Pith. sign in

REVIEW 3 major objections 4 minor 37 references

Astrometry-Only Detection of Microlensing Events with Gaia

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read GAME Filter recovers microlensing parameters from Gaia astrometry alone.

desk verdict Solid, well-scoped methods paper with public code; the headline recovery numbers are best-case because the mock data lack blending and use DR3 visit counts. read the letter →

arxiv 2412.14844 v2 pith:VHNH5EN5 submitted 2024-12-19 astro-ph.SR astro-ph.IM

T. Jankovič , A. Gomboc , {L}. Wyrzykowski , U. Kostić , M. Karlica , M. Larma , T. Petrushevska , M. Bronikowski
show 2 more authors
M. Jab{l}ońska Z. Kaczmarek
This is my paper · ORCID
classification astro-ph.SRastro-ph.IM
keywords astrometricmicrolensingGaiaDR4GAMEFilterparallaxlensmassanddistancemockobservationsbinaryfalsepositivesEinsteinradius
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

This paper develops and tests the GAME Filter, a software tool that hunts for astrometric microlensing events in Gaia's astrometric time series using only positional measurements, with no photometric light curve. The claim is that for strong events the filter recovers the lensing parameters—Einstein radius $\theta_E$, event timescale $t_E$, impact parameter $u_0$, time of closest approach $t_0$, and the two microlensing-parallax components—and that the astrometric signal breaks the $\pi_{E,E}$–$\pi_{E,N}$ degeneracy that limits photometric microlensing. The paper validates the filter on mock Gaia DR4 datasets of 50,000 events each, varying source brightness, sky position, visit count, and whether the event peak falls inside or outside the mission window, and it reports that lenses of roughly 1–20 $M_\odot$ out to about 6 kpc can be characterized. This matters because Gaia DR4, expected in 2026, will release full astrometric time series, and astrometry-only detection would let dark or faint lenses—including stellar remnants—be found without waiting for a photometric brightening.

What carries the argument

The carrying mechanism is the astrometric microlensing centroid shift $\delta\theta_C = \frac{u(t)}{u(t)^2 + 2}\,\theta_E$, where $u(t)$ is the lens-source separation measured in units of the Einstein radius $\theta_E$; a passing lens makes the unresolved light centroid trace an ellipse over the source's straight proper-motion track. GAME Filter evaluates the predicted along-scan deviation for trial parameters and minimizes the renormalized microlensing unit weight error (MUWE), a $\chi^2$-like goodness-of-fit, using the L-BFGS-B algorithm with multiple initial guesses for $u_0$, $\pi_{E,E}$, and $\pi_{E,N}$. An event is declared recovered only when the minimized MUWE lies between 0.9 and 1.1 and the L2 optimality error is small, so the fitted ellipse is the object whose shape and orientation carry the lens mass and distance.

What would settle it

A decisive test is to run GAME Filter on real Gaia DR4 astrometric time series for the two known microlensing events OB110462 and GDR3-001 and compare the recovered $\theta_E$ and $\boldsymbol{\pi}_E$ with the published values; agreement within the mock-data error bars would support the filter, and disagreement beyond those errors would falsify the central claim. A complementary calculation is to regenerate the mock observations with $f_{\rm bl} < 1$ and the actual DR4 scanning law: if the recovery rates and parameter accuracies fall far below the reported values, the quoted sensitivity range does not transfer to real data.

Watch

Extended reading notes

Core claim

The paper's central discovery is that a microlensing event can be identified and its lens characterized from the astrometric signal alone: the centroid of the two unresolved images shifts by $\delta\theta_C = \frac{u}{u^2+2}\,\theta_E$ with respect to the unlensed track, tracing an ellipse whose size, shape, and orientation encode the six microlensing parameters. Fitting the along-scan residuals $\Delta x_{\rm obs}$ to this ellipse, and accepting only fits whose normalized unit-weight error is close to 1 and whose optimality error is small, the filter recovers true parameters for 61% of the simulated strong events at $G_0 = 14$, and recovers the microlensing parallax components as a single cluster, not the two degenerate clusters seen in photometry. The authors present this as the basis for characterizing lenses of roughly 0.1–20 $M_\odot$ out to about 7 kpc when the signal is strong, with sensitivity concentrated at nearby high-mass lenses.

Load-bearing premise

The simulated Gaia DR4 observations—Gaussian scatter, no light blending, and visit counts taken from DR3 instead of the not-yet-available DR4 schedule—faithfully represent the real astrometric time series the mission will deliver.

Editorial extensions

If this is right

  • For bright, well-sampled events at $G_0 = 14$ with 281 visits, about 61% pass the filter; 55% of astrometry-only events ($|u_0|>1$) are recovered, rising to 82% when a photometric signal is also present.
  • Astrometry resolves the microlensing-parallax degeneracy: the recovered $\pi_{E,E}$ and $\pi_{E,N}$ form one cluster, because the two components set the ellipse orientation rather than just the light-curve shape.
  • Fainter sources weaken the signal, but at $G_0 = 19$ the parameters for events with $\theta_E \gtrsim 2$ mas are still recovered with accuracy around 80%.
  • Events whose peak falls outside the 2014.5–2020 DR4 window are still recovered, with the recovery fraction dropping about 15 percentage points and parameter errors rising.
  • Binary systems contaminate at roughly 5% for $G_0 = 14$, and the misclassified binaries appear as weak, low-$\theta_E$, high-$|u_0|$ microlensing-like signals.

Reading between the lines

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

  • A natural next step is to fit the blending fraction $f_{\rm bl}$ as an extra parameter; until then, the quoted mass and distance estimates should be treated as lower bounds in fields where the lens or a neighbour contributes light.
  • Because the mock visit counts use the DR3 scanning law while DR4's schedule is not yet public, re-running the filter on the true DR4 epoch geometry is the immediate validation test once the time series arrive.
  • The false-positive estimate was made with uniformly sampled binary orbits; converting the 5% confusion rate into a DR4 contamination prediction would require a realistic Galactic binary population, which the paper defers to future work.
  • If the recovery rates survive real DR4 data, the same residual-ellipse filtering strategy transfers to astrometric time series from other surveys, with epoch sampling being the main difference.
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

3 major / 4 minor

Summary. Jankovič et al. present the GAME Filter, a software pipeline that fits a single-star-plus-microlensing model to Gaia epoch astrometry (along-scan positions) and selects candidate events using a reduced chi-square statistic (MUWE), an L-BFGS-B convergence criterion (Lopt), and boundary checks. They generate mock Gaia DR4 datasets with astromet for three source magnitudes, three sky positions with different visit counts, and an extended t0 range, plus a binary-star dataset to assess false positives. For each 50,000-event dataset they report the fraction of recovered events (Prec) and the fractions with parameters within 10% and 20% of the truth (P10, P20), study parameter degeneracies, and construct a lens mass-distance sensitivity map. The paper concludes that strong events are recovered with about 60% efficiency at G=14, fainter sources down to G=19 can still be characterized when the Einstein radius exceeds about 2 mas, visit counts above about 90 have little impact, events with their peak outside the DR4 window remain partially recoverable, and binary contamination is low.

Significance. If the reported performance transfers to real Gaia DR4 epoch astrometry, this would be a valuable tool for the first systematic astrometry-only microlensing search and would enable lens mass and distance estimates for events that have no photometric signal. The claim that astrometry breaks the photometric microlensing-parallax degeneracy is supported by the corner plots, which show a single cluster for (pi_EE, pi_EN) rather than the two clusters typical of photometric fits. The paper's strengths include the large mock-data campaigns (50,000 events per dataset), the public release of the code, the explicit false-positive binary test, and a transparent list of caveats. The significance is moderated by the idealized nature of the mock observations: the recovery rates and the 1-20 M_sun / 6 kpc reach should be viewed as upper bounds until the filter is tested on more realistic DR4 epoch astrometry with blending and the actual DR4 scanning law.

major comments (3)
  1. [Section 2.2, footnote 6; Table 3] The mock DR4 datasets use visit counts Nv obtained from the DR3 scanning law because scanninglaw does not provide DR4; this is stated in footnote 6. Since the recovery statistics in Table 3 and the sensitivity map in Figure 11 depend directly on the time sampling of the epoch astrometry, the reported rates may not be representative of the actual DR4 cadence. Please rerun the analysis with the best available DR4 scanning-law prediction, or alternatively demonstrate robustness by testing several plausible DR4 cadences (e.g., different starting epochs or updated scheduling simulations) and show that Prec, P20, and P10 change only mildly.
  2. [Section 4.3, Eq. (11)] The simulations assume fbl = 1, i.e., no blending from a luminous lens or unresolved neighbours, and Section 4.3 states that this assumption sets the lower boundary for the Einstein radius and hence for the derived lens mass and distance. With fbl < 1 the centroid shift is diluted and, as Eq. (11) shows, the measured Gaia parallax becomes a luminosity-weighted mix of source and lens parallax. Because the central claim that GAME Filter characterizes lenses at 1-20 M_sun and distances up to about 6 kpc is computed from these fbl = 1 simulations only, the reported reach is an upper bound. I request quantitative experiments in which fbl is varied over [0,1] or unresolved blended neighbours are added, showing how Prec, P20, and P10 degrade; if such simulations are not feasible, the abstract and conclusions should explicitly state that the quoted mass and distance ranges assume negligible blending.
  3. [Section 2.3 and Appendix A.2] The acceptance thresholds (0.9 < MUWEmin < 1.1, Lopt < 0.015, and Lthresh = 0.01) are selected from histograms of the same mock datasets that are then used to report recovery rates. This in-sample calibration can only bias the reported Prec, P20, and P10 optimistically. Please validate the thresholds on an independent subset of the mock data (e.g., a cross-validation split or a separately generated calibration set), or fix the thresholds a priori and then report the recovery rates on unseen data. The relation between the Lopt < 0.015 acceptance criterion and the Lthresh = 0.01 restart criterion should also be clarified, since the reader cannot tell whether these are two different thresholds or the same threshold quoted with different precision.
minor comments (4)
  1. [Figure 7 and Figure 8 captions] The captions refer to a dataset named 'lens_G14_N209_extended', but Table 3 lists the extended dataset as 'lens_G14_N281_extended'; Section 3.4 also uses 'lens_G14_N281_extend'. Please make the dataset names consistent throughout.
  2. [Section 4.1, item (i)] The sentence 'The strongest microlensing signal occurs at u0 = sqrt(2) theta_E and t0' is dimensionally inconsistent because u0 is dimensionless while theta_E is an angle. The maximum of Eq. (9) occurs at u = sqrt(2), i.e., at angular separation sqrt(2) theta_E; please rephrase (e.g., 'at u0 = sqrt(2) for a closest approach that sits at the maximum, or at separation u(t) = sqrt(2)').
  3. [Equation (10)] The MUWE denominator is N - 11; when N <= 11 this reduced chi-square is undefined or negative. Since short events with sparse sampling are part of the claimed domain of applicability, the paper should state how such cases are handled in the code or exclude them explicitly from the analysis.
  4. [Abstract vs. Conclusions, Section 4.2] The abstract states lens masses from approximately 1 to 20 M_sun, while the first bullet of the Conclusions states 0.1 to 20 M_sun. Figure 11 shows that sensitivity is below 10% for ML < 2 M_sun beyond 4 kpc, so the broader range in the Conclusions is not supported by the same sensitivity criterion. Please harmonize the quoted ranges and define the sensitivity threshold used for the headline claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's validation is an injection–recovery simulation whose output is not equivalent to its input, and the caveats are external-validity limitations rather than circular reasoning.

full rationale

The paper's central claims are validated by generating mock Gaia observations with known microlensing parameters and then testing whether GAME Filter recovers those parameters from noisy, sparsely sampled along-scan measurements. This is a genuine injection–recovery test: the filter fails for a large fraction of events (e.g., Prec drops to 16.3% at G0 = 19 in Table 3), so success is not guaranteed by construction. The recovery criteria (MUWEmin and Lopt thresholds) are calibrated on the same mock datasets in Section 2.3, which is an in-sample calibration that could make the reported Prec/P20/P10 rates optimistic, but this does not make the central derivation circular: the mass/distance sensitivity in Section 4.2 is based on counting fits within 20% of true parameter values, independent of those thresholds, and the degeneracy-breaking claim (Section 3.1) follows from the fitted astrometric ellipse orientation, not from the thresholds. There is no load-bearing self-citation: the coauthor citations (Jabłońska et al. 2022 for GDR3-001, Wyrzykowski et al. 2023 for the DR3 catalog, and the jaxtromet code) are published, externally available results used as context or benchmarks, not as justification for the filter's validity; the future-work self-citation (Kaczmarek et al., in prep.) makes no evidential claim. No uniqueness theorem from the authors is invoked, and no ansatz is smuggled in via citation: the astrometric microlensing equations are standard (Dominik & Sahu 2000; Paczynski 1986). The genuine limitations, clearly flagged in the paper, are external-validity concerns rather than circularity: footnote 6 uses DR3 visit counts because the scanninglaw package lacks DR4, Section 2.2 adopts a Gaussian along-scan noise model, and Section 4.3 assumes fbl = 1 (no blending). These affect whether the reported performance transfers to real Gaia DR4 data, but they do not make any prediction equivalent to its input by construction.

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

The analysis rests on standard microlensing equations, a simplified no-blending noise model, uniform parameter distributions for testing, and DR3 visit counts as a stand-in for DR4. No new physical entities are introduced.

free parameters (3)
  • MUWE acceptance bounds = 0.9 < MUWEmin < 1.1
    Chosen by eye from histograms, where histograms show a sharp decline, of the same mock datasets used to compute recovery rates. These bounds define what counts as a recovered event.
  • Lopt acceptance threshold = 0.015 in main text, 0.01 in Appendix A.2
    Set as the value below which 90 percent of the Lopt distribution lies for the lens_G14_N281_DR4 dataset. Used to decide whether to restart minimization with new initial guesses.
  • theta_E initial guess calibration = theta_E0 ~ 3 * max - 2.25
    Empirical linear fit to simulated residual amplitudes (Figure A.1, lower panel). It seeds the minimizer and influences convergence.
assumptions (5)
  • standard math Standard point-source microlensing equations (Eqs. 8 and 9) describe the astrometric centroid shift.
    Taken from Dominik & Sahu (2000) and Belokurov & Evans (2002); the paper does not re-derive them.
  • domain assumption No blending: the lens and unrelated sources contribute no light, so fbl = 1.
    Stated in Section 2.2; the caveat in Section 4.3 acknowledges that blending would change the observable centroid and inferred masses and distances.
  • domain assumption Gaia measurement noise is Gaussian with magnitude-dependent variance along the scan direction.
    Mock observations are generated by scattering positions according to magnitude-dependent error bars; real Gaia DR4 systematics are not modeled.
  • domain assumption Uniform distributions of microlensing and binary parameters are adequate for testing filter sensitivity.
    Used to explore parameter space; the authors state this is not a realistic Galactic model and defers event-rate estimates.
  • domain assumption Gaia DR3 visit counts approximate DR4 scanning statistics.
    Footnote 6: the scanninglaw package does not provide DR4 data, so Nv values from DR3 were used.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Astrometry-Only Detection of Microlensing Events with Gaia." pith.science (2026). https://pith.science/paper/VHNH5EN5

@misc{pith2026241214844,
  author       = {Pith},
  title        = {Pith review of: Astrometry-Only Detection of Microlensing Events with Gaia},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VHNH5EN5}},
  note         = {Machine review of arXiv:2412.14844}
}
read the original abstract

Astrometric microlensing events occur when a massive object passes between a distant source and the observer, causing a shift of the light centroid. The precise astrometric measurements of the Gaia mission provide an unprecedented opportunity to detect and analyze these events, revealing properties of lensing objects such as their mass and distance. We develop and test the Gaia Astrometric Microlensing Events (GAME) Filter, a software tool to identify astrometric microlensing events and derive lensing object properties. We generated mock Gaia observations for different magnitudes, number of Gaia visits, and events extending beyond Gaia's observational run. We applied GAME Filter to these datasets and validated its performance. We also assessed the rate of false positives where binary astrometric systems are misidentified as microlensing events. GAME Filter successfully recovers microlensing parameters for strong events. Parameters are more difficult to recover for short events and those extending beyond Gaia's run, where only a fraction of the events is observed. The astrometric effect breaks the degeneracy in the microlensing parallax present in photometric microlensing. For fainter sources, the observed signal weakens, reducing recovered events and increasing parameter errors. However, even for Gaia G-band magnitude 19, parameters can be recovered for Einstein radii above two mas. Observing regions with varying numbers of Gaia visits has minimal impact on filter accuracy when the number of visits exceeds 90. Additionally, even if the peak of a microlensing event lies outside Gaia's run, microlensing parameters can still be recovered. GAME Filter characterizes lenses with astrometry-only data for lens masses from approximately 1 to 20 solar masses and distances up to 6 kpc.

Figures

Figures reproduced from arXiv: 2412.14844 by the authors.

Figure 1
Figure 1. Upper panel: Tracks of the light centroid for a single [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. 2D histograms of values of the parameters obtained [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Corner plots for ratios R between true and minimised values of individual parameters for recovered events in the mock dataset lens_G14_N281_DR4 (see Tables 1 and 3). fect is most apparent for events with a short tE. Furthermore, the histogram of u0 transitions from a bimodal distribution to a sin￾gle peak as the t0 range increases. There are also fewer recovered events for large θE, as larger θE corresponds to longe… view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: Relative errors for recovered events for mock [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 7
Figure 7. Figure 7: Distributions of parameter values after minimisation [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 9
Figure 9. Figure 9: Distributions of parameter values after minimisation [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 10
Figure 10. Figure 10: Gaia along scan measurements xobs as a function of tobs for G0 = 14 (blue circle symbols) and 19 (red diamond symbols). The green vertical dashed lines correspond to t0. We assume the same single source and microlensing parameters as in [PITH_FULL_IMAGE:figures/full_…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 29 canonical work pages

  1. [1]

    Belokurov, V. A. & Evans, N. W. 2002, MNRAS, 331, 649

    work page 2002

  2. [2]

    2021, arXiv preprint arXiv:2105.15183

    Blondel, M., Berthet, Q., Cuturi, M., et al. 2021, arXiv preprint arXiv:2105.15183

    arXiv 2021

  3. [3]

    2001, MNRAS, 327, 868

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

    work page 2001

  4. [4]

    Chen, I.-K., Kongsore, M., & Tilburg, K. V. 2023, Journal of Cosmology and Astroparticle Physics, 2023, 037

    work page 2023

  5. [5]

    & Sahu , K

    Dominik , M. & Sahu , K. C. 2000, , 534, 213

    work page 2000

  6. [6]

    2019, The Messenger, 178, 45

    Dong , S., M \'e rand , A., Delplancke-Str \"o bele , F., Gould , A., & Zang , W. 2019, The Messenger, 178, 45

    work page 2019

  7. [7]

    1936, Science, 84, 506

    Einstein , A. 1936, Science, 84, 506

    work page 1936

  8. [8]

    2024, The Open Journal of Astrophysics, 7, 100

    El-Badry , K., Lam , C., Holl , B., et al. 2024, The Open Journal of Astrophysics, 7, 100

    work page 2024

Show all 37 references
  1. [9]

    E., et al

    Fardeen , J., McGill , P., Perkins , S. E., et al. 2024, , 965, 138

    2024

  2. [10]

    , Babusiaux, C

    Gaia Collaboration , Arenou, F. , Babusiaux, C. , et al. 2023, A&A, 674, A34

    2023

  3. [11]

    Gaia Collaboration , Brown, A. G. A. , Vallenari, A. , et al. 2018, A&A, 616, A1

    2018

  4. [12]

    , de Bruijne, J

    Gaia Collaboration , Prusti, T. , de Bruijne, J. H. J. , et al. 2016, A&A, 595, A1

    2016

  5. [13]

    Gaia Collaboration , Smart, R. L. , Sarro, L. M. , et al. 2021, A&A, 649, A6

    2021

  6. [14]

    2004, , 606, 319

    Gould , A. 2004, , 606, 319

    2004

  7. [15]

    2023, , 674, A9

    Halbwachs , J.-L., Pourbaix , D., Arenou , F., et al. 2023, , 674, A9

    2023

  8. [16]

    D., & Polnarev , A

    Hog , E., Novikov , I. D., & Polnarev , A. G. 1995, , 294, 287

    1995

  9. [17]

    Hunter, J. D. 2007, Computing in Science & Engineering, 9, 90

    2007

  10. [18]

    , Wyrzykowski, Ł

    Jabłońska, M. , Wyrzykowski, Ł. , Rybicki, K. A. , et al. 2022, A&A, 666, L16

    2022

  11. [19]

    2022, Astron

    Klüter, J., Bastian, U., Demleitner, M., & Wambsganss, J. 2022, Astron. J., 163, 176

    2022

  12. [20]

    , Bastian, U

    Klüter, J. , Bastian, U. , Demleitner, M. , & Wambsganss, J. 2018, A&A, 620, A175

    2018

  13. [21]

    & Paczyński, B

    Mao, S. & Paczyński, B. 1996, The Astrophysical Journal, 473, 57

    1996

  14. [22]

    C., Evans, N

    McGill, P., Smith, L. C., Evans, N. W., Belokurov, V., & Smart, R. L. 2018, MNRAS, 478, L29

    2018

  15. [23]

    1996, , 470, L113

    Miralda-Escude , J. 1996, , 470, L113

    1996

  16. [24]

    & Yoshii , Y

    Miyamoto , M. & Yoshii , Y. 1995, Astron. J., 110, 1427

    1995

  17. [25]

    2022, , 937, L24

    Mr \'o z , P., Udalski , A., & Gould , A. 2022, , 937, L24

    2022

  18. [26]

    1986, , 301, 503

    Paczynski , B. 1986, , 301, 503

    1986

  19. [27]

    Penoyre , Z., Belokurov , V., & Evans , N. W. 2022, , 513, 2437

    2022

  20. [28]

    1964, MNRAS, 128, 307

    Refsdal, S. 1964, MNRAS, 128, 307

    1964

  21. [29]

    C., Anderson, J., Casertano, S., et al

    Sahu, K. C., Anderson, J., Casertano, S., et al. 2017, Science, 356, 1046

    2017

  22. [30]

    C., Anderson, J., Casertano, S., et al

    Sahu, K. C., Anderson, J., Casertano, S., et al. 2022, , 933, 83

    2022

  23. [31]

    2023, MNRAS, 527, 1177

    Su, J., Wang, J., Zhang, Y., Cheng, X., & Yang, L. 2023, MNRAS, 527, 1177

    2023

  24. [32]

    2003, , 53, 291

    Udalski , A. 2003, , 53, 291

    2003

  25. [33]

    Walker , M. A. 1995, , 453, 37

    1995

  26. [34]

    , Kruszyńska, K

    Wyrzykowski, Ł. , Kruszyńska, K. , Rybicki, K. A. , et al. 2023, A&A, 674, A23

    2023

  27. [35]

    2018, MNRAS, 480, 236

    Zurlo, A., Gratton, R., Mesa, D., et al. 2018, MNRAS, 480, 236

    2018

  28. [36]

    , " * write output.state after.block = add.period write newline

    ENTRY address archiveprefix author booktitle chapter edition editor howpublished institution eprint journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sent...

  29. [37]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.