Pith. sign in

REVIEW 4 major objections 6 minor 69 references

Automated Computer Vision Cluster Identification in the Fireworks Galaxy

T0 review · 4 major / 6 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read An automated edge-detection census of NGC 6946 finds its young star clusters follow a single steep power-law luminosity function with slope ≈2.2 in two UV bands, with no bright-end break.

desk verdict A large, honestly characterized cluster catalog for NGC 6946 with plausible LF slopes, but the completeness calibration is the main thing to probe in review. read the letter →

arxiv 2607.26330 v1 pith:SHU3SMJP submitted 2026-07-28 astro-ph.GA

Debby Tran , Benjamin F. Williams , Emily M. Levesque , Tobin M. Wainer , Emerson Bowles , Bo-Eun Choi , L. Clifton Johnson , Anil C. Seth This is my paper · ORCID
classification astro-ph.GA
keywords youngstarclustersNGC6946FireworksGalaxyluminosityfunctionautomatedclusterdetectionkerneldensityestimationedgeultravioletimaging
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

The authors set out to build a fast, repeatable way to find young star clusters in resolved ultraviolet images, and to use it to measure how many clusters exist at each brightness in the Fireworks Galaxy. They adapt a classic computer-vision edge-detection algorithm to work on maps of stellar density, then calibrate it with artificial clusters to set thresholds and measure completeness. Applied to NGC 6946, the method produces a catalog of 6,410 cluster candidates. For the brightest, most reliable subset, the luminosity function is a steep power law with slope about 2.2 in both F275W and F336W, and it continues as a power law down to about M=-6, suggesting no turnover at the bright end. If the completeness estimates hold, the young cluster population of this galaxy is described by one smooth power law, with consequences for how clusters form and disperse.

What carries the argument

The central object is a shape-adaptive cluster-detection pipeline: a stellar density map is built from resolved ultraviolet stellar positions via kernel density estimation, then unsharp-masked to subtract the diffuse background. The sharpened map is passed through an edge-detection chain—gradient computation, non-maximum suppression, and hysteresis thresholding with automatic threshold selection—to trace contours. A mean-shift clustering step and convex hull construction convert contours into polygonal cluster footprints. This replaces fixed circular apertures with data-driven outlines, and its thresholds are calibrated using 1,000 artificial clusters drawn from stellar evolution models and

What would settle it

A direct test would be to have expert classifiers visually identify young clusters in a sub-region of the same ultraviolet images and compare with the algorithm's candidates, measuring the actual false-positive rate and completeness near the claimed limit. Alternatively, re-running the insertion tests with less centrally concentrated or hierarchical synthetic cluster profiles would show whether the recovery rate and the inferred power-law slope change materially.

Watch

Extended reading notes

Core claim

The central claim is that the young (≤25 Myr) cluster population of NGC 6946 has a luminosity function consistent with a single power law of slope α = 2.26 ± 0.08 in F275W and 2.22 ± 0.07 in F336W, with no evidence for a break above -7.75 mag. This is established with a new automated cluster-finding method that combines kernel density estimation with edge detection to outline clusters non-parametrically, yielding 6,410 candidates. The authors argue the completeness is conservative, based on recovery of synthetic clusters, and that the power law extends at least one magnitude fainter than the reliable sample, based on the behavior of the luminosity function itself.

Load-bearing premise

The load-bearing premise is that the artificial clusters used to tune thresholds and measure completeness—built with a centrally concentrated density profile typical of old globular clusters—represent the real, more irregular young clusters well enough that the detection rates and completeness limits transfer; if real clusters are systematically less concentrated, the completeness limit and the inferred luminosity function slope could be wrong.

Editorial extensions

If this is right

  • If the slope α≈2.2 is correct, NGC 6946's young cluster luminosity function is steeper than those reported for several other galaxies in similar ultraviolet bands, implying galaxy-to-galaxy variation in cluster formation or disruption.
  • The absence of a bright-end break above -7.75 mag argues against a truncation in the young cluster mass function at the high-mass end in this galaxy at this age.
  • The quantitative recovery rate and false-positive rate from synthetic clusters provide a selection function, making the catalog usable for statistical studies of cluster demographics.
  • Because the method only needs resolved stellar positions and densities, it can be applied to other nearby galaxies with resolved ultraviolet catalogs, enabling uniform cluster censuses across environments.
  • The catalog of 6,410 candidates with integrated magnitudes, colors, and half-light radii provides a large sample for testing mass-radius, age-radius, and environmental correlations within a single galaxy.

Reading between the lines

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

  • The calibration uses synthetic clusters with a centrally concentrated density profile typical of old globular clusters; real young clusters tend to be less concentrated and more hierarchical, so the true completeness limit may be fainter than the conservative claim, which would reinforce the faint-end power law but could also shift the measured slope.
  • The steeper slope compared with past ultraviolet studies of other galaxies may stem from this method's selection function—for instance, favoring compact sub-structures within larger associations—rather than from a physical difference; a side-by-side comparison on the same galaxy would separate these.
  • Since the algorithm outlines cluster boundaries rather than imposing circular apertures, the same data could be used to correlate morphology (asymmetry, sub-structure) with age and environment, offering a route to test cluster disruption mechanisms.
  • A natural extension would be porting the pipeline to optical or infrared resolved catalogs, or to more distant galaxies where clusters are semi-resolved, to test whether the single-power-law result persists across wavelengths and environments.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper presents an automated cluster-candidate identification algorithm based on Canny edge detection applied to KDE stellar-density maps of resolved F275W/F336W photometry of NGC 6946. The algorithm identifies 6,410 young cluster candidates. The authors test the method with 756 synthetic King-profile clusters inserted into one region, report a recovery fraction of 60.7% and a 'conservative false positive rate of 27.3%', and derive a conservative 50% completeness limit near F275W~22 mag. They fit the luminosity function (LF) of a bright sample (M_F275W<-7.65, M_F336W<-7.9) and obtain slopes α=2.26±0.08 and 2.22±0.07, with no evidence for a break above M≈-7.75. They also fit steeper? shallower? slopes α~2.04-2.05 when extending to M~-6, and interpret the LF's continuation as evidence that the catalog is highly complete about one magnitude fainter than the conservative limit.

Significance. If the completeness and LF results are robust, the paper would provide a new, reproducible automated method for finding young clusters in semi-resolved UV imaging, plus a large public catalog of NGC 6946 cluster candidates with photometry and radii. The injection-testing framework is a genuine strength, and the public data products (catalog on MAST) are valuable. However, the current completeness calibration is not independent of the threshold tuning, the faint sample is explicitly unverified, and the central LF conclusions are drawn from data at or below the stated 50% completeness limit without a completeness correction. The method may still be useful as a candidate-finding tool, but the quantitative LF claims and the no-break conclusion are not yet established.

major comments (4)
  1. [Secs. 3.1, 3.2, 3.4] The completeness and recovery numbers are obtained from the same synthetic clusters used to tune the Canny thresholds: Sec. 3.2 states that the high/low thresholds are 'chosen to optimize for the number of synthetic clusters detected', and Sec. 3.4 then uses those same clusters to set the 50% completeness limit. This circularity means the recovery fraction is an optimistically biased estimate on the training set. The post-hoc exclusion of synthetic clusters with masses above 10^5.75 Msun (Sec. 3.1) removes precisely the bright, massive regime most relevant to the no-break LF claim. I recommend using a held-out set of synthetic clusters for completeness evaluation, including the bright models with a saturation treatment, and quantifying how threshold tuning biases the reported recovery and completeness.
  2. [Sec. 5, Fig. 9] The bright LF sample is M_F275W<-7.65 and M_F336W<-7.9. With distance modulus 29.4, the stated 50% completeness limits F275W~22 and F336W~21.5 correspond to M~-7.4 and M~-7.9, respectively. Thus the F275W fit starts only 0.25 mag brighter than the 50% completeness point, and the F336W fit starts essentially at it. The LF is fit without applying any completeness correction. If completeness varies within this magnitude range, the fitted slopes α=2.26±0.08 and 2.22±0.07 are biased, and the conclusion of no break above M≈-7.75 is not supported. The authors should either fit with a completeness function derived from the injection tests or restrict the fit to a brighter, demonstrably complete sample.
  3. [Secs. 3.4, 5] The claim that 'the luminosity function suggests that the catalog may be highly complete at least 1 magnitude fainter' is circular: it assumes the underlying LF is a single power law, which is exactly the hypothesis being tested. The extended fits down to M~-6 (α=2.04±0.03 and 2.05±0.03) are performed on data below the stated conservative completeness limit without correction. The absence of a turnover is interpreted as completeness, but an intrinsic break or flattening would produce the same appearance. An independent completeness test at these magnitudes (e.g., injecting more realistic clusters in multiple regions, or comparing with visual/other catalogs) is needed before these faint-end slopes or the no-break claim can be accepted.
  4. [Secs. 3.1, 3.3] The synthetic clusters are King profiles with tidal-to-core radius ratio 30, and the authors themselves note that young clusters tend to be less centrally concentrated than globular clusters. The injection test is also run on a single selected region, assumed representative of all environments. Therefore the recovery fractions and false-positive rate may not transfer to the irregular, hierarchical real young clusters. In addition, the false-positive definition in Sec. 3.3 counts real, non-injected clusters as false positives (96% of them), so the abstract's 'conservative false positive rate of 27.3%' is not a contamination rate for the real catalog. Please test with more realistic morphologies, use multiple regions, and report the rate of non-cluster artifacts separately from real clusters that were simply not part of the injection set.
minor comments (6)
  1. [Abstract/Sec. 3.3] The term 'false positive rate' is misleading given that 96% of these objects are potential real clusters. Please clarify in the abstract and text whether this is a missed-injection rate or an actual contamination rate.
  2. [Sec. 3.1] The text refers to 'F275TOT and F336TOT magnitudes' without defining these variables; presumably F275W and F336W integrated magnitudes. Please define or rephrase.
  3. [Sec. 3.5] The text says the output magnitude is 'roughly half the total integrated magnitude' and then gives fitted offsets of 0.11 and 0.41 mag. A factor of two in flux is 0.75 mag, so this wording is inconsistent with the stated offsets. Please clarify what is meant.
  4. [Sec. 5 / Fig. 9] The notation 'M_V=-6.25 in F336W' is confusing; it should read M_F336W. Also, 'Integrated Absolute Magnitude [mag]' on the axis label is nonstandard; use 'Absolute Magnitude' or 'M_F275W'/'M_F336W'.
  5. [Sec. 6] The summary bullet states 'conservative 50% completeness down to 10^3.25 Msun, dependent on age bin and 21.75 and 21.5 mag'. This conflates mass completeness and magnitude completeness; please state both separately and specify the age dependence.
  6. [Sec. 2] The abstract and intro mention ~81,000 resolved young massive stars from the FUVS catalog, while Sec. 6 says the algorithm input is roughly 300,000 sources. Clarify the relationship between these numbers and the SNR>=4 selection.

Circularity Check

2 steps flagged · score 5.0 of 10

Recovery/completeness are measured on the same synthetic clusters used to tune thresholds, and the faint-end completeness claim is inferred from the luminosity-function shape it is then used to measure; the bright-end slope itself is an independent fit.

  1. fitted input called prediction [Section 3.2 (Description of Algorithm) and Section 3.4 (Completeness)]
    "These thresholds are chosen to optimize for the number of synthetic clusters detected. ... We use the synthetic clusters from Section 3.1 to estimate the lower limit of the 50% completeness of the algorithm. ... Of the 756 inserted synthetic clusters, 463 were found by the algorithm."

    The Canny high/low thresholds are explicitly fitted to maximize recovery of the same 756 synthetic clusters that are then used to measure the recovery fraction and the 50% completeness limit. The reported performance is therefore the optimization target itself, not an independent prediction: recovery/completeness is forced by the fitting procedure rather than validated against an unseen sample.

  2. self definitional [Section 3.4 (Completeness) and Section 5 (Luminosity Function)]
    "Thus, while the results from the synthetic clusters in Section 3.1 show that our most reliable candidates are brightward of F275W∼22, the luminosity function suggests that the catalog may be highly complete at least 1 magnitude fainter (see Section 5 for details). ... Because the luminosity function of the catalog follows a consistent power-law to luminosities fainter than our most reliable sample, we also fit a slope down toM V =−6.25 in F336W. This result is 2.05±0.03 down toMV =−6 in F275W and 2.04±0.03, similar to the slope of the most-reliable sample, suggesting that our candidates catalo"

    The faint-end completeness/reliability is inferred from the power-law shape of the observed luminosity function, and the power-law slope fitted to that same faint sample is then presented as confirmation that the catalog is reliable there. The LF shape is used both as the evidence for completeness and as the result, so the faint-end slope is not an independent measurement; the reliability claim and the fitted slope are mutually supporting rather than separately established.

full rationale

The central bright-end LF slope (α≈2.26/2.22) is an empirical binned fit to the observed catalog, not a quantity derived from the synthetic clusters by construction, so the main slope measurement is not circular. However, two parts of the validation chain are. First, the Canny thresholds are explicitly set to maximize recovery of the 756 synthetic clusters, and the same sample is then used to report the recovery fraction and the 50% completeness limit; that performance number is the optimization target, so it cannot serve as an independent test. Second, the claim that the catalog is 'highly complete' one magnitude fainter is inferred from the power-law shape of the LF, and the faint-end power-law slope fitted to that same magnitude range is then cited as evidence that the sample is reliable; the completeness and the faint-end slope are mutually supporting rather than independently measured. The paper's own caveat that the synthetic King-profile clusters differ morphologically from real young clusters, and its explicit decision not to construct a completeness function, mean the no-break claim above M≈−7.75 is fragile; but that is a robustness issue, not circularity. Self-citations (Tran et al. 2022, 2023) provide algorithm and photometry details and are not load-bearing for the LF result.

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

The catalog and LF rest on standard stellar-population models, imaging algorithms, and a set of hand-tuned or synthetic-recovery-optimized parameters (thresholds, bandwidths, polygon cuts). No new physical entities are introduced. The most load-bearing external inputs are the representativeness of the King-profile synthetic clusters and the assumed extinction/distance.

free parameters (5)
  • Canny high/low threshold ratio = low=Otsu threshold; high=2x Otsu threshold
    Sec 3.2: 'These thresholds are chosen to optimize for the number of synthetic clusters detected.' Tuned on synthetic cluster recovery.
  • KDE bandwidths
    Two Gaussian bandwidths (narrow for density map, wide for unsharp masking); hand-selected. Sec 6 notes 'requires careful selection thresholds and bandwidth parameters.'
  • Minimum polygon stellar content = 4 high-quality stars
    Sec 3.2: 'we remove polygons containing less than 4 high-quality stars'—a hand-chosen cut affecting the catalog.
  • Synthetic cluster mass upper limit = 10^5.75 M_sun
    Sec 3.1: clusters more massive than 10^5.75 M_sun were excluded as 'too bright and saturate' and 'not representative of our data upon visual inspection'—a post hoc exclusion affecting completeness at the highest masses.
  • LF fitting magnitude limits = bright: M<-7.65 (F275W), M<-7.9 (F336W); faint: down to M=-6
    Sec 5: fitting ranges chosen where the LF appears power-law; the faint limit is justified by the LF flattening, i.e., by the data being fit.
assumptions (5)
  • standard math Canny edge detection, Otsu thresholding, mean-shift clustering, and convex-hull operations behave as documented in the cited computer-vision literature.
    The algorithm pipeline (Sec 3.2) relies on these standard algorithms without proof or re-derivation.
  • domain assumption PARSEC isochrones and the Kroupa IMF accurately represent the resolved UV stellar populations of young clusters in NGC 6946.
    Used to generate synthetic clusters (Sec 3.1); the realism of the injected clusters depends on these models.
  • domain assumption A King profile with tidal-to-core radius ratio of 30 is an adequate spatial model for the synthetic young clusters.
    Sec 3.1: 'young clusters tend to be less centrally concentrated than globular clusters, where the King profile is derived'; the authors acknowledge the mismatch but proceed.
  • domain assumption Schlafly & Finkbeiner (2011) extinction AV=0.938 and distance modulus 29.4 (7.8 Mpc) apply to NGC 6946.
    Used to populate synthetic clusters (Sec 3.1) and to convert photometry to absolute magnitudes (Sec 5).
  • ad hoc to paper The luminosity function of the bright reliable sample can be extrapolated to assess completeness of the fainter sample.
    Sec 3.4 and Sec 5: 'the luminosity function suggests that the catalog may be highly complete at least 1 magnitude fainter'; the faint-end slope is then fit to that same faint sample, a self-consistency argument.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Automated Computer Vision Cluster Identification in the Fireworks Galaxy." pith.science (2026). https://pith.science/paper/SHU3SMJP

@misc{pith2026260726330,
  author       = {Pith},
  title        = {Pith review of: Automated Computer Vision Cluster Identification in the Fireworks Galaxy},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SHU3SMJP}},
  note         = {Machine review of arXiv:2607.26330}
}
abstract

We present the integrated photometry, radii, and spatial distribution of young ($\leq$ 25 Myr) star cluster candidates in NGC 6946. NGC 6946, also known as the Fireworks Galaxy, is a highly star-forming galaxy with numerous young massive clusters. We have developed a modified computer vision algorithm using photometry from images taken with Hubble Space Telescope (HST) Wide Field Camera 3 Ultraviolet channel (WFC3/UVIS) F275W and F336W filters to identify and outline candidate clusters. We describe our technique in detail, including extensive testing with artificial clusters, where the algorithm recovers 60.7% of synthetic clusters and has a conservative false positive rate of 27.3% down to luminosities of M$_{F336W} \sim -6$. We identify 6410 cluster candidates down to much fainter magnitudes (M$_{F336W} \sim -4$) via the aforementioned algorithm which are more difficult to verify, but are still of interest as the luminosity function of these candidates is consistent with a standard power law with a slope of $\sim$2.

Figures

Figures reproduced from arXiv: 2607.26330 by the authors.

Figure 1
Figure 1. This four-panel plot illustrates the various input parameters for synthetic cluster creation. Top Left: Input masses ranging from 102.5 − 105.75M⊙; Top Right: Effective radius of clusters, between 1-8 parsecs; Bottom Right: Roughly flat age distribution from 106.6 − 107.5 ; Bottom Left: Integrated F275W and F336W magnitudes from 16-28 mag dataset using 2-D Gaussian kernel density estimation. This is further describe… view at source ↗
Figure 2
Figure 2. Top left: Centroids from FUVS photometry catalog; Top second from the left: Stellar density map from KDE; Top third from the left: Smoothed stellar density map; Top right: Unsharp masked stellar density map; Bottom left: Gradient of stellar density map; Bottom second from the left: Non-maximum suppression with interpolation to get edges to be one￾pixel width; Bottom third from the left: Double threshold hysteresis t… view at source ↗
Figure 3
Figure 3. The cyan circles are the artificial clusters inserted into the image, where the size of the circle indicates their effective radius. The pink polygons are artificial clusters identified by the algorithm. Cyan circles without an accompanying pink polygon are synthetic clusters that were missed by the algorithm. The green polygons are objects not inserted as synthetic clusters, but are identified by the algorithm. We … view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Left: Input magnitude of each synthetic cluster versus its output magnitude in both F275W and F336W. The output magnitude from integrating the counts within the edges with respect to the input magnitude is determined via the algorithm is modeled by a lines of best fit …
Figure 5
Figure 5. Figure 5: Plot of the algorithm’s radius calculation ver￾sus the residual between input and output half light radius. There is scatter in this residual plot up to 2 parsecs or 0.05 arcseconds at a distance of 7.8 Mpc. There is a slight corre￾lation between the residual and the a…
Figure 6
Figure 6. Figure 6: Comparison between the radius determined by the algorithm versus the input effective radius. Shown are a selection of synthetic clusters with a wide range of magnitude and radii. The cyan circles mark the input effective radius and the green polygons mark the edges det…
Figure 7
Figure 7. Figure 7: The spatial distribution of the cluster candidates detected overlaid on an F275W image of NGC 6946. A majority of these clusters are found in the spiral arms of NGC 6946. North is up and east is right. The color of each cluster indicates the unique cluster found. tribu…
Figure 8
Figure 8. Figure 8: Top Left: Histogram of integrated F275W magnitude of the detected cluster candidates, with median magnitude of 23.91 mag. Top Right: Histogram of integrated F336W magnitude of the detected cluster candidates, with median magnitude of 23.62 mag. Bottom Left: Histogram o…
Figure 9
Figure 9. Figure 9: Cluster luminosity functions of NGC 6946 in F275W and F336W, fit using absolute magnitudes, then converted to α using Equation 4, with the asterisk slopes denoting the fit down to where the luminosity function turns over. Triangle markers are the F275W bins, circle mar…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

69 extracted references · 18 canonical work pages

  1. [1]

    Adamo, A., Kruijssen, J. M. D., Bastian, N., Silva-Villa, E., & Ryon, J. 2015, MNRAS, 452, 246, doi: 10.1093/mnras/stv1203

    doi:10.1093/mnras/stv1203 2015

  2. [2]

    E., Messa, M., et al

    Adamo, A., Ryon, J. E., Messa, M., et al. 2017, ApJ, 841, 131, doi: 10.3847/1538-4357/aa7132 NGC 6946 Clusters I13 12 11 10 9 8 7 6 5 4 Integrated Absolute Magnitude [mag] 1.5 2.0 2.5 3.0 3.5 4.0 4.5log(dN/dM)+const F275W = 2.26 ± 0.08 F336W = 2.22 ± 0.07 F275W * = 2.04 ± 0.03 F336W * = 2.05 ± 0.03 F275W F336W Figure 9.Cluster luminosity functions of NGC ...

    doi:10.3847/1538-4357/aa7132 2017

  3. [3]

    S., Rizzi, L., & Tully, R

    Anand, G. S., Rizzi, L., & Tully, R. B. 2018, AJ, 156, 105, doi: 10.3847/1538-3881/aad3b2 Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. 2013, A&A, 558, A33, doi: 10.1051/0004-6361/201322068 Astropy Collaboration, Price-Whelan, A. M., Sip˝ ocz, B. M., et al. 2018, AJ, 156, 123, doi: 10.3847/1538-3881/aabc4f Astropy Collaboration, Price-...

    doi:10.3847/1538-3881/aad3b2 2018

  4. [4]

    L., van Zee, L., Cˆ ot´ e, S., & Schade, D

    Barnes, K. L., van Zee, L., Cˆ ot´ e, S., & Schade, D. 2012, ApJ, 757, 64, doi: 10.1088/0004-637X/757/1/64

    work page doi:10.1088/0004-637x/757/1/64 2012

  5. [5]

    2007, MNRAS, 379, 1302, doi: 10.1111/j.1365-2966.2007.12064.x

    Bastian, N., Ercolano, B., Gieles, M., et al. 2007, MNRAS, 379, 1302, doi: 10.1111/j.1365-2966.2007.12064.x

    arXiv 2007

  6. [6]

    S., Trancho, G., et al

    Bastian, N., Konstantopoulos, I. S., Trancho, G., et al. 2012a, A&A, 541, A25, doi: 10.1051/0004-6361/201219079

    work page doi:10.1051/0004-6361/201219079

  7. [7]

    2011, MNRAS, 417, L6, doi: 10.1111/j.1745-3933.2011.01103.x —

    Bastian, N., Adamo, A., Gieles, M., et al. 2011, MNRAS, 417, L6, doi: 10.1111/j.1745-3933.2011.01103.x —. 2012b, MNRAS, 419, 2606, doi: 10.1111/j.1365-2966.2011.19909.x

    arXiv 2011

  8. [8]

    C., Hubeny, I., & Rauch, T

    Bohlin, R. C., Hubeny, I., & Rauch, T. 2020, The Astronomical Journal, 160, 21, doi: 10.3847/1538-3881/ab94b4

    doi:10.3847/1538-3881/ab94b4 2020

Show all 69 references
  1. [9]

    2000, Dr

    Bradski, G. 2000, Dr. Dobb’s Journal of Software Tools

    2000

  2. [10]

    Brown, G., & Gnedin, O. Y. 2021, MNRAS, 508, 5935, doi: 10.1093/mnras/stab2907 14

    2021 doi

  3. [11]

    1986, IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-8, 679, doi: 10.1109/TPAMI.1986.4767851

    Canny, J. 1986, IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-8, 679, doi: 10.1109/TPAMI.1986.4767851

    1986

  4. [12]

    Cartwright, A., & Whitworth, A. P. 2004, Monthly Notices of the Royal Astronomical Society, 348, 589, doi: 10.1111/j.1365-2966.2004.07360.x

    2004

  5. [13]

    C., Kim, H., et al

    Chandar, R., Whitmore, B. C., Kim, H., et al. 2010, ApJ, 719, 966, doi: 10.1088/0004-637X/719/1/966

    2010 doi

  6. [14]

    Choksi, N., & Kruijssen, J. M. D. 2021, MNRAS, 507, 5492, doi: 10.1093/mnras/stab2514

    2021 doi

  7. [15]

    2002, IEEE Transactions on Pattern Analysis and Machine Intelligence, 24, 603, doi: 10.1109/34.1000236

    Comaniciu, D., & Meer, P. 2002, IEEE Transactions on Pattern Analysis and Machine Intelligence, 24, 603, doi: 10.1109/34.1000236

    2002 doi

  8. [16]

    O., Lee, J

    Cook, D. O., Lee, J. C., Adamo, A., et al. 2019, Monthly Notices of the Royal Astronomical Society, 484, 4897, doi: 10.1093/mnras/stz331

    2019 doi

  9. [17]

    2019a, A&A, 631, A128, doi: 10.1051/0004-6361/201936409 —

    Costa, G., Girardi, L., Bressan, A., et al. 2019a, A&A, 631, A128, doi: 10.1051/0004-6361/201936409 —. 2019b, MNRAS, 485, 4641, doi: 10.1093/mnras/stz728

    doi

  10. [18]

    Elmegreen, B. G. 2018, ApJ, 853, 88, doi: 10.3847/1538-4357/aaa252

    2018 doi

  11. [19]

    C., Weisz, D

    Fouesneau, M., Johnson, L. C., Weisz, D. R., et al. 2014, ApJ, 786, 117, doi: 10.1088/0004-637X/786/2/117 Gaia Collaboration, Prusti, T., de Bruijne, J. H. J., et al. 2016, A&A, 595, A1, doi: 10.1051/0004-6361/201629272 Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al...

    2014 doi

  12. [20]

    2009, MNRAS, 394, 2113, doi: 10.1111/j.1365-2966.2009.14473.x

    Gieles, M. 2009, MNRAS, 394, 2113, doi: 10.1111/j.1365-2966.2009.14473.x

    2009

  13. [21]

    Gieles, M., Lamers, H. J. G. L. M., & Portegies Zwart, S. F. 2007, ApJ, 668, 268, doi: 10.1086/520795

    2007 doi

  14. [22]

    S., Bastian, N., & Stein, I

    Gieles, M., Larsen, S. S., Bastian, N., & Stein, I. T. 2006, A&A, 450, 129, doi: 10.1051/0004-6361:20053589

    2006 doi

  15. [23]

    A., Schmeja, S., Klessen, R

    Gouliermis, D. A., Schmeja, S., Klessen, R. S., de Blok, W. J. G., & Walter, F. 2010, ApJ, 725, 1717, doi: 10.1088/0004-637X/725/2/1717

    2010 doi

  16. [24]

    A., Elmegreen, B

    Gouliermis, D. A., Elmegreen, B. G., Elmegreen, D. M., et al. 2017, MNRAS, 468, 509, doi: 10.1093/mnras/stx445 Grudi´ c, M. Y., Kruijssen, J. M. D., Faucher-Gigu` ere, C.-A., et al. 2021, MNRAS, 506, 3239, doi: 10.1093/mnras/stab1894

    2017 doi

  17. [25]

    R., Gieles, M., Scheepmaker, R

    Haas, M. R., Gieles, M., Scheepmaker, R. A., Larsen, S. S., & Lamers, H. J. G. L. M. 2008, A&A, 487, 937, doi: 10.1051/0004-6361:20078831

    2008 doi

  18. [26]

    A., Schult, D

    Hagberg, A. A., Schult, D. A., & Swart, P. J. 2008, in Proceedings of the 7th Python in Science Conference, ed. G. Varoquaux, T. Vaught, & J. Millman, Pasadena, CA USA, 11 – 15

    2008

  19. [27]

    R., Millman, K

    Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357, doi: 10.1038/s41586-020-2649-2

    2020 doi

  20. [28]

    A., Faber, S

    Holtzman, J. A., Faber, S. M., Shaya, E. J., et al. 1992, AJ, 103, 691, doi: 10.1086/116094

    1992 doi

  21. [29]

    Hunter, J. D. 2007, Computing in Science and Engineering, 9, 90, doi: 10.1109/MCSE.2007.55

    2007 doi

  22. [30]

    R., Koplitz, B., Williams, B

    Johnson, J. R., Koplitz, B., Williams, B. F., et al. 2023, ApJ, 945, 108, doi: 10.3847/1538-4357/acb775

    2023 doi

  23. [31]

    C., Seth, A

    Johnson, L. C., Seth, A. C., Dalcanton, J. J., et al. 2012, ApJ, 752, 95, doi: 10.1088/0004-637X/752/2/95 —. 2015, ApJ, 802, 127, doi: 10.1088/0004-637X/802/2/127 —. 2017, ApJ, 839, 78, doi: 10.3847/1538-4357/aa6a1f

    2012 doi

  24. [32]

    C., Wainer, T

    Johnson, L. C., Wainer, T. M., Torresvillanueva, E. E., et al. 2022, ApJ, 938, 81, doi: 10.3847/1538-4357/ac8def

    2022 doi

  25. [33]

    V., Fleischmann, M., et al

    Jordahl, K., den Bossche, J. V., Fleischmann, M., et al. 2020, doi: 10.5281/zenodo.3946761

    2020 doi

  26. [34]

    1962, AJ, 67, 471, doi: 10.1086/108756

    King, I. 1962, AJ, 67, 471, doi: 10.1086/108756

    1962 doi

  27. [35]

    2001, MNRAS, 322, 231, doi: 10.1046/j.1365-8711.2001.04022.x

    Kroupa, P. 2001, MNRAS, 322, 231, doi: 10.1046/j.1365-8711.2001.04022.x

    2001

  28. [36]

    R., McKee, C

    Krumholz, M. R., McKee, C. F., & Bland-Hawthorn, J. 2019, ARA&A, 57, 227, doi: 10.1146/annurev-astro-091918-104430

    2019 doi

  29. [37]

    Larsen, S. S. 2002, AJ, 124, 1393, doi: 10.1086/342381

    2002 doi

  30. [38]

    Larsen, S. S. 2006, in Planets to Cosmology: Essential Science in the Final Years of the Hubble Space Telescope, ed. M. Livio & S. Casertano, Vol. 18, 35 —. 2009, A&A, 494, 539, doi: 10.1051/0004-6361:200811212

    2006 doi

  31. [39]

    L., D ´ ıaz-Santos, T., Armus, L., et al

    Larson, K. L., D ´ ıaz-Santos, T., Armus, L., et al. 2020, The Astrophysical Journal, 888, 92, doi: 10.3847/1538-4357/ab5dc3

    2020 doi

  32. [40]

    L., Lee, J

    Larson, K. L., Lee, J. C., Thilker, D. A., et al. 2023, MNRAS, 523, 6061, doi: 10.1093/mnras/stad1600

    2023 doi

  33. [41]

    2019, A Complete Inventory of the Fireworks Galaxy’s O-Type

    Levesque, E., Dalcanton, J., Khan, R., et al. 2019, A Complete Inventory of the Fireworks Galaxy’s O-Type

    2019

  34. [42]

    Cycle 27, ID

    Stars, HST Proposal. Cycle 27, ID. #15877 Ma ´ ız Apell´ aniz, J., &´Ubeda, L. 2005, ApJ, 629, 873, doi: 10.1086/431458

    2005 doi

  35. [43]

    2018, Vision Research, 153, 30, doi: https://doi.org/10.1016/j.visres.2018.09.007

    McIlhagga, W. 2018, Vision Research, 153, 30, doi: https://doi.org/10.1016/j.visres.2018.09.007

    2018 doi

  36. [44]

    2010, in Proceedings of the 9th Python in Science Conference, ed

    McKinney, W. 2010, in Proceedings of the 9th Python in Science Conference, ed. St´ efan van der Walt & Jarrod Millman, 56 – 61, doi: 10.25080/Majora-92bf1922-00a

    2010 doi

  37. [45]

    2018, MNRAS, 473, 996, doi: 10.1093/mnras/stx2403

    Messa, M., Adamo, A., ¨Ostlin, G., et al. 2018, MNRAS, 473, 996, doi: 10.1093/mnras/stx2403

    2018 doi

  38. [46]

    W., Khan, R., Williams, B., et al

    Murphy, J. W., Khan, R., Williams, B., et al. 2018, ApJ, 860, 117, doi: 10.3847/1538-4357/aac2be

    2018 doi

  39. [47]

    T., Costa, G., Girardi, L., et al

    Nguyen, C. T., Costa, G., Girardi, L., et al. 2022, A&A, 665, A126, doi: 10.1051/0004-6361/202244166

    2022 doi

  40. [48]

    1979, IEEE Transactions on Systems, Man, and Cybernetics, 9, 62, doi: 10.1109/TSMC.1979.4310076 pandas development team, T

    Otsu, N. 1979, IEEE Transactions on Systems, Man, and Cybernetics, 9, 62, doi: 10.1109/TSMC.1979.4310076 pandas development team, T. 2020, doi: 10.5281/zenodo.3509134 NGC 6946 Clusters I15

    1979

  41. [49]

    2011, Journal of Machine Learning Research, 12, 2825 Rodr ´ ıguez, M

    Pedregosa, F., Varoquaux, G., Gramfort, A., et al. 2011, Journal of Machine Learning Research, 12, 2825 Rodr ´ ıguez, M. J., Baume, G., & Feinstein, C. 2019, A&A, 626, A35, doi: 10.1051/0004-6361/201935291

    2011 doi

  42. [50]

    1976, ApJ, 203, 297, doi: 10.1086/154079

    Schechter, P. 1976, ApJ, 203, 297, doi: 10.1086/154079

    1976 doi

  43. [51]

    F., & Finkbeiner, D

    Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ, 737, 103, doi: 10.1088/0004-637X/737/2/103

    2011 doi

  44. [52]

    2011, Astronomische Nachrichten, 332, 172, doi: 10.1002/asna.201011484

    Schmeja, S. 2011, Astronomische Nachrichten, 332, 172, doi: 10.1002/asna.201011484

    2011 doi

  45. [53]

    1968, A 3x3 Isotropic Gradient Operator for Image Processing

    Sobel, I., & Feldman, G. 1968, A 3x3 Isotropic Gradient Operator for Image Processing

    1968

  46. [54]

    A., Whitmore, B

    Thilker, D. A., Whitmore, B. C., Lee, J. C., et al. 2022, MNRAS, 509, 4094, doi: 10.1093/mnras/stab3183

    2022 doi

  47. [55]

    E., Kadlec, K., et al

    Tran, D., Tuttle, S. E., Kadlec, K., et al. 2022, Journal of Astronomical Telescopes, Instruments, and Systems, 8, 045004, doi: 10.1117/1.JATIS.8.4.045004

    2022 doi

  48. [56]

    2023, ApJ, 954, 211, doi: 10.3847/1538-4357/aced44

    Tran, D., Williams, B., Levesque, E., et al. 2023, ApJ, 954, 211, doi: 10.3847/1538-4357/aced44

    2023 doi

  49. [57]

    A., Dale, D

    Turner, J. A., Dale, D. A., Lee, J. C., et al. 2021, MNRAS, 502, 1366, doi: 10.1093/mnras/stab055 van der Walt, S., Sch¨ onberger, J. L., Nunez-Iglesias, J., et al. 2014, PeerJ, 2, e453, doi: 10.7717/peerj.453

    2021 doi

  50. [58]

    E., et al

    Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature Methods, 17, 261, doi: 10.1038/s41592-019-0686-2

    2020 doi

  51. [59]

    M., Johnson, L

    Wainer, T. M., Johnson, L. C., Seth, A. C., et al. 2022, arXiv e-prints, arXiv:2201.04161. https://arxiv.org/abs/2201.04161

    2022 arXiv

  52. [60]

    Waskom, M. L. 2021, Journal of Open Source Software, 6, 3021, doi: 10.21105/joss.03021

    2021 doi

  53. [61]

    Whitmore, B. C. 2003, in A Decade of Hubble Space Telescope Science, ed. M. Livio, K. Noll, & M. Stiavelli, Vol. 14, 153–178

    2003

  54. [62]

    C., Chandar, R., Bowers, A

    Whitmore, B. C., Chandar, R., Bowers, A. S., et al. 2014, AJ, 147, 78, doi: 10.1088/0004-6256/147/4/78

    2014 doi

  55. [63]

    C., Schweizer, F., Kundu, A., & Miller, B

    Whitmore, B. C., Schweizer, F., Kundu, A., & Miller, B. W. 2002, AJ, 124, 147, doi: 10.1086/340808

    2002 doi

  56. [64]

    1993, AJ, 106, 1354, doi: 10.1086/116732

    Robert, C. 1993, AJ, 106, 1354, doi: 10.1086/116732

    1993 doi

  57. [65]

    C., Zhang, Q., Leitherer, C., et al

    Whitmore, B. C., Zhang, Q., Leitherer, C., et al. 1999, AJ, 118, 1551, doi: 10.1086/301041

    1999 doi

  58. [66]

    C., Chandar, R., Schweizer, F., et al

    Whitmore, B. C., Chandar, R., Schweizer, F., et al. 2010, AJ, 140, 75, doi: 10.1088/0004-6256/140/1/75

    2010 doi

  59. [67]

    F., Lang, D., Dalcanton, J

    Williams, B. F., Lang, D., Dalcanton, J. J., et al. 2014, Astrophysical Journal, Supplement Series, 215, doi: 10.1088/0067-0049/215/1/9

    2014 doi

  60. [68]

    N., et al

    Yadav, J., Das, M., Patra, N. N., et al. 2021, ApJ, 914, 54, doi: 10.3847/1538-4357/abf8c1

    2021 doi

  61. [69]

    Zhang, Q., & Fall, S. M. 1999, ApJL, 527, L81, doi: 10.1086/312412

    1999 doi

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.