Pith. sign in

REVIEW 3 major objections 6 minor 38 references

Mapping the Inner Milky Way with Infrared-Derived Distances to AGB Stars

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A machine-learning model trained on infrared magnitudes alone assigns statistical distances to more than 36,000 dust-obscured AGB stars, and the resulting catalog recovers the Milky Way's bar and separates young and old bulge Miras.

desk verdict A genuinely useful distance catalog for 36k obscured AGB stars, with an honest but under-supported extrapolation from the MSX training set that needs a quantitative check before the 6% accuracy claim is taken at face value. read the letter →

arxiv 2608.10577 v1 pith:KP352DIT submitted 2026-08-11 astro-ph.GA astro-ph.IMastro-ph.SR

Rajorshi Bhattacharya , Subhajit Kar , Loránt O. Sjouwerman , Anupam Bhardwaj This is my paper · ORCID
classification astro-ph.GAastro-ph.IMastro-ph.SR
keywords MilkyWayGalaxyAsymptoticgiantbranchstarsGalacticstructureInfraredphotometryMachinelearningMiravariablesAKARIsurveyStatisticaldistances
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 tries to show that statistical distances to heavily obscured oxygen-rich asymptotic giant branch (AGB) stars can be derived from raw infrared photometry alone. A gradient-boosted regression model trained on about 10,000 stars from the BAaDE survey's MSX-selected sample, whose distances come from fitting their spectral energy distributions (SEDs), reaches a test-set accuracy of about 6% and is then applied to 36,134 stars selected from the AKARI infrared satellite survey, producing a distance catalog with a combined uncertainty of about 36% over 0.5-20 kpc. The authors treat the catalog as a population-level product rather than precise per-star distances. If the distances are trustworthy, they recover the expected bar-like asymmetry of the inner Galaxy, show a bulge vertical dispersion about 30% larger than the disk, and show that long-period Miras outline the bar while short-period Miras form a smoother, older component. This matters because direct astrometry and period-luminosity relations are unreliable for dust-obscured AGB stars, so a photometric distance scale would make a large hidden population usable for mapping Galactic structure.

What carries the argument

The mechanism is the empirical photometry-to-distance mapping learned by the XGBoost regressor. The feature vector is five apparent magnitudes (2MASS $J,H,K_s$ and AKARI 9 and 18 $\mu$m) with no extinction correction; the training labels are SED-based distances for BAaDE/MSX stars, so the ML outputs inherit that distance scale. The transfer works because the AKARI-selected sources occupy similar near- and mid-IR color-magnitude space to the training sample, and the paper checks this by comparing predicted distances with independent estimates. A secondary mechanism is the dimensionless vertical dispersion $\tilde{\sigma}_z = \sigma_z / Z_{\max}$, which normalizes the measured vertical dispersion by the maximum accessible height under the latitude cut, allowing bulge and disk thickness to be compared despite the $|b| \lesssim 6^\circ$ selection.

What would settle it

Measure VLBI parallaxes for 20-30 AKARI-selected O-rich AGB stars with SiO masers located beyond 5 kpc and compare them with the ML distances: if the median residual exceeds the quoted ~36% uncertainty or correlates with Galactic longitude, the transferability assumption is falsified.

Watch

Extended reading notes

Core claim

The central claim is that apparent multi-band infrared photometry - 2MASS $J,H,K_s$ and AKARI 9 and 18 $\mu$m - contains enough information to predict statistical distances to oxygen-rich AGB stars in the inner Milky Way. The paper builds an XGBoost (gradient-boosted decision tree) regressor on BAaDE/MSX stars whose distances come from an earlier SED-based analysis; the model reaches $R^2 \approx 0.98$ and a mean absolute percentage error near 6% on an independent test set, with a binned fit of predicted versus SED distance consistent with slope $1.01 \pm 0.01$ and zero offset. Applying the model to the AKARI/IRC color-selected sample yields distances for 36,134 sources with a combined ~35.5% uncertainty. The authors validate the scale against Gaia parallaxes for high-latitude OH/IR stars, bolometric luminosities from an OH/IR catalog, and several Mira period-luminosity relations, finding broad agreement with Galactic calibrations and systematic offsets from LMC-calibrated ones that they attribute to known distance-scale and metallicity differences. They use the catalog to recover the bar's near-far distance asymmetry, measure a dimensionless vertical dispersion $\tilde{\sigma}_z \approx 0.40$ in the bulge versus about 0.30 in the disk, and show that Miras with periods longer than 400 days trace the bar while shorter-period Miras do not; they explicitly decline to treat the far-side enhancement in short-period Miras as a physical result.

Load-bearing premise

The load-bearing premise is that the photometry-to-distance relation learned on the brighter, near-side-heavy MSX sample transfers directly to the fainter AKARI sample with broader far-side longitude coverage, so any undetected difference in color-magnitude space or distance distribution between the two samples would shift the distances and the structural conclusions.

Editorial extensions

If this is right

  • The published catalog extends statistical distance coverage for dust-obscured O-rich AGB stars from about 0.5 to 20 kpc, roughly tripling the number of inner-Galaxy AGB stars with distance information compared with the MSX-based training sample.
  • If the distance scale is correct, the near-far asymmetry in the longitude-selected distance distributions is an independent confirmation that the Milky Way's bulge is barred, with the bar's near end at positive longitudes.
  • Long-period Miras ($P>400$ days) become established as tracers of a younger, more massive bar population, while short-period Miras ($P\le 400$ days) trace a smoother, older spheroidal component.
  • The dimensionless vertical dispersion profile, high and flat in the bulge and lower and flat in the disk with the bulge about 30% larger, provides a population-level measure of the inner Galaxy's vertical structure that is insensitive to the adopted scale-height conversion.
  • Agreement with Galactic-calibrated period-luminosity relations, at the ~12% level for mid-IR relations, supports using these statistical distances as a prior or cross-check for individual Mira distances in the inner Galaxy.

Reading between the lines

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

  • If the MSX-to-AKARI transfer holds, the same photometry-to-distance training strategy could be extended to other infrared surveys to reach fainter and more crowded sightlines, including the nuclear stellar disk that this sample cannot adequately cover.
  • The paper's Gaia validation is limited to 166 high-latitude sources outside the $|b|<5^\circ$ footprint, so the in-plane transferability is not directly tested by Gaia; VLBI maser parallaxes for AKARI-selected stars inside the footprint would close that gap and could also settle whether the short-period Mira far-side enhancement is real.
  • Because pulsation period is not a model input, the period-stratified bar morphology is an emergent result, but period-correlated changes in dust-shell emission could still leak into the distances; comparing ML distances for Miras of different periods at fixed apparent magnitudes would separate a population effect from an SED-driven artifact.
  • Combining the catalog with line-of-sight velocities from the BAaDE survey could upgrade the projected structural maps into kinematic constraints on the bar, since density alone fixes neither the pattern speed nor the three-dimensional orientation.
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 / 6 minor

Summary. The paper develops an XGBoost regression model that maps 2MASS and AKARI infrared photometry to heliocentric distances for oxygen-rich AGB stars, trained on MSX-selected stars with SED-based distances from Bhattacharya et al. (2024). The model achieves R^2 ≈ 0.98 and a MAPE of about 6% on a held-out test set drawn from the same MSX sample, and is then applied to 36,134 AKARI-selected AGB candidates that do not overlap the training sample. The resulting distance catalog is validated by comparison with Gaia parallaxes (for 166 sources at |b| > 5 deg), with several period-luminosity relations, and with OH/IR bolometric luminosities. Using these distances, the authors report that the combined AKARI+MSX sample recovers the Galactic bar through an asymmetry in near/far distance distributions, finds a roughly constant vertical dispersion in the bulge that is about 30% larger than in the disk, and shows that long-period Miras (P > 400 d) trace a barred morphology more strongly than short-period Miras. The paper is careful to describe the distances as statistical, with a quoted per-source uncertainty of about 35.5%, and explicitly declines to interpret a far-side enhancement of short-period Miras as a physical result.

Significance. If the distance catalog is reliable at the population level, it would be a useful resource for mapping the inner Milky Way in heavily obscured regions where Gaia astrometry and kinematic distances are unavailable or biased. The methodology is straightforward, reproducible in principle, and the paper includes several diagnostic checks: 5-fold cross-validation, Optuna-based hyperparameter optimization, a held-out test set, binned 1:1 diagnostics, and comparisons with multiple independent distance estimators. The paper also explicitly acknowledges that the ML distances are tied to the SED-based distance scale of the training set and cannot be more accurate than those training labels. The main scientific claims, that long-period Miras preferentially trace the bar and that bulge and disk vertical dispersions differ, are consistent with prior work and are presented with appropriate caution. However, the central transfer from the MSX training sample to the AKARI application sample, and the interpretation of the 6% MAPE as a measure of accuracy, need stronger quantitative support before the catalog and its structural inferences can be fully endorsed.

major comments (3)
  1. [Section 3.1, with implications for Sections 4.2, 5.2, and 5.3.2] The load-bearing assumption that the AKARI-selected sources occupy a color-magnitude space 'similar' to the MSX training sample is asserted without a quantitative demonstration. Figure 1 shows that the MSX training set has a pronounced far-side deficit for -30 deg < l < 0 deg, while the AKARI sample provides substantially more far-side coverage, and Section 2.1 states that AKARI reaches ~2-3 mag fainter than MSX. Because XGBoost is an interpolator, predictions for AKARI sources that are fainter or on the far side can be unconstrained and may inherit the training set's selection bias. I recommend adding a quantitative feature-space overlap analysis (e.g., density distributions of AKARI sources versus the training set in the five-dimensional photometric space, or a nearest-neighbor distance statistic) and a side-split or far-side validation in which the model is trained on one longitude range and tested on the other. Without such a check, the far-side structural results in Figures 8, 12, 13, 14, and 15 may be artifacts of extrapolation.
  2. [Table 3 and Section 4.2.2] The headline '6% MAPE' measures the model's ability to reproduce the SED-derived training labels, not the absolute accuracy of the distances. The paper states this in Section 3.1, but the abstract and Section 4.1.3 present the 6% value in a way that can easily be misread as the total distance error. In addition, the per-source error budget of 35.5% is obtained by summing the maximum SED uncertainty (~35%) and the mean ML MAPE (~6%) in quadrature. This is not a proper propagation of per-source uncertainties: the 35% value is an upper limit, not a random error, and the 6% is a mean absolute percentage error, not a standard deviation. I recommend rephrasing the abstract to say that the model reproduces SED-based distances with a MAPE of about 6%, and either providing per-source error estimates or explicitly labeling the 35.5% as a conservative ensemble-level envelope rather than an individual 1-sigma uncertainty.
  3. [Section 5.3.2, Figures 14-15] The paper correctly disclaims the far-side enhancement of short-period Miras as a physical result, but the same extrapolation risk applies to the long-period Mira bar signal and to the vertical-dispersion map. Although period is not an explicit model feature, period correlates with mid-IR colors and circumstellar dust emission, so period-correlated SED differences can induce apparent period-dependent spatial structure through the model's feature mapping. The argument that the model 'has no explicit mechanism' to assign period-dependent distances is therefore not sufficient. I recommend testing whether the bar morphology (Figure 14) and the vertical-dispersion profile (Figures 12-13) persist when the analysis is restricted to the distance and longitude range where the training sample has good coverage, or when the model is retrained on a longitude-balanced subset of the MSX sample.
minor comments (6)
  1. [References and Section 5.1.1] The author names for the Gaia OH/IR catalogs are inconsistent: the text uses 'B. L. Marti et al. (2025)' and 'B. Lopez Marti et al. (2025)', while the reference list has both 'Lopez Marti, B., Jimenez-Esteban, F. M., Engels, D., & Garcia-Lario, P. 2025' and 'Marti, B. L., Jimenez-Esteban, F., Engels, D., & Garcia-Lario, P. 2025'. Please unify the citation style and remove the duplicate reference.
  2. [Abstract and Section 4.2.2] The abstract quotes 'a 36% total error margin' while Section 4.2.2 derives 35.5%; please make these numbers consistent or explain the rounding.
  3. [Equation (1) and Section 4.1.1] The paper reports a 'MAPE-based accuracy metric' of ~94% and then translates it to 'a mean deviation of ~6%'. Reporting the MAPE directly (about 6%) would be clearer and would avoid the impression that the model achieves 94% accuracy in an absolute sense.
  4. [Figure 5 caption] The caption says 'the blue points show the binned median trend, indicating the 16th-84th percentile spread within each bin,' but it is unclear whether the blue points are the medians and the shaded band represents the percentile spread, or whether both are shown. Please clarify the caption and ensure the figure legend matches.
  5. [Section 5.1.2] The bolometric luminosities derived from ML distances are compared with Feast et al. (1989) P-L expectations without explicitly stating in the comparison paragraph that the bolometric fluxes are not corrected for interstellar extinction; the paper later notes that extinction correction would raise the luminosities, but this caveat should appear before the comparison is interpreted.
  6. [Figures 14 and 16] The colorbar units are given as N kpc^-2 in Figure 14 and N kpc^-3 in Figure 16; if these are intentional (projected versus volume-corrected densities), please state this in the captions, but currently the units appear inconsistent at a glance.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the ML distance scale is explicitly tied to externally anchored SED distances, and the headline structural results survive independent external checks and period splits.

full rationale

The training labels are the SED-based distances of R. Bhattacharya et al. (2024), but this is not a circular reduction because the paper openly defines the ML output as a surrogate for that SED distance scale: 'the ML-derived distances should be interpreted as empirical distances tied to the SED-based distance scale of the training set' (Sect. 3.1), and the SED scale itself is anchored by VLBI parallax sources and Galactic-center constraints external to the ML fit. The ~6% test MAPE is a fidelity measure of the photometric mapping to those labels, not a claim of independent physical accuracy; the abstract's 'independent test set' refers to a held-out split of the label sample, and the paper separately validates against Gaia parallaxes (Sect. 5.1.1), the external Sanders (2023) P-L relation, and Catchpole et al. (2016) Mira distances. The period-dependent bar result is not built in because period is not an input feature (Sect. 5.3.2), making the short/long-period split an independent check. The MSX-to-AKARI transfer is a stated assumption whose failure would bias results, but it is an extrapolation/robustness limitation, not a reduction of the output to the input; the authors also explicitly decline to interpret the far-side short-period enhancement as physical. Self-citations are numerous but are anchored to external measurements, so they do not make the derivation circular.

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

The central distance scale relies on a chain of assumptions: the SED-derived training distances, the photometric-space overlap between MSX and AKARI samples, the absence of extinction-driven bias in raw magnitudes, and the attribution of external comparison offsets to known biases in Gaia and LMC-calibrated P-L relations. The XGBoost hyperparameters and sample-definition thresholds are fitted or chosen values that affect the catalog.

free parameters (5)
  • XGBoost hyperparameters = n_estimators=360, max_depth=7, learning_rate=0.03144, min_child_weight=4, subsample=0.66957
    Tuned via Optuna on the training set; the distance predictions depend on these choices.
  • AKARI color selection range = -0.6 < [9]-[18] < 0.6
    Chosen to isolate O-rich AGB stars; determines the photometric space of the target catalog.
  • Mira period threshold = P = 400 d
    Used to split short- and long-period Miras for the bar morphology comparison.
  • Vertical dispersion scaling factor c = not fitted (left general)
    Converts dimensionless sigma_z to physical scale height; the claimed bulge/disk difference is independent of c but the physical interpretation depends on it.
  • Zmax normalization = Zmax = d sin(6 deg)
    Normalizes the vertical dispersion by the maximum accessible height, which affects the dimensionless sigma_z values.
assumptions (5)
  • domain assumption The SED-based distances from Bhattacharya et al. 2024 are accurate enough to serve as training labels.
    The model is trained on these distances, so any systematic error in them propagates to all ML distances.
  • domain assumption The AKARI-selected sample occupies the same photometric-distance space as the MSX training sample.
    Stated in Sect. 3.1 to justify applying the model without large extrapolation.
  • domain assumption Apparent IR magnitudes without extinction correction can map to distance without systematic extinction bias.
    The model uses raw magnitudes; the authors test residual correlation with A_Ks for a subsample and find no significant trend.
  • domain assumption Non-AGB contamination (~2.4%) is negligible for population-level analysis.
    Based on SIMBAD classifications; small fractions of YSOs and C-rich stars remain.
  • ad hoc to paper Systematic offsets in Gaia and P-L comparisons are attributable to known biases in those methods, not to the ML distances.
    Used to reconcile the offsets in Sects. 5.1.1 and 5.1.3, e.g., Lutz-Kelker bias and LMC zero-point differences.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mapping the Inner Milky Way with Infrared-Derived Distances to AGB Stars." pith.science (2026). https://pith.science/paper/KP352DIT

@misc{pith2026260810577,
  author       = {Pith},
  title        = {Pith review of: Mapping the Inner Milky Way with Infrared-Derived Distances to AGB Stars},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KP352DIT}},
  note         = {Machine review of arXiv:2608.10577}
}
read the original abstract

The structure and evolution of the Milky Way (MW) can be traced with distance estimates to the evolved stellar populations in the inner galactic region. However, direct astrometric distances remain unavailable or highly uncertain for the majority of these sources due to instrumental limitations, large angular diameters, and complex variability. In this work, we develop a supervised machine-learning model to estimate statistical distances to oxygen-rich asymptotic giant branch (AGB) stars selected from the AKARI mid-infrared survey. We build an XGBoost regression model that maps multi-band IR photometry to distance using a training set of AGB stars with previously derived SED-based distances achieving a mean absolute percentage error (MAPE) of 6% on an independent test set. Distance estimates for over 36,000 AGB sources are obtained within a 36% total error margin, greatly expanding distance coverage (0.5-20 kpc) for dust-obscured AGB populations in the Galactic plane. We find good agreement with reported distances for Galactic Mira variables and independent period-luminosity relations. Utilizing the expanded distance set, we further investigate the spatial distribution of Mira variables in the Galactic bulge and disk. Longer-period Miras preferentially trace the bulge's barred morphology compared to their shorter-period counterparts that populate the disk. We also find roughly constant, but differing, relative scale heights for the bulge and disk, with the bulge vertical dispersion about 30% larger. These findings show that IR photometry-derived statistical distances can recover large-scale Galactic structures and establish long-period Mira variables as efficient tracers of stellar populations in heavily obscured regions of the MW.

Figures

Figures reproduced from arXiv: 2608.10577 by the authors.

Figure 1
Figure 1. Histogram showing the distribution for galactic longitude (l) in degrees of the SED-distance (MSX) sample, the AKARI sample and the combined sample used in this study. Note that the AKARI sample is restricted within |b| ≤ 5 ◦ to match the MSX/BAaDE footprint. itudes of |b| < 5 ◦ , i.e., to match the MSX footprint. We also assessed the contamination level by cross-matching our sample with SIMBAD. After removing dupli… view at source ↗
Figure 2
Figure 2. Sky distribution of the MSX-selected and AKARI-selected samples in Galactic coordinates (l, b). The MSX-selected sample is shown as blue open circles, while the AKARI-selected sample is shown as gray triangles. The AKARI sample extends over a broader longitude range than the MSX sample, while both samples occupy a similar low-latitude Galactic footprint. This comparison complements the longitude histogram in [PITH_… view at source ↗
Figure 3
Figure 3. Correlation matrix between the non–extinction-corrected IR photometric 2MASS and AKARI bands and the SED–derived distances for the MSX-selected sample (training + test set). ple (∼ 36,000 sources) would require extensive ancillary data and could introduce additional systematic uncer￾tainties. We would like to note that the ML-derived distances should be interpreted as empirical distances tied to the SED-based distan… view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: ML-estimated or predicted distances for the test set versus the input original SED-derived distance (R. Bhat￾tacharya et al. 2024). Red points show individual sources and the dashed line marks the one-to-one relation. The close alignment indicates strong model performa…
Figure 5
Figure 5. Figure 5: Comparison of ML-predicted distances with SED-based distances for the test sample. Individual sources are binned in two dimensions to reduce scatter in high-den￾sity regions. The black dashed diagonal line represents the 1:1 relation, while the blue points show the bin…
Figure 6
Figure 6. Figure 6: Training and test accuracy as a function of boost￾ing rounds for the XGBRegressor model. The curves display how model performance evolves over the course of training, with early boosting rounds producing rapid improvements followed by convergence toward a stable platea…
Figure 7
Figure 7. Figure 7: Projected two-dimensional (X, Y ) density maps of AGB candidates in the Galactic plane (i.e., |b| < 5 ◦ ). The Galactic Centre (GC) is marked by a filled circle at (0, 0), and the Sun is indicated by a star at (0, 8.277 kpc). An ellipse centered on the GC, with a semi-…
Figure 8
Figure 8. Figure 8: Histogram highlighting the distance-based pop￾ulation distribution of the combined AKARI AGB sample and MSX-selected sources for the near (8◦ < l < 16◦ ) and far-side (−12◦ < l < −4 ◦ ) of the MW. populations. We further discuss this comparison and its implications in …
Figure 10
Figure 10. Figure 10: Bolometric magnitude distribution for 143 Gaia OH/IR sources in our sample without Gaia parallaxes satis￾fying a fractional parallax error < 20%, whose luminosities are derived using ML-based distances, compared with 222 other sources from B. L. Mart´ı et al. (2025) t…
Figure 11
Figure 11. Figure 11: Comparison between ML-predicted distances and P-L-based distances for different cross-matched samples. The panels are ordered clockwise from the top-left: OGLE-based near-IR P-L distances from P. Iwanek et al. (2023), mid-IR MSX-based P-L distances from M. O. Lewis et…
Figure 12
Figure 12. Figure 12: Heatmap of the dimensionless vertical dispersion, ˜σz, in the Galactic plane. The left panel shows ˜σz computed in (X, Y ) bins, where the colorbar represents the value of ˜σz, while the right panel shows the corresponding number of sources per bin. The Sun is marked …
Figure 13
Figure 13. Figure 13: Radial variation of the dimensionless vertical dispersion, ˜σz, as a function of Galactocentric radius, RGC. The values are derived by averaging over spatial bins in the (X, Y ) plane. The error bars represent the uncertainty in each radial bin, which depends on the n…
Figure 14
Figure 14. Figure 14: Spatial distributions of Mira variables in the Galactic bulge as a function of pulsation period. The left panel shows the distribution of short-period Miras (P < 400 d), while the right panel shows the distribution of long-period Miras (P > 400 d), using ML-based dist…
Figure 15
Figure 15. Figure 15: Normalized heliocentric-distance distributions of Mira variables separated by pulsation period. The left panel shows sources within −15◦ < l < 15◦ , while the right panel includes all Galactic longitudes; both panels are restricted to −4.5 ◦ < b < 4.5 ◦ following R. M…
Figure 16
Figure 16. Figure 16: Volume-corrected spatial distributions of Mira variables in the Galactic bulge as a function of pulsation period. The left panel shows the distribution of short-period Miras (P < 400 d), while the right panel shows the distribution of long-period Miras (P > 400 d), us…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 28 canonical work pages

  1. [1]

    2019, Astronomy & Astrophysics, 625, L10

    Abuter, R., Amorim, A., Baub¨ ock, M., et al. 2019, Astronomy & Astrophysics, 625, L10

    work page 2019

  2. [2]

    2019, in Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, 2623–2631 21 Albarrac ´ ın, R., Zoccali, M., Carvajal, J

    Akiba, T., Sano, S., Yanase, T., Ohta, T., & Koyama, M. 2019, in Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, 2623–2631 21 Albarrac ´ ın, R., Zoccali, M., Carvajal, J. O., et al. 2024, Astronomy & Astrophysics, 693, A28

    work page 2019

  3. [3]

    Andriantsaralaza, M., Ramstedt, S., Vlemmings, W. H. T., & De Beck, E. 2022, A&A, 667, A74, doi: 10.1051/0004-6361/202243670

    doi:10.1051/0004-6361/202243670 2022

  4. [4]

    Bailer-Jones, C. A. 2015, Publications of the Astronomical Society of the Pacific, 127, 994

    work page 2015

  5. [5]

    2017, Astronomy & Astrophysics, 605, A100

    Bhardwaj, A., Rejkuba, M., Minniti, D., et al. 2017, Astronomy & Astrophysics, 605, A100

    work page 2017

  6. [6]

    M., Pihlstr¨ om, Y

    Bhattacharya, R., Medina, B. M., Pihlstr¨ om, Y. M., et al. 2024, ApJ, 969, 109, doi: 10.3847/1538-4357/ad463e

    work page doi:10.3847/1538-4357/ad463e 2024

  7. [7]

    C., et al

    Braga, V., Bhardwaj, A., Ramos, R. C., et al. 2018, Astronomy & Astrophysics, 619, A51

    work page 2018

  8. [8]

    2007, Astronomy & Astrophysics, 465, 825

    Cabrera-Lavers, A., Hammersley, P., Gonz´ alez-Fern´ andez, C., et al. 2007, Astronomy & Astrophysics, 465, 825

    work page 2007

Show all 38 references
  1. [9]

    M., Whitelock, P

    Catchpole, R. M., Whitelock, P. A., Feast, M. W., et al. 2016, Monthly Notices of the Royal Astronomical Society, 455, 2216

    2016

  2. [10]

    Chen, T., & Guestrin, C. 2016, in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’16 (New York, NY, USA: Association for Computing Machinery), 785–794, doi: 10.1145/2939672.2939785

    2016

  3. [11]

    2020, Astronomy & Astrophysics, 640, A23

    Chiavassa, A., Kravchenko, K., Millour, F., et al. 2020, Astronomy & Astrophysics, 640, A23

    2020

  4. [12]

    2006, Caltech, Pasadena

    Cutri, R., Skrutskie, M., Van Dyk, S., et al. 2006, Caltech, Pasadena

    2006

  5. [13]

    1999, The Midcourse Space Experiment Point Source Catalog Version 1.2 Explanatory Guide,, Technical Report, AD-A381933; AFRL-VS-TR-1999-1522

    Tedesco, E. 1999, The Midcourse Space Experiment Point Source Catalog Version 1.2 Explanatory Guide,, Technical Report, AD-A381933; AFRL-VS-TR-1999-1522

    1999

  6. [14]

    W., Glass, I., Whitelock, P., & Catchpole, R

    Feast, M. W., Glass, I., Whitelock, P., & Catchpole, R. 1989, Monthly Notices of the Royal Astronomical Society, 241, 375 Gaia Collaboration, Vallenari, A., Brown, A. G. A., et al. 2023, A&A, 674, A1, doi: 10.1051/0004-6361/202243940

    1989 doi

  7. [15]

    H., & Friedman, J

    Hastie, T., Tibshirani, R., Friedman, J. H., & Friedman, J. H. 2009, The elements of statistical learning: data mining, inference, and prediction, Vol. 2 (Springer) H¨ ofner, S., & Olofsson, H. 2018, A&A Rv, 26, 1, doi: 10.1007/s00159-017-0106-5

    2009 doi

  8. [16]

    2011, Astronomy & Astrophysics, 534, A79

    Ishihara, D., Kaneda, H., Onaka, T., et al. 2011, Astronomy & Astrophysics, 534, A79

    2011

  9. [17]

    2010, Astronomy & Astrophysics, 514, A1

    Ishihara, D., Onaka, T., Kataza, H., et al. 2010, Astronomy & Astrophysics, 514, A1

    2010

  10. [18]

    2023, ApJS, 264, 20, doi: 10.3847/1538-4365/acad7a Jim´ enez-Esteban, F., & Engels, D

    Iwanek, P., Poleski, R., Koz lowski, S., et al. 2023, ApJS, 264, 20, doi: 10.3847/1538-4365/acad7a Jim´ enez-Esteban, F., & Engels, D. 2015, Astronomy & Astrophysics, 579, A76

    2023 doi

  11. [19]

    Lewis, M. O. 2021, PhD thesis, The University of New Mexico

    2021

  12. [20]

    O., Bhattacharya, R., Sjouwerman, L

    Lewis, M. O., Bhattacharya, R., Sjouwerman, L. O., et al. 2023a, A&A, 677, A153, doi: 10.1051/0004-6361/202346568

    doi

  13. [21]

    O., Pihlstr¨ om, Y

    Lewis, M. O., Pihlstr¨ om, Y. M., Sjouwerman, L. O., et al. 2020, ApJ, 892, 52, doi: 10.3847/1538-4357/ab7920

    2020 doi

  14. [22]

    O., Pihlstr¨ om, Y

    Lewis, M. O., Pihlstr¨ om, Y. M., Sjouwerman, L. O., et al. 2020, The Astrophysical Journal, 892, 52 L´ opez Mart ´ ı, B., Jim´ enez-Esteban, F. M., Engels, D., & Garc ´ ıa-Lario, P. 2025, A&A, 698, A109, doi: 10.1051/0004-6361/202453125 Mart ´ ı, B. L., Jim´ enez-Esteban, F.,...

    2020 doi

  15. [23]

    2007, Publications of the Astronomical Society of Japan, 59, S369

    Murakami, H., Baba, H., Barthel, P., et al. 2007, Publications of the Astronomical Society of Japan, 59, S369

    2007

  16. [24]

    Sanders, J. L. 2023, Monthly Notices of the Royal Astronomical Society, 523, 2369

    2023

  17. [25]

    L., Matsunaga, N., Kawata, D., et al

    Sanders, J. L., Matsunaga, N., Kawata, D., et al. 2022, Monthly Notices of the Royal Astronomical Society, 517, 257

    2022

  18. [26]

    2017, Astronomy & Astrophysics, 600, A14

    Schultheis, M., Rojas-Arriagada, A., Garc ´ ıa P´ erez, A., et al. 2017, Astronomy & Astrophysics, 600, A14

    2017

  19. [28]

    O., Capen, S

    Sjouwerman, L. O., Capen, S. M., & Claussen, M. J. 2009, The Astrophysical Journal, 705, 1554, doi: 10.1088/0004-637X/705/2/1554

    2009 doi

  20. [29]

    C., Pihlstr¨ om, Y

    Stroh, M. C., Pihlstr¨ om, Y. M., Sjouwerman, L. O., et al. 2019b, The Astrophysical Journal Supplement Series, 244, 25, doi: 10.3847/1538-4365/ab3c35

    doi

  21. [30]

    2025, Universe, 11, 132

    Carraro, G. 2025, Universe, 11, 132

    2025

  22. [31]

    2018, The Astrophysical Journal, 861, 75

    Trapp, A., Rich, R., Morris, M., et al. 2018, The Astrophysical Journal, 861, 75

    2018

  23. [32]

    2020, The Astrophysical Journal, 891, 50

    Urago, R., Omodaka, T., Nagayama, T., et al. 2020, The Astrophysical Journal, 891, 50

    2020

  24. [33]

    L., Henden, A

    Watson, C. L., Henden, A. A., & Price, A. 2006, Society for Astronomical Sciences Annual Symposium, 25, 47

    2006

  25. [34]

    2015, Monthly Notices of the Royal Astronomical Society, 450, 4050

    Wegg, C., Gerhard, O., & Portail, M. 2015, Monthly Notices of the Royal Astronomical Society, 450, 4050

    2015

  26. [35]

    1992, The Center, Bulge and Disk of the Milky Way, 103 22

    Whitelock, P., & Catchpole, R. 1992, The Center, Bulge and Disk of the Milky Way, 103 22

    1992

  27. [36]

    1991, Monthly Notices of the Royal Astronomical Society (ISSN 0035-8711), vol

    Whitelock, P., Feast, M., & Catchpole, R. 1991, Monthly Notices of the Royal Astronomical Society (ISSN 0035-8711), vol. 248, Jan. 15, 1991, p. 276-312., 248, 276

    1991

  28. [37]

    A., Feast, M

    Whitelock, P. A., Feast, M. W., & Van Leeuwen, F. 2008, MNRAS, 386, 313, doi: 10.1111/j.1365-2966.2008.13032.x

    2008

  29. [38]

    A., Feast, M

    Whitelock, P. A., Feast, M. W., van Loon, J. T., & Zijlstra, A. A. 2003, MNRAS, 342, 86, doi: 10.1046/j.1365-8711.2003.06514.x

    2003

  30. [39]

    J., Zheng, X., & Wang, G

    Xu, S., Zhang, B., Reid, M. J., Zheng, X., & Wang, G. 2019, ApJ, 875, 114, doi: 10.3847/1538-4357/ab0e83

    2019 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.