Pith. sign in

REVIEW 4 major objections 5 minor 33 references

Twinkle: A GPU-based binary-lens microlensing code with contour integration method

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Twinkle is a GPU-based binary-lens microlensing code that removes catastrophic cancellation in the lens-equation coefficients and claims correct magnifications over $0.5 < s < 3$, $q > 10^{-6}$ at about 100x single-thread CPU speed.

desk verdict Useful engineering: a GPU contour-integration code with a legitimate coefficient refactorization and a clever ghost detector, but the headline robustness claim leans on self-consistent error estimates rather than independent ground truth. read the letter →

arxiv 2501.03322 v1 pith:GHGHNJHJ submitted 2025-01-06 astro-ph.IM astro-ph.EPastro-ph.GAastro-ph.SR

Suwei Wang , Lile Wang , Subo Dong This is my paper · ORCID
classification astro-ph.IMastro-ph.EPastro-ph.GAastro-ph.SR
keywords gravitationalmicrolensingbinarylenscontourintegrationGPUcomputingfinite-sourceeffectscatastrophiccancellationlow-mass-ratioplanetsopen-sourcesoftware
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Twinkle is a contour-integration code for computing the magnification of a two-body gravitational lens made of a host star and a planet. The paper's central claim is that existing contour-integration implementations lose precision when the planet/star mass ratio is small, because evaluating the coefficients of the fifth-degree lens polynomial subtracts large numbers that nearly cancel; this can produce wrong image contours, spikes in light curves, and failed fits exactly in the low-mass-ratio regime relevant to Earth-mass planets. Twinkle refactorizes those coefficients so quantities of order $q$ are computed directly, adds a two-part detector that distinguishes real images from ghost images (spurious polynomial roots) using a zero-sum residual identity and a perpendicular-slope jump near caustics, and maps the computation onto GPUs with one thread per point source and one block per source. The paper reports that a single consumer GPU runs about 100 times faster than a single-threaded CPU run of the standard reference code, while remaining correct throughout $0.5 < s < 3$ and $q > 10^{-6}$. A CPU-only version retains the numerical improvements, so the fixes are not tied to the GPU implementation.

What carries the argument

The load-bearing objects are the refactorized polynomial coefficients of the binary-lens equation, the residual zero-sum identity, and the slope detector. The coefficient refactorization introduces $v_c=\bar{y}+s$ and $v_p=1+s\bar{y}$ and rewrites the fifth-degree coefficients so that no term of order $q$ is produced by subtracting two much larger nearly equal numbers; this is what keeps image positions accurate when the source grazes a planetary caustic. The residual detector uses the theorem that the five residuals of the polynomial equation always sum to zero, so any pair of ghost images must have equal residual magnitudes, giving a numerically stable criterion to separate real from ghost images. The slope detector uses the fact that near the critical curve a pair of real images turns into a pair of ghost images with a $\pi/2$ jump in the direction of their separation, so adjacent image pairs that fail the perpendicularity test are reclassified or deleted. Adaptive sampling then redistributes contour points according to the estimated local error, and the GPU implementation assigns one thread to each point source and one block to each extended source, using streams and shared-memory reductions so that many sources are computed concurrently.

What would settle it

Take the paper's hard case of $s=0.5$, $q=10^{-6}$, with a source boundary grazing a planetary caustic cusp, and compute the magnification with an independent method such as dense inverse ray-shooting or quadruple-precision evaluation of the contour integral. If the independent value differs from Twinkle by more than the stated $10^{-4}$ relative tolerance while Twinkle reports an error below tolerance, the robustness claim is falsified.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the failures of binary-lens contour integration at small mass ratios are not inherent to the method but come from how the lens polynomial's coefficients are evaluated. The conventional coefficient expressions contain subtractions of nearly equal large terms whose true values are powers of $q$, so floating-point arithmetic destroys the precision needed when the source boundary sits near a planetary caustic with $q = 10^{-6}$. Twinkle rewrites the coefficients using intermediate variables so each term is formed at its true order; it also proves a zero-sum identity for the five residuals of the polynomial, so when two ghost images exist their residual magnitudes must be equal and can be identified reliably, and it adds a slope detector that exploits the perpendicular approach of real and ghost image pairs near the critical curve to repair misclassified image pairs. The paper then claims that Twinkle correctly and efficiently computes all magnifications in the parameter range $0.5 < s < 3$ and $q > 10^{-6}$, extends reliable calculation to $q \ge 10^{-9}$ for $0.1 \le s \le 10$ at a relative tolerance of $10^{-4}$, and on a single consumer GPU is about 100 times faster than a single-threaded CPU implementation of the standard algorithm.

Load-bearing premise

The low-mass-ratio accuracy claim depends on trusting Twinkle's internal error estimates and on using Twinkle itself as the reference for judging the comparison code, rather than on an independent ground-truth calculation.

Editorial extensions

If this is right

  • Microlensing fits for low-mass-ratio events will no longer be derailed by the spikes and wrong image connections that occur when a source boundary passes near a planetary caustic.
  • A single consumer GPU can replace a small CPU cluster for contour-integration modeling, opening wider parameter-space searches and faster Markov-chain exploration for current surveys.
  • The refactorized coefficients and ghost-image detector can be ported into any code that solves the binary-lens polynomial, improving accuracy even without GPU hardware.
  • The CPU-only version of Twinkle keeps the same numerical robustness while matching the speed of the standard code, so the low-$q$ abnormal region is removed on both architectures.
  • For the planned space-based microlensing survey whose simulated low-$q$ detections lie in the region where the standard code slows or fails, Twinkle removes a known computational bottleneck.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The zero-sum residual identity is a cheap, self-contained consistency check that could also be used in ray-shooting codes to flag unreliable polynomial solves, since it requires no external reference.
  • Because each extended source is an independent block, Twinkle's design should show nearly linear weak scaling across multiple GPUs or nodes; the paper reports only single-GPU scaling.
  • The coefficient-refactorization strategy could be applied to higher-order lens configurations with more than two masses, where cancellation problems are likely to be worse, though the paper does not test this.
  • A direct head-to-head with quadruple-precision arithmetic or dense ray-shooting near cusps would turn the parameter-space robustness claim into a calibration rather than an internal consistency check.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper presents Twinkle, a GPU-accelerated contour-integration code for binary-lens microlensing magnification. The main contributions are a refactorization of the binary-lens polynomial coefficients to avoid catastrophic cancellation at low mass ratios, a two-part ghost-image detector based on a zero-residual-sum theorem and a slope criterion near critical curves, and a GPU-oriented implementation with adaptive sampling and stream-based task dispatch. The authors report a speedup of roughly two orders of magnitude over single-threaded VBBL on an RTX 4090 and claim reliable magnification over the parameter space 0.5 < s < 3 and q > 1e-6, with additional tests claimed for 0.1 <= s <= 10 and q >= 1e-9. The accuracy validation is based primarily on comparison with VBBL, and in the low-q regime the VBBL error rate is defined relative to Twinkle itself.

Significance. If the claims are correct, Twinkle would be a valuable tool for binary-lens microlensing modeling: the coefficient refactorization is a simple, portable algebraic improvement; the residual-sum proof in Eq. (14) is elegant and checkable; and the open-source GPU and CPU releases are useful community assets. The demonstrated speed advantage and robustness in the low-mass-ratio regime are directly relevant to Roman and ongoing ground-based microlensing programs. However, the central correctness claim currently rests on self-consistent error estimates and on comparisons in which VBBL's failures are defined as disagreement with Twinkle. Independent validation is needed before the robustness claim can be accepted.

major comments (4)
  1. [§4.3, Figure 7 and §4.2, Figure 6] The central robustness claim — that Twinkle 'can correctly and efficiently solve all magnifications within the parameter space [0.5 < s < 3 and q > 1e-6]' — is not supported by an independent accuracy test. In the bottom right panel of Figure 7, the VBBL error rate is defined as the proportion of points where VBBL disagrees with Twinkle, and the white region means agreement with Twinkle; this measures consistency, not correctness. In Figure 6, the VBBL spikes show that VBBL and Twinkle differ at q = 1e-6, but no independent value demonstrates that Twinkle's smooth curve is the true magnification. If both codes fail differently in the abnormal region (q ≲ 1e-5, s ∼ 0.6), the conclusion would be false. Please add an independent reference for the low-q regime, e.g., ray-shooting with a high ray density, arbitrary-precision evaluation of the contour integrals, or an analytic test case.
  2. [§2.5, Eqs. (17)–(18)] The error estimate used as the stopping criterion is a heuristic, not a rigorous upper bound. The four error terms E1 through E4 are empirical measures based on parabolic-correction differences and local quantities, and the iteration stops when the estimated error falls below tolerance. If this estimator is biased low near cusps or critical curves, Twinkle can report convergence to an incorrect image area. The correctness claims therefore depend on validating this estimator against an external reference, but no such validation is provided.
  3. [§4.3, last paragraph] The statement that 'we have conducted additional tests demonstrating that our code can reliably calculate magnifications for binary separations in the range 0.1 <= s <= 10 and mass ratios q >= 1e-9 with a relative tolerance of 1e-4' is unsupported. No data, procedure, or acceptance criteria are given for these tests. Since this extends the parameter-space claim well beyond Figures 5–7, the authors should either provide supporting results (e.g., a figure or table analogous to Figure 7) or soften the claim to what is actually shown.
  4. [§4.1, Figure 5] The two VBBL outliers that exceed the tolerance are excluded from the comparison, with the explanation that they are rare and occur at high magnification. These are precisely the points where the reference code fails, and excluding them weakens the agreement test. The authors should either analyze these outliers in detail or replace VBBL with an independent reference at those locations, so that the comparison is not biased toward agreement.
minor comments (5)
  1. [Abstract and Figure 1 caption] There are missing spaces and incorrect cross-references: the abstract has 'andTwinkle' (should be 'and Twinkle'), and the Figure 1 caption refers to 'Equation 4 in §2.1' although Equation (4) appears in §2.2.
  2. [Figure 4] The x-axis label is rendered as '/2' in the caption; please clarify that it is θ/(2π) or specify the normalization explicitly.
  3. [§2.4.1] The criterion for judging that two residual lengths are 'equivalent' is not defined; please provide the exact threshold or statistical test used in the residual detector.
  4. [§4.3] The sentence 'Twinkle can correctly and efficiently solve all magnification within the parameter space' should read 'all magnifications'.
  5. [§2.2, Eq. (6)] The algebraic equivalence between the refactored coefficients in Equation (6) and the original coefficients in Equation (4) is not demonstrated in the text; a short derivation or a high-precision coefficient comparison would aid reproducibility and reader confidence.

Circularity Check

1 steps flagged · score 4.0 of 10

VBBL error rates are defined relative to Twinkle, so the low-q correctness validation is partially self-referential; the speed and coefficient-refactorization results remain independently grounded.

  1. self definitional [Section 4.3, Figure 7 (bottom-right panel), echoed in Section 5]
    "The bottom right panel in Figure 7 shows the error rates of VBBL, defined as the proportion of erroneous points among all sampled points, in which the white region indicates where the magnifications calculated by Twinkle and VBBL are consistent. ... VBBL fails to yield correct magnifications ... Twinkle can correctly and efficiently solve all magnification within the parameter space."

    The panel operationally defines a VBBL 'erroneous point' as one where VBBL disagrees with Twinkle, so the displayed VBBL error rate is, by construction, a Twinkle-VBBL disagreement rate rather than an independent measure of VBBL error. The paper then uses this map, together with Twinkle's own internal error estimator, to conclude that VBBL fails and that Twinkle is correct in the low-q abnormal region. That conclusion assumes Twinkle is the ground truth, which is exactly what is at issue where VBBL is asserted to fail; no independent benchmark (ray-shooting, high-precision arithmetic, or analytic test case) is supplied to break the loop.

full rationale

The paper contains no parameter fitting, no load-bearing self-citation chain, and no imported uniqueness theorem. The coefficient refactorization in Section 2.2 is an explicit algebraic manipulation designed to avoid catastrophic cancellation, with a concrete example in Figure 1; this is checkable and independent. The residual-detector proof in Section 2.4 is self-contained, and the speed claim is supported by direct timing comparisons of Twinkle on one GPU versus single-threaded VBBL on one CPU, so that claim is independently grounded. The partial circularity is confined to the validation of correctness in the regime where VBBL is said to fail: the bottom-right panel of Figure 7 defines 'erroneous' VBBL points as those differing from Twinkle, making the error-rate map a consistency map rather than an absolute-error map. The additional assertion that Twinkle works for 0.1 <= s <= 10 and q >= 1e-9 is stated without a described procedure or shown data, and the internal error estimator in Section 2.5 is a heuristic based on parabolic-correction differences rather than a proven bound. These are genuine correctness risks, but only the VBBL-error definition is a structural circularity, and it affects a validation claim rather than the core algorithmic derivation. On that basis a score of 4 is appropriate: partial circularity in one central support, with substantial independent content remaining.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central robustness claim rests on the reliability of the adopted parabolic error estimators, the local expansion used by the slope detector, and the scoped binary-lens model. No new physical entities are introduced; all free parameters are algorithmic defaults rather than fitted physical constants.

free parameters (3)
  • Relative error tolerance = 1e-4 (default)
    User-set accuracy target used in all performance tests; the robustness claim depends on this tolerance.
  • Number of adaptive sampling quantiles = 64 (default)
    Chosen by hand for GPU thread-block size; affects speed and accuracy.
  • Maximum contour sampling points = 4096 per source
    Determined by available GPU shared memory (64 kB), not by physics; can limit accuracy in extreme cross-caustic cases with tolerance below 1e-6.
assumptions (5)
  • domain assumption The binary-lens equation (Eq. 1) accurately models the lensing geometry of two point masses; no higher-order effects such as parallax, orbital motion, or extended lenses are included.
    Standard model used throughout; the code's claim is scoped to this model.
  • standard math The polynomial transformation from the lens equation to a degree-5 polynomial has exactly the same roots as the original equation in the transformed coordinate system.
    Skowron & Gould 2012 algorithm; necessary for the coefficient refactorization.
  • domain assumption For a uniform-brightness source, magnification equals total image area divided by source area (Stokes theorem), and the parabolic correction terms from Bozza (2010, 2024) provide reliable error estimates.
    Foundation of contour integration; the error estimator is adopted from prior literature and not independently validated in this paper.
  • domain assumption Near critical curves, ghost-image and real-image pairs are locally perpendicular (Eq. 16), justifying the slope detector.
    Based on local expansion of the lens equation; assumed valid for the parameter range tested.
  • standard math The sum of residuals of the five polynomial roots is zero in the center-of-mass frame (Eq. 14), and this remains true under rotation and translation.
    Proved using Vieta's theorem; used by the residual detector.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Twinkle: A GPU-based binary-lens microlensing code with contour integration method." pith.science (2026). https://pith.science/paper/GHGHNJHJ

@misc{pith2026250103322,
  author       = {Pith},
  title        = {Pith review of: Twinkle: A GPU-based binary-lens microlensing code with contour integration method},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GHGHNJHJ}},
  note         = {Machine review of arXiv:2501.03322}
}
read the original abstract

With the rapidly increasing rate of microlensing planet detections, microlensing modeling software faces significant challenges in computation efficiency. Here, we develop the Twinkle code, an efficient and robust binary-lens modeling software suite optimized for heterogeneous computing devices, especially GPUs. Existing microlensing codes have the issue of catastrophic cancellation that undermines the numerical stability and precision, and Twinkle resolves them by refining the coefficients of the binary-lens equation. We also devise an improved method for robustly identifying ghost images, thereby enhancing computational reliability. We have advanced the state of the art by optimizing Twinkle specifically for heterogeneous computing devices by taking into account the unique task and cache memory dispatching patterns of GPUs, while the compatibility with the traditional computing architectures of CPUs is still maintained. Twinkle has demonstrated an acceleration of approximately 2 orders of magnitude (>~100 times) on contemporary GPUs. The enhancement in computational speed of Twinkle will translate to the delivery of accurate and highly efficient data analysis for ongoing and upcoming microlensing projects. Both GPU and CPU versions of Twinkle are open-source and publicly available.

Figures

Figures reproduced from arXiv: 2501.03322 by the authors.

Figure 1
Figure 1. An example demonstrating the enhanced accuracy introduced by the improvements in generating the coefficients of the lens equation (Equation 4 in §2.1). The top panel shows the overall images (green) on the image plane (θx-θy). The bottom left and right panels are the zoom-ins of the region marked by the square in the top panel, showing the results using conventional (bottom left panel; Equation 4) and our improved (… view at source ↗
Figure 2
Figure 2. Visualization of an extended source and the images. The left and right panels show the source and image planes, respectively. The caustic is shown in red, and the critical curve is displayed in black. The two black plus signs mark the positions of two lens objects. The source boundary in the left panel crosses the caustic twice, creating three complete image boundaries (1, 2, and 3 in the right panel) and two broken… view at source ↗
Figure 3
Figure 3. Schematic examples of ghost image pair misjudgment near the critical curve. The magenta dots and the blue crosses are real images with positive and negative parity, while the gray diamonds are ghost images. The black line is the critical curve, on which a pair of real images could transfer into a pair of ghost images. Image pairs marked with the same text come from the same point sources. In the left panel, pair 1 a… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The inverse transform method generates eight new samples. θ is the position angle of the sampling point source. The green line is the cumulative distribution function induced by the area error. The cyan quantiles distribute uni￾formly on the vertical axis, mapping to θ…
Figure 5
Figure 5. Figure 5: Calculation performance of Twinkle compared with VBBL. All the origin points are moved to the position of planetary caustic (or the resonant caustic for s = 1). The left column takes s = 0.5, q = 10−4 , ρ = 5 × 10−5 , ∆θx = (θx + 1.5), ∆θy = (θy − 0.035). The middle co…
Figure 6
Figure 6. Figure 6: An example of Twinkle robustness. The upper left panel shows the magnification calculated by Twinkle and VBBL, and the lower left panel shows the relative difference, with the horizontal pink lines representing tolerance and the horizontal axis representing the coordin…
Figure 7
Figure 7. Figure 7: Calculation performance in parameter space. The upper row shares the same logarithmic color mapping, in which the blank region marks where VBBL takes abnormally long calculation time. The bottom left panel is acceleration defined as the time ratio of Twinkle and single…
Figure 8
Figure 8. Figure 8: Scaling relationship of computations performed by Twinkle. The horizontal axis is the number of sources in tasks, and the vertical axis is the computational time normalized by the linear fitting coefficient. Each source is dispatched to one block. The green data points…
Figure 9
Figure 9. Figure 9: CPU version performance in parameter space. The left panel shows the computational time used by the CPU version of Twinkle, with the same color mapping as the upper row of [PITH_FULL_IMAGE:figures/full_fig_p016_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 19 canonical work pages

  1. [1]

    Bennett, D. P. 2010, ApJ, 716, 1408

    work page 2010

  2. [2]

    A., Abe, F., Dodd, R

    Bond, I. A., Abe, F., Dodd, R. J., et al. 2001, MNRAS, 327, 868

    2001

  3. [3]

    2010, MNRAS, 408, 2188

    Bozza, V. 2010, MNRAS, 408, 2188

    work page 2010

  4. [4]

    2018, MNRAS, 479, 5157

    Bozza, V., Bachelet, E., Bartolić, F., et al. 2018, MNRAS, 479, 5157

    work page 2018

  5. [5]

    VBMicroLensing: three algorithms for multiple lensing with contour integration

    Bozza, V., Saggese, V., Covone, G., Rota, P., & Zhang, J. 2024, arXiv e-prints, arXiv:2410.13660

    work page Pith review arXiv 2024

  6. [6]

    2017, MNRAS, 468, 3993

    Cassan, A. 2017, MNRAS, 468, 3993

    work page 2017

  7. [7]

    1995, A&AS, 109, 597 —

    Dominik, M. 1995, A&AS, 109, 597 —. 1998, A&A, 333, L79 —. 2007, MNRAS, 377, 1679

    work page 1995

  8. [8]

    L., Gaudi, B

    Dong, S., DePoy, D. L., Gaudi, B. S., et al. 2006, ApJ, 642, 842

    2006

Show all 33 references
  1. [9]

    A., Gould, A., et al

    Dong, S., Bond, I. A., Gould, A., et al. 2009, ApJ, 698, 1826

    2009

  2. [10]

    Gaudi, B. S. 2012, ARA&A, 50, 411

    2012

  3. [11]

    1994, ApJ, 421, L75 —

    Gould, A. 1994, ApJ, 421, L75 —. 2008, ApJ, 681, 1593

    1994

  4. [12]

    1997, ApJ, 477, 580

    Gould, A., & Gaucherel, C. 1997, ApJ, 477, 580

    1997

  5. [13]

    1992, ApJ, 396, 104

    Gould, A., & Loeb, A. 1992, ApJ, 396, 104

    1992

  6. [14]

    1986, A&A, 166, 36

    Kayser, R., Refsdal, S., & Stabell, R. 1986, A&A, 166, 36

    1986

  7. [15]

    2016, JKAS, 49, 37

    Kim, S.-L., Lee, C.-U., Park, B.-G., et al. 2016, JKAS, 49, 37

    2016

  8. [16]

    Kuang, R., Mao, S., Wang, T., Zang, W., & Long, R. J. 2021, MNRAS, 503, 6143

    2021

  9. [17]

    1991, ApJ, 374, L37

    Mao, S., & Paczynski, B. 1991, ApJ, 374, L37

    1991

  10. [18]

    W., Thorne, K

    Misner, C. W., Thorne, K. S., & Wheeler, J. A. 1973, Gravitation (Princeton University Press)

    1973

  11. [19]

    2023, arXiv e-prints, arXiv:2310.07502 NASA Exoplanet Archive

    Mroz, P., & Poleski, R. 2023, arXiv e-prints, arXiv:2310.07502 NASA Exoplanet Archive. 2024, Microlensing Planets Table, vVersion: 2024-10-22, NExScI-Caltech/IPAC, doi:10.26133/NEA38. https://catcopy.ipac.caltech.edu/ dois/doi.php?id=10.26133/NEA38

    2023 arXiv

  12. [20]

    J., & Wickramasinghe, W

    Nemiroff, R. J., & Wickramasinghe, W. A. D. T. 1994, ApJ, 424, L21

    1994

  13. [21]

    2009, ApJ, 690, 1772

    Pejcha, O., & Heyrovský, D. 2009, ApJ, 690, 1772

    2009

  14. [22]

    T., Gaudi, B

    Penny, M. T., Gaudi, B. S., Kerins, E., et al. 2019, ApJS, 241, 3 17

    2019

  15. [23]

    Rhie, S. H. 2001, arXiv e-prints, astro

    2001

  16. [24]

    Schneider, P., Ehlers, J., & Falco, E. E. 1992, Gravitational Lenses (Springer Berlin, Heidelberg), doi:10.1007/978-3-662-03758-4

    1992 doi

  17. [25]

    1986, A&A, 164, 237 —

    Schneider, P., & Weiss, A. 1986, A&A, 164, 237 —. 1987, A&A, 171, 49

    1986

  18. [26]

    1987, A&A, 174, 361

    Schramm, T., & Kayser, R. 1987, A&A, 174, 361

    1987

  19. [27]

    2012, arXiv e-prints, arXiv:1203.1034

    Skowron, J., & Gould, A. 2012, arXiv e-prints, arXiv:1203.1034

    2012 arXiv

  20. [28]

    K., & Szymański, G

    Udalski, A., Szymański, M. K., & Szymański, G. 2015, Acta Astron., 65, 1

    2015

  21. [29]

    1997, MNRAS, 284, 172

    Wambsganss, J. 1997, MNRAS, 284, 172

    1997

  22. [30]

    Witt, H. J. 1990, A&A, 236, 311

    1990

  23. [31]

    J., & Mao, S

    Witt, H. J., & Mao, S. 1994, ApJ, 430, 505

    1994

  24. [32]

    C., & Gould, A

    Yee, J. C., & Gould, A. 2023, arXiv e-prints, arXiv:2306.15037

    2023 arXiv

  25. [33]

    2021, ARA&A, 59, 291

    Zhu, W., & Dong, S. 2021, ARA&A, 59, 291

    2021

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.