383-day series.
  3. [Evaluation] Evaluation and Monte Carlo section: the iterative rejection step uses weighted RV scatter computed from the identical time series that supplies the reported 1.122 m/s RMS; although Monte Carlo reproducibility checks are mentioned, no explicit cross-validation split or held-out solar subset is detailed to separate selection from evaluation.
minor comments (2)
  1. [Methods] Clarify whether the CCF comparison uses the identical line list or a standard mask, and report the exact weighting scheme applied to the RV scatter in the iterative step.
  2. [Abstract] The abstract states 'we perform Monte Carlo simulations and show FLARES is robust'; add a brief description of the simulation design (e.g., number of trials, what is randomized) to the main text.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and constructive review. We address each major comment below, indicating where revisions will be made to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that FLARES produces line lists 'suitable for future extreme-precision RV measurements' rests on an untested assumption that solar-derived metrics (depth, SNR, detector position, and iterative scatter thresholds) will identify high-performing lines on other stars; no independent stellar dataset or non-solar test is described.

    Authors: Solar data from NEID provides the highest-SNR, most homogeneous time series available for isolating line-selection effects without stellar activity contamination. The metrics used (depth, SNR, detector position, scatter) are physical properties expected to generalize across targets observed with the same spectrograph. We agree, however, that the abstract claim is stronger than the solar-only validation supports. We will revise the abstract to describe the FLARES lines as 'promising candidates for' extreme-precision RV measurements and add a sentence noting that extension to stellar targets is planned future work. revision: partial

  2. Referee: [FLARES algorithm] FLARES algorithm description: the target line count (24) and extreme-value thresholds for candidate selection are treated as fixed inputs without a sensitivity study or pre-specification, leaving open whether they were adjusted after inspecting the final RMS on the same 383-day series.

    Authors: The target count of 24 was chosen to yield a compact list comparable in size to typical CCF line masks while still achieving substantial RMS reduction; the extreme-value thresholds were set to isolate clear outliers in the four line metrics. These choices were made prior to the final reported RMS. To remove any ambiguity, we will add a sensitivity analysis in the revised manuscript that varies both the target count (e.g., 10–50 lines) and the percentile thresholds, showing that the performance gain remains robust in the vicinity of the adopted values. revision: yes

  3. Referee: [Evaluation] Evaluation and Monte Carlo section: the iterative rejection step uses weighted RV scatter computed from the identical time series that supplies the reported 1.122 m/s RMS; although Monte Carlo reproducibility checks are mentioned, no explicit cross-validation split or held-out solar subset is detailed to separate selection from evaluation.

    Authors: The Monte Carlo procedure already draws random temporal subsets to test selection stability, but we acknowledge that it does not constitute a formal train/test split that fully decouples selection from evaluation. We will revise the Evaluation section to include an explicit k-fold cross-validation: line selection is performed on training folds only, the resulting line list is applied to the held-out fold, and the average out-of-sample RMS is reported. This will be added alongside the existing Monte Carlo results. revision: yes

Circularity Check

1 steps flagged

FLARES reported RMS is the direct output of iterative optimization on the identical solar time series used for evaluation

specific steps
  1. fitted input called prediction [Abstract]
    "FLARES selects candidate spectral lines with extreme values of multiple line metrics and properties such as depth, signal-to-noise ratio, and detector position, and preferentially rejects lines whose removal produces the largest decrease in the weighted RV scatter. FLARES achieves an RV RMS of 1.122 m s^{-1} using just 24 lines and performs better than the benchmark methods."

    The rejection step directly optimizes the weighted RV scatter computed from the 383-day solar series; the quoted 1.122 m/s RMS is the value of that same scatter after the optimization has been performed on the identical observations. No held-out partition or external stellar dataset is used for the final number.

full rationale

The paper derives line metrics and the iterative rejection rule from the 383-day NEID solar observations, then applies the rejection criterion (largest decrease in weighted RV scatter) to produce a 24-line list whose scatter is reported as 1.122 m/s. Because the selection explicitly minimizes the same scatter metric on the same dataset, the headline performance number is the fitted result rather than an independent test. Monte Carlo runs address only internal reproducibility within that sample. No self-citation chains, definitional loops, or imported uniqueness theorems appear; the circularity is confined to the fitted-input-called-prediction pattern in the evaluation step.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claim rests on empirical correlations between observable line properties and RV scatter measured on solar data, plus the assumption that iterative removal based on scatter reduction identifies generally useful lines.

free parameters (2)
  • target line count (24)
    The number of lines retained is the outcome of the iterative process that minimizes weighted RV scatter.
  • extreme-value thresholds for candidate selection
    Thresholds on depth, SNR, and detector position used to pre-filter candidate lines before iteration.
axioms (2)
  • domain assumption Line depth, SNR, and detector position are predictive of a line's contribution to RV scatter.
    Used to define the pool of candidate lines that FLARES considers.
  • domain assumption Weighted RV scatter is a reliable scalar metric for line quality.
    The quantity minimized during iterative rejection.

pith-pipeline@v0.9.1-grok · 5954 in / 1285 out tokens · 27329 ms · 2026-06-30T01:59:02.353420+00:00 · methodology

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

64 extracted references · 54 canonical work pages · 2 internal anchors

  1. [1]

    Valenti, J. A. 2022, A&A, 664, A34, doi: 10.1051/0004-6361/202243276 Al Moulla, K., Dumusque, X., Figueira, P., et al. 2023, A&A, 669, A39, doi: 10.1051/0004-6361/202244663 Anglada-Escud´ e, G., & Tuomi, M. 2012, A&A, 548, A58, doi: 10.1051/0004-6361/201219910

    work page doi:10.1051/0004-6361/202243276 2022

  2. [2]

    1996, A&AS, 119, 373

    Baranne, A., Queloz, D., Mayor, M., et al. 1996, A&AS, 119, 373

    1996

  3. [3]

    2014, A&A, 564, A46, doi: 10.1051/0004-6361/201322383

    Bodichon, R. 2014, A&A, 564, A46, doi: 10.1051/0004-6361/201322383

    work page doi:10.1051/0004-6361/201322383 2014

  4. [4]

    2011, A&A, 528, A4, doi: 10.1051/0004-6361/201014354

    Boisse, I., Bouchy, F., H´ ebrard, G., et al. 2011, A&A, 528, A4, doi: 10.1051/0004-6361/201014354

    work page doi:10.1051/0004-6361/201014354 2011

  5. [5]

    2001, A&A, 374, 733, doi: 10.1051/0004-6361:20010730

    Bouchy, F., Pepe, F., & Queloz, D. 2001, A&A, 374, 733, doi: 10.1051/0004-6361:20010730

    work page doi:10.1051/0004-6361:20010730 2001

  6. [6]

    A., Dumusque, X., & Halverson, S

    Burt, J. A., Dumusque, X., & Halverson, S. 2025, arXiv e-prints, arXiv:2511.01954, doi: 10.48550/arXiv.2511.01954

    work page doi:10.48550/arxiv.2511.01954 2025

  7. [7]

    P., Marcy, G

    Butler, R. P., Marcy, G. W., Williams, E., et al. 1996, PASP, 108, 500, doi: 10.1086/133755

    work page doi:10.1086/133755 1996

  8. [8]

    1985, A&A, 150, 256

    Cavallini, F., Ceppatelli, G., & Righini, A. 1985, A&A, 150, 256

    1985

  9. [9]

    and McElroy, Douglas L

    Christiansen, J. L., McElroy, D. L., Harbut, M., et al. 2025, PSJ, 6, 186, doi: 10.3847/PSJ/ade3c2 Collier Cameron, A., Mortier, A., Phillips, D., et al. 2019, MNRAS, 487, 1082, doi: 10.1093/mnras/stz1215

    work page doi:10.3847/psj/ade3c2 2025

  10. [10]

    Ground-based and Airborne Instrumentation for Astronomy IV , year = 2012, editor =

    Cosentino, R., Lovis, C., Pepe, F., 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. McLean, S. K. Ramsay, & H. Takami, 84461V, doi: 10.1117/12.925738

    work page doi:10.1117/12.925738 2012

  11. [11]

    S., Leifer, S., et al

    Crass, J., Gaudi, B. S., Leifer, S., et al. 2021, arXiv e-prints, arXiv:2107.14291, doi: 10.48550/arXiv.2107.14291

    work page doi:10.48550/arxiv.2107.14291 2021

  12. [12]

    2020, A&A, 633, A76, doi: 10.1051/0004-6361/201936548

    Lovis, C. 2020, A&A, 633, A76, doi: 10.1051/0004-6361/201936548

    work page doi:10.1051/0004-6361/201936548 2020

  13. [13]

    C., & Pepe, F

    Cretignier, M., Dumusque, X., Hara, N. C., & Pepe, F. 2021, A&A, 653, A43, doi: 10.1051/0004-6361/202140986

    work page doi:10.1051/0004-6361/202140986 2021

  14. [14]

    1982, ARA&A, 20, 61, doi: 10.1146/annurev.aa.20.090182.000425

    Dravins, D. 1982, ARA&A, 20, 61, doi: 10.1146/annurev.aa.20.090182.000425

    work page doi:10.1146/annurev.aa.20.090182.000425 1982

  15. [15]

    1981, A&A, 96, 345

    Dravins, D., Lindegren, L., & Nordlund, A. 1981, A&A, 96, 345

    1981

  16. [16]

    A&A , author =

    Dumusque, X. 2018, A&A, 620, A47, doi: 10.1051/0004-6361/201833795 30K. Pandey, et al

    work page doi:10.1051/0004-6361/201833795 2018

  17. [17]

    Dumusque, X., Boisse, I., & Santos, N. C. 2014, ApJ, 796, 132, doi: 10.1088/0004-637X/796/2/132

    work page doi:10.1088/0004-637x/796/2/132 2014

  18. [18]

    2011a, A&A, 527, A82, doi: 10.1051/0004-6361/201015877

    Bonfils, X. 2011a, A&A, 527, A82, doi: 10.1051/0004-6361/201015877

    work page doi:10.1051/0004-6361/201015877

  19. [19]

    Monteiro, M. J. P. F. G. 2011b, A&A, 525, A140, doi: 10.1051/0004-6361/201014097

    work page doi:10.1051/0004-6361/201014097

  20. [20]

    2025, arXiv e-prints, arXiv:2510.27635, doi: 10.48550/arXiv.2510.27635

    Dumusque, X., Al Moulla, K., Cretignier, M., et al. 2025, arXiv e-prints, arXiv:2510.27635, doi: 10.48550/arXiv.2510.27635

    work page doi:10.48550/arxiv.2510.27635 2025

  21. [21]

    A., & Krolik, J

    Edelson, R. A., & Krolik, J. H. 1988, ApJ, 333, 646, doi: 10.1086/166773

    work page doi:10.1086/166773 1988

  22. [22]

    B., Bender, C

    Ford, E. B., Bender, C. F., Blake, C. H., et al. 2024, arXiv e-prints, arXiv:2408.13318, doi: 10.48550/arXiv.2408.13318

    work page doi:10.48550/arxiv.2408.13318 2024

  23. [23]

    R., Howard, A

    Gibson, S. R., Howard, A. W., Rider, K., et al. 2024, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 13096, Ground-based and Airborne Instrumentation for Astronomy X, ed. J. J

    2024

  24. [24]

    Motohara, & J

    Bryant, K. Motohara, & J. R. D. Vernet, 1309609, doi: 10.1117/12.3017841

    work page doi:10.1117/12.3017841

  25. [25]

    B., Halverson, S., et al

    Gilbertson, C., Ford, E. B., Halverson, S., et al. 2024, arXiv e-prints, arXiv:2408.17289, doi: 10.48550/arXiv.2408.17289

    work page doi:10.48550/arxiv.2408.17289 2024

  26. [26]

    Gray, D. F. 2005, The Observation and Analysis of Stellar Photospheres, doi: 10.1017/CBO9781316036570

    work page doi:10.1017/cbo9781316036570 2005

  27. [27]

    2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol

    Halverson, S., Terrien, R., Mahadevan, S., et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9908, Ground-based and Airborne Instrumentation for Astronomy VI, ed. C. J

    2016

  28. [28]

    Simard, & H

    Evans, L. Simard, & H. Takami, 99086P, doi: 10.1117/12.2232761

    work page doi:10.1117/12.2232761

  29. [29]

    D., Collier Cameron, A., Unruh, Y

    Haywood, R. D., Collier Cameron, A., Unruh, Y. C., et al. 2016, MNRAS, 457, 3637, doi: 10.1093/mnras/stw187

    work page doi:10.1093/mnras/stw187 2016

  30. [30]

    2020, arXiv e-prints, arXiv:2005.14083, doi: 10.48550/arXiv.2005.14083

    Holzer, P., Cisewski-Kehe, J., Fischer, D., & Zhao, L. 2020, arXiv e-prints, arXiv:2005.14083, doi: 10.48550/arXiv.2005.14083

    work page doi:10.48550/arxiv.2005.14083 2020

  31. [31]

    2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol

    Jurgenson, C., Fischer, D., McCracken, T., et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9908, Ground-based and Airborne Instrumentation for Astronomy VI, ed. C. J

    2016

  32. [32]

    and Fischer, D

    Evans, L. Simard, & H. Takami, 99086T, doi: 10.1117/12.2233002

    work page doi:10.1117/12.2233002

  33. [33]

    Kjeldsen, H., & Bedding, T. R. 1995, A&A, 293, 87, doi: 10.48550/arXiv.astro-ph/9403015

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.astro-ph/9403015 1995

  34. [34]

    R., & Christensen-Dalsgaard, J

    Kjeldsen, H., Bedding, T. R., & Christensen-Dalsgaard, J. 2008, ApJL, 683, L175, doi: 10.1086/591667

    work page doi:10.1086/591667 2008

  35. [35]

    2021, A&A, 652, A28, doi: 10.1051/0004-6361/202140605

    Lafarga, M., Ribas, I., Reiners, A., et al. 2021, A&A, 652, A28, doi: 10.1051/0004-6361/202140605

    work page doi:10.1051/0004-6361/202140605 2021

  36. [36]

    2010, A&A, 512, A38, doi: 10.1051/0004-6361/200913071

    Lagrange, A.-M., Desort, M., & Meunier, N. 2010, A&A, 512, A38, doi: 10.1051/0004-6361/200913071

    work page doi:10.1051/0004-6361/200913071 2010

  37. [37]

    2011, A&A, 528, L9, doi: 10.1051/0004-6361/201016354

    Lagrange, A.-M., Meunier, N., Desort, M., & Malbet, F. 2011, A&A, 528, L9, doi: 10.1051/0004-6361/201016354

    work page doi:10.1051/0004-6361/201016354 2011

  38. [38]

    S., Naylor, T., Haywood, R

    Lakeland, B. S., Naylor, T., Haywood, R. D., et al. 2024, MNRAS, 527, 7681, doi: 10.1093/mnras/stad3723

    work page doi:10.1093/mnras/stad3723 2024

  39. [39]

    F., Molaro, P., Monaco, L., & Haywood, R

    Lanza, A. F., Molaro, P., Monaco, L., & Haywood, R. D. 2016, A&A, 587, A103, doi: 10.1051/0004-6361/201527379

    work page doi:10.1051/0004-6361/201527379 2016

  40. [40]

    Lin, A. S. J., Monson, A., Mahadevan, S., et al. 2022, AJ, 163, 184, doi: 10.3847/1538-3881/ac5622

    work page doi:10.3847/1538-3881/ac5622 2022

  41. [41]

    The HARPS search for southern extra-solar planets. XXXI. Magnetic activity cycles in solar-type stars: statistics and impact on precise radial velocities

    Lovis, C., Dumusque, X., Santos, N. C., et al. 2011, arXiv e-prints, arXiv:1107.5325, doi: 10.48550/arXiv.1107.5325

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1107.5325 2011

  42. [42]

    K., Wright, J

    Luhn, J. K., Wright, J. T., Howard, A. W., & Isaacson, H. 2020, AJ, 159, 235, doi: 10.3847/1538-3881/ab855a

    work page doi:10.3847/1538-3881/ab855a 2020

  43. [43]

    , year = 1995, month = nov, volume =

    Mayor, M., & Queloz, D. 1995, Nature, 378, 355, doi: 10.1038/378355a0

    work page doi:10.1038/378355a0 1995

  44. [44]

    S., Moore, T

    McMillan, R. S., Moore, T. L., Perry, M. L., & Smith, P. H. 1993, ApJ, 403, 801, doi: 10.1086/172251

    work page doi:10.1086/172251 1993

  45. [45]

    2010, A&A, 512, A39, doi: 10.1051/0004-6361/200913551

    Meunier, N., Desort, M., & Lagrange, A.-M. 2010, A&A, 512, A39, doi: 10.1051/0004-6361/200913551

    work page doi:10.1051/0004-6361/200913551 2010

  46. [46]

    2015, A&A, 583, A118, doi: 10.1051/0004-6361/201525721

    Meunier, N., Lagrange, A.-M., Borgniet, S., & Rieutord, M. 2015, A&A, 583, A118, doi: 10.1051/0004-6361/201525721

    work page doi:10.1051/0004-6361/201525721 2015

  47. [47]

    2017, A&A, 607, A124, doi: 10.1051/0004-6361/201731017 O’Sullivan, N

    Meunier, N., Mignon, L., & Lagrange, A.-M. 2017, A&A, 607, A124, doi: 10.1051/0004-6361/201731017 O’Sullivan, N. K., Aigrain, S., Cretignier, M., et al. 2025, MNRAS, 541, 3942, doi: 10.1093/mnras/staf1168

    work page doi:10.1051/0004-6361/201731017 2017

  48. [48]

    L., Ford, E

    Palumbo, M. L., Ford, E. B., Gonzalez, E. B., et al. 2024, AJ, 168, 46, doi: 10.3847/1538-3881/ad4c6d

    work page doi:10.3847/1538-3881/ad4c6d 2024

  49. [49]

    2026, Supplementary Material for Improving the Precision of Line-by-Line Radial Velocities: A Data-Driven Iterative Algorithm for Spectral Line

    Pandey, K. 2026, Supplementary Material for Improving the Precision of Line-by-Line Radial Velocities: A Data-Driven Iterative Algorithm for Spectral Line

    2026

  50. [50]

    Selection, Zenodo, doi: 10.5281/zenodo.20937741

    work page doi:10.5281/zenodo.20937741

  51. [51]

    A&A , author =

    Pepe, F., Mayor, M., Galland, F., et al. 2002, A&A, 388, 632, doi: 10.1051/0004-6361:20020433

    work page doi:10.1051/0004-6361:20020433 2002

  52. [52]

    2004, A&A, 423, 385, doi: 10.1051/0004-6361:20040389

    Pepe, F., Mayor, M., Queloz, D., et al. 2004, A&A, 423, 385, doi: 10.1051/0004-6361:20040389

    work page doi:10.1051/0004-6361:20040389 2004

  53. [53]

    2013, The Messenger, 153, 6

    Pepe, F., Cristiani, S., Rebolo, R., et al. 2013, The Messenger, 153, 6

    2013

  54. [54]

    L., et al

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

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

  55. [55]

    Saar, S. H. 2003, in Astronomical Society of the Pacific Conference Series, Vol. 294, Scientific Frontiers in Research on Extrasolar Planets, ed. D. Deming & S. Seager, 65–70

    2003

  56. [56]

    H., & Donahue, R

    Saar, S. H., & Donahue, R. A. 1997, ApJ, 485, 319, doi: 10.1086/304392

    work page doi:10.1086/304392 1997

  57. [57]

    B., & Zhao, L

    Salzer, J., Cisewski-Kehe, J., Ford, E. B., & Zhao, L. L. 2025, AJ, 170, 179, doi: 10.3847/1538-3881/adf29d A Data-Driven Iterative Algorithm for Spectral Line Selection31

    work page doi:10.3847/1538-3881/adf29d 2025

  58. [58]

    2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol

    Schwab, C., Rakich, A., Gong, Q., et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9908, Ground-based and Airborne Instrumentation for Astronomy VI, ed. C. J. Evans, L. Simard, & H. Takami, 99087H, doi: 10.1117/12.2234411

    work page doi:10.1117/12.2234411 2016

  59. [59]

    L., & Schwab, C

    Seifahrt, A., St¨ urmer, J., Bean, J. L., & Schwab, C. 2018, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 10702, Ground-based and Airborne Instrumentation for Astronomy VII, ed. C. J

    2018

  60. [60]

    Simard, & H

    Evans, L. Simard, & H. Takami, 107026D, doi: 10.1117/12.2312936

    work page doi:10.1117/12.2312936

  61. [61]

    N., & Fabrycky, D

    Winn, J. N., & Fabrycky, D. C. 2015, ARA&A, 53, 409, doi: 10.1146/annurev-astro-082214-122246

    work page doi:10.1146/annurev-astro-082214-122246 2015

  62. [62]

    T., & Kanodia, S

    Wright, J. T., & Kanodia, S. 2020, PSJ, 1, 38, doi: 10.3847/PSJ/ababa4

    work page doi:10.3847/psj/ababa4 2020

  63. [63]

    A&A , author =

    Zechmeister, M., Reiners, A., Amado, P. J., et al. 2018, A&A, 609, A12, doi: 10.1051/0004-6361/201731483

    work page doi:10.1051/0004-6361/201731483 2018

  64. [64]

    L., Dumusque, X., Ford, E

    Zhao, L. L., Dumusque, X., Ford, E. B., et al. 2023, AJ, 166, 173, doi: 10.3847/1538-3881/acf83e

    work page doi:10.3847/1538-3881/acf83e 2023