Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

Chemical segregation analysed with unsupervised clustering

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Density-based clustering of molecular-line pixels, with spatial coordinates removed, reveals a chemical segregation between c-C3H2 and CH3CCH in all three cores that is not visible in the emission maps.

desk verdict The method demonstration is solid as a positive control, but the claimed new c-C3H2/CH3CCH segregation is likely an artifact of duplicated transitions per pixel and does not survive scrutiny. read the letter →

arxiv 2507.22482 v1 pith:ELOBPSEN submitted 2025-07-30 astro-ph.GA

classification astro-ph.GA
keywords chemicalsegregationdensecoresDBSCANHDBSCANc-C3H2CH3CCHmolecularlineemissionH2columndensitygradient
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 paper sets out to show that unsupervised density-based clustering (DBSCAN and HDBSCAN) can uncover chemical structure in dense cores from very small molecular datasets. Each emission pixel is described by its physical environment—integrated intensity, velocity offset, linewidth, H2 column density, distance to the dust peak, and column-density gradient—rather than by its position on the sky. The analysis reproduces the known segregation between c-C3H2 and CH3OH in B68, L1521E, and L1544, and it further claims a new segregation between c-C3H2 and CH3CCH in all three cores, a separation that is not apparent from the emission maps themselves. The authors interpret this as evidence that c-C3H2 traces lower-density outer gas while CH3OH and CH3CCH trace inner or accreting gas, and they use the clustering plus chemical modelling to argue that the CH3CCH peak in L1544 marks the landing point of chemically fresh accreted material.

What carries the argument

The machinery is density-based clustering—DBSCAN, which groups points whose mutual distance is below a threshold epsilon and labels smaller groupings as noise, and HDBSCAN, its hierarchical extension that handles clusters of varying density. The input is constructed by removing spatial coordinates and representing each observed pixel by six physical features: integrated intensity, velocity offset from the source's systemic velocity, linewidth, H2 column density, projected distance to the dust peak, and the magnitude of the H2 column-density gradient. This feature-space representation lets the same analysis run across cores of different sizes and locations, and the clusters are then read by looking at which molecule dominates each imbalanced cluster. Chemical simulations of the CH3CCH formation and destruction network (formation via dissociative recombination of C3H5+ and destruction by atomic carbon) provide the mechanism invoked to explain why the molecule survives at the L1544 accretion landing point.

What would settle it

The decisive test is a clustering rerun on B68 and L1521E with one data point per molecule per pixel, for example averaging the CH3CCH transitions before building the feature space; if the previously imbalanced clusters persist, the segregation is chemical, and if they vanish, it is an artifact of duplicated points.

Watch

Extended reading notes

Core claim

At the paper's center is a claim about observable chemistry: in the starless cores B68 and L1521E and the prestellar core L1544, the carbon-chain molecules c-C3H2 and CH3CCH occupy different physical layers even where their projected emission maps overlap. The argument is made by combining both molecules without labels in a feature space built from each pixel's intensity, velocity offset, linewidth, H2 column density, distance to the dust peak, and column-density gradient; clusters that come out imbalanced in one molecule correspond to regions of chemical segregation. This approach reproduces the already known c-C3H2/CH3OH segregation, validates the method, and then reveals a c-C3H2/CH3CCH segregation in all three cores. The paper also reports that CH3OH and CH3CCH cluster similarly relative to c-C3H2, that the most informative features are intensity, velocity offset, and column-density-related quantities, and that in L1544 the CH3CCH peak sits at a shielded northwest location where fresh accreted gas can form CH3CCH before atomic carbon destroys it. Abundance measurements at dust peaks add an evolutionary reading: CH3CCH is about one order of magnitude more abundant in starless than in prestellar cores, with L1544 as the exception.

Load-bearing premise

The result depends on treating each spectral transition of a molecule as an independent sample, which in B68 and L1521E duplicates CH3CCH at the same pixels and can create clusters through point density rather than physical chemistry—a choice the paper flags but does not correct.

Editorial extensions

If this is right

  • Dense cores are chemically layered in a way that single-molecule maps obscure: c-C3H2 traces a lower-density outer shell while CH3CCH and CH3OH concentrate in inner or freshly accreted gas.
  • Chemical-differentiation studies need not wait for large line surveys; a handful of targeted molecular transitions can expose segregation with density-based clustering.
  • Because velocity offset dominates the cluster splits, static chemical models are inadequate for anisotropic chemical structures, and dynamical models including accretion must be used.
  • CH3CCH abundance, about one order of magnitude lower in prestellar than starless cores except at L1544, can serve as a combined probe of evolutionary stage and local accretion environment.
  • Using the H2 column density gradient as a feature in place of sky coordinates lets cores in different environments be compared directly, and the varying relevance of this feature across cores reflects different illumination conditions.

Reading between the lines

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

  • A natural robustness check: repeat the clustering with one data point per molecule per pixel in B68 and L1521E to test whether the new c-C3H2/CH3CCH segregation survives removal of duplicated transitions; the paper acknowledges the duplication but does not perform this test.
  • The same feature-space recipe could be applied to archival multi-molecule maps of other cores to hunt for hidden segregation wherever projected emission overlaps, giving a census of chemical layers across many clouds.
  • If the accretion interpretation holds, c-C3H2-to-CH3CCH or c-C3H2-to-CH3OH ratio maps might serve as a practical tracer of inflowing, chemically fresh gas, complementing kinematic methods.
  • The prominence of velocity offset suggests a testable dynamical prediction: synthetic observations from collapsing-core models with time-dependent chemistry should reproduce the same cluster separations only when accretion and photochemistry are included.
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

3 major / 6 minor

Summary. The paper applies the density-based clustering algorithms DBSCAN and HDBSCAN to IRAM 30 m observations of c-C3H2, CH3OH, and CH3CCH toward the starless cores B68 and L1521E and the prestellar core L1544. The spatial coordinates of each emission pixel are discarded and replaced by six physical features (integrated intensity, velocity offset, linewidth, H2 column density, distance to the dust peak, and H2 column density gradient). Four case studies are considered: c-C3H2 versus CH3OH (Case 1), c-C3H2 versus CH3CCH (Case 2), CH3OH versus CH3CCH (Case 3), and all three molecules together (Case 4). The analysis reproduces the known c-C3H2/CH3OH segregation in all three cores and, as the main new result, claims to identify a segregation between c-C3H2 and CH3CCH that is not apparent from the emission maps. The authors also measure CH3CCH abundances at the dust peaks of several cores and use a gas-grain chemical model to argue that the CH3CCH peak in L1544 traces the landing point of freshly accreted gas.

Significance. If the central new claim is robust, the paper demonstrates that density-based clustering on small molecular datasets can uncover subtle chemical segregation, and the c-C3H2/CH3CCH segregation in the less-evolved cores B68 and L1521E would be an observationally new result with implications for chemical layering and accretion flows. The successful reproduction of the previously known c-C3H2/CH3OH segregation serves as a useful positive control, and the measured evolutionary trend of CH3CCH abundances from starless to prestellar cores is an interesting addition. The authors also make their detailed clustering figures publicly available on Zenodo, which improves reproducibility. However, the principal new claim rests on a sampling assumption that is acknowledged but not tested, and the imbalance criterion used throughout lacks a statistical significance test; these issues need to be resolved before the result can be accepted.

major comments (3)
  1. [§3.3, Table 4; §5.1] The central new claim of a c-C3H2/CH3CCH segregation in B68 and L1521E is not established because Case 2 includes two or three CH3CCH transitions per spatial pixel while c-C3H2 contributes only one point per pixel. Since DBSCAN and HDBSCAN are density-based, these duplicated CH3CCH points inflate the local density in feature space at CH3CCH-bright positions, biasing cluster formation toward those regions. The paper acknowledges this effect in §5.1 but does not provide an equal-per-pixel control (e.g., using a single representative transition per molecule, or averaging transitions, or down-weighting duplicated points). Until such a control is performed, the reported segregation and the interpretation that CH3CCH traces an inner layer while c-C3H2 traces an outer shell are not robust against this sampling artifact.
  2. [§4.3, Table C.1] The definition of an 'imbalanced' cluster as a deviation of at least 10% from the initial molecular ratio is not accompanied by any significance test. Several clusters contain very few points (N of order 6-10; e.g., Table C.1, combi 2, L1521E, cluster 2 and combi 8, B68, cluster 4), where a 10% shift can be consistent with Poisson counting noise. The paper should either restrict the imbalance designation to clusters whose ratios differ significantly from the input ratio (e.g., via a chi-square or permutation test) or state the expected binomial scatter around the input ratio. This matters because the qualitative summaries in §4.3.2 and the final conclusions on molecular segregation rely on these labels.
  3. [§3.1, §3.3] The clustering treats each emission-pixel sample as an independent data point even though adjacent pixels are strongly correlated given the 8 arcsec pixel size and 32 arcsec beam, and it treats multiple transitions of the same molecule as independent feature vectors. This pseudo-replication can affect the local density estimates that DBSCAN and HDBSCAN rely on, particularly when transitions of one molecule are duplicated in Case 2 and Case 3. The manuscript should discuss this limitation explicitly and, ideally, include a test in which the data are smoothed or thinned so that each independent beam is represented by a single sample per molecule.
minor comments (6)
  1. [Abstract, §5.1] The abstract and conclusions list integrated intensity, velocity offset, H2 column density, and H2 column density gradient as the key features driving the clustering, but the paper does not present a quantitative feature-relevance analysis; this claim appears to be based on visual inspection of two-dimensional feature projections. Please clarify the basis for this statement or add a feature-importance measure.
  2. [§4.4] The excitation temperature is fixed at 8 K for all cores, and the statement that a lower or higher excitation temperature only shifts the abundances without changing the overall trend is not quantified. Please provide the magnitude of the shift for a reasonable range (e.g., 5-10 K) to support the robustness claim.
  3. [Table 2, Fig. A.1] The transition labels are inconsistent: Table 2 lists CH3OH 20,2-10,1 (E2) and 21,2-11,1 (A+), while Fig. A.1 labels the maps '202-101E0' and '212-111E2'. Please harmonize the notation.
  4. [Fig. 2 caption] The caption of Fig. 2 states that dashed contours represent 30%, 50%, and 90% of the H2 column density peak, but the citation to Spezzano et al. (2020) appears after the period in an awkward way. Please fix the sentence structure.
  5. [Table 4] Table 4 gives the molecular ratios as percentages but does not list the total number of data points per dataset. Adding N would help the reader evaluate the statistical weight of the clusters shown in Table C.1.
  6. [§3.3] For Case 4, the text says the dataset combines Case 1 and Case 2, but it is not stated explicitly which transitions are used for each molecule in that combined dataset. Please specify this to avoid ambiguity, especially because the number of CH3CCH transitions differs between cores.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; the main caveat is an acknowledged data-density imbalance, not a circular reduction.

full rationale

The claimed derivation chain is: line maps are converted to per-pixel physical features, density-based clustering is run without molecule labels, and cluster molecular ratios are interpreted as chemical segregation. No step is equivalent to its inputs by construction. Case 1 (c-C3H2 vs CH3OH) uses one transition per molecule and reproduces the previously published segregation (Spezzano et al. 2016, 2020) as validation, not as an input. In L1544, Case 2 also uses one transition per molecule, so the new c-C3H2 vs CH3CCH segregation there is not affected by the duplication issue. For B68 and L1521E, Table 4 and Section 5.1 explicitly state that two or three CH3CCH transitions were included to balance the number of data points, which increases the local density of CH3CCH points and the likelihood of forming clusters in CH3CCH-bright regions; this is a real sampling-bias caveat requiring an equal-per-pixel control, and the paper itself acknowledges it. However, this is not circularity: the cluster compositions and spatial patterns are not equal to the input duplication by construction, and no fitted parameter is later relabelled as a prediction. The chemical modelling (pyRate plus the KIDA network) is independent of the clustering results. Self-citations are to data sources and earlier observational results; none is invoked as a uniqueness theorem or as an ansatz that forces the present conclusions. Overall, no significant circularity.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The paper introduces no new physical entities or fitted physical constants. The main assumptions are domain-specific: single velocity components, optically thin lines, reliability of Herschel column densities, validity of the chemical network and physical model, and the ad hoc treatment of multiple transitions as independent samples. The latter is the most consequential because it directly biases the clustering results.

free parameters (4)
  • Assumed excitation temperature Tex = 8 K
    Adopted for all cores when converting line intensities to column densities in Sec. 4.4. Authors note that a different Tex shifts abundances but not the trend, so it is a nuisance parameter rather than a fitted constant.
  • Imbalance threshold = 10%
    Clusters with a molecular ratio deviating by 10% or more from the input ratio are deemed 'imbalanced' in Sec. 4.3. The threshold is chosen by hand and has no statistical justification.
  • Clustering hyperparameter grid ranges = epsilon 0.05-0.155; min_samples 10-100; min_cluster_size 5-20
    Grid search ranges for DBSCAN and HDBSCAN in Sec. 3.2. The selection criteria (2-5 clusters, at least 50% coverage) are post hoc and used to pick the best result, introducing subjective choices.
  • H2 column density gradient kernel width = 2 telescope beams (64 arcsec)
    Used to compute the N(H2) gradient in Sec. 4.1 following Soler et al. (2013). The choice could affect the feature values and thus the clustering, but it is not varied or tested.
assumptions (6)
  • domain assumption The molecular emission lines are optically thin enough for Gaussian fitting and column density derivation.
    Used throughout; for CH3CCH the authors select the most optically thin transitions but do not explicitly verify optical depths. See Secs. 3.1 and 4.4.
  • domain assumption Each core has a single velocity component along the line of sight.
    Stated in Sec. 3.1, justifying the one-dimensional Gaussian fit per pixel and the use of velocity offset as a feature.
  • domain assumption The H2 column density maps from Herschel SPIRE reliably trace the core structure and the dust peak.
    Used as a feature and to define the dust peak (Sec. 3.1), sourced from Spezzano et al. (2016, 2020). Any systematic error in these maps would propagate into the clustering features.
  • domain assumption The KIDA 2014 gas-phase network and the Semenov et al. (2010) grain-surface network correctly model CH3CCH chemistry.
    Used in Sec. 5.2 to support the proposed formation and destruction pathways for CH3CCH. The model is not validated against the observed abundance in detail.
  • domain assumption The physical model of L1544 from Keto et al. (2015) is a valid background for the chemical simulation.
    Used in Sec. 5.2 to run the pyRate chemical model. If the physical structure is incorrect, the inferred chemistry would be affected.
  • ad hoc to paper Multiple observed transitions of the same molecule can be treated as independent data points in the clustering input.
    Used in Cases 2-4 to balance data point counts (Table 4). This is questionable because duplicate pixels bias density-based clustering. The authors acknowledge the bias but do not correct it.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Chemical segregation analysed with unsupervised clustering." pith.science (2026). https://pith.science/paper/ELOBPSEN

@misc{pith2026250722482,
  author       = {Pith},
  title        = {Pith review of: Chemical segregation analysed with unsupervised clustering},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ELOBPSEN}},
  note         = {Machine review of arXiv:2507.22482}
}
abstract

Molecular emission is a powerful tool for studying the physical and chemical structures of dense cores. The distribution and abundance of different molecules provide information on the chemical composition and physical properties in these cores. We study the chemical segregation of three molecules (c-C$_3$H$_2$, CH$_3$OH, CH$_3$CCH) in the starless cores B68 and L1521E, and the prestellar core L1544. We applied the density-based clustering algorithms DBSCAN and HDBSCAN to identify chemical and physical structures within these cores. To enable cross-core comparisons, the input samples were characterised based on their physical environment, discarding the 2D spatial information. The clustering analysis showed significant chemical differentiation across the cores, successfully reproducing the known molecular segregation of c-C$_3$H$_2$ and CH$_3$OH in all three cores. Furthermore, it identifies a segregation between c-C$_3$H$_2$ and CH$_3$CCH, which is not apparent from the emission maps. Key features driving the clustering are integrated intensity, velocity offset, H$_2$ column density, and H$_2$ column density gradient. Different environmental conditions are reflected in the variations in the feature relevance across the cores. This study shows that density-based clustering provides valuable insights into chemical and physical structures of starless cores. It demonstrates that already small datasets of two or three molecules can yield meaningful results. This new approach revealed similarities in the clustering patterns of CH$_3$OH and CH$_3$CCH relative to c-C$_3$H$_2$, suggesting that c-C$_3$H$_2$ traces regions of lower density than to the other two molecules. This allowed for insight into the CH$_3$CCH peak in L1544, which appears to trace a landing point of chemically fresh gas that is accreted to the core, highlighting the impact of accretion processes on molecular distributions.

Figures

Figures reproduced from arXiv: 2507.22482 by the authors.

Figure 1
Figure 1. Comparison of different features for B68 (left), L1521E (middle), and L1544 (right), with c-C3H2 given in green (circle), CH3OH in blue (cross), and CH3CCH in red (diamond). into north/south, and on-centre/off-centre regions. However, un￾like in L1521E, these structures do not consistently correspond to a specific molecular segregation. A separation of the two molecules is visible in Voffset and ∇NH2 in combination … view at source ↗
Figure 2
Figure 2. Clustering results for the starless core B68 for the dataset of Case 1 (left) and the dataset of Case 2 (right) for feature combinations 2, 3, 4, 9, and 10. Each row represents a different combination of features (see Table. 3): combination 2 (intensity, Voffset, and dist2dust), combination 3 (intensity, Voffset, and NH2 ), combination 4 (intensity, Voffset, and ∇NH2 ), combination 9 (intensity, dist2dust, and NH2 )… view at source ↗
Figure 3
Figure 3. Abundances of CH3CCH and c-C3H2 at the dust peaks of the cores. The molecular column density was calculated assuming Tex = 8 K (see text for details). The starless cores are marked with an asterisk. 5. Discussion 5.1. Density-based clustering Segregation between c-C3H2 and CH3OH. Density-based clustering is able to find molecular differentiation in our dataset. In particular, the clustering with the dataset containi… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Centroid velocities (top) and linewidths (bottom) of c-C3H2 (left), CH3CCH (right), and CH3OH (middle) towards the prestellar core L1544. Black contours show 50% and 90% of the respective molecular emission peak. White contours show 30%, 50%, and 90% of the H2 column d…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. End-to-end differentiable retrieval of molecular spectra using hydrodynamics, chemistry, and radiative transfer

    astro-ph.IM 2026-07 conditional novelty 6.0 of 10

    An end-to-end differentiable JAX pipeline couples 1D hydrodynamics, time-dependent chemistry, and radiative transfer, and recovers shock and rate parameters from synthetic HCO+ spectra.

Reference graph

Works this paper leans on

56 extracted references · 38 canonical work pages · cited by 1 Pith paper

  1. [1]

    , " * 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.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1 ...

  2. [2]

    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 global.max substring 't := if while FUNCTION word.in bbl.in " " * FUNCTION format....

  3. [3]

    Alves , F. O. & Franco , G. A. P. 2007, , 470, 597

  4. [4]

    2010, , 518, L102

    Andr \'e , P., Men'shchikov , A., Bontemps , S., et al. 2010, , 518, L102

  5. [5]

    2000, in Protostars and Planets IV, ed

    Andre , P., Ward-Thompson , D., & Barsony , M. 2000, in Protostars and Planets IV, ed. V. Mannings , A. P. Boss , & S. S. Russell , 59

  6. [6]

    & Burie , J

    Bauer , A. & Burie , J. 1969, C. R. Acad. Sci. Paris, B 268, 800

  7. [7]

    2018, , 610, A12

    Bron , E., Daudon , C., Pety , J., et al. 2018, , 610, A12

  8. [8]

    Campello, R. J. G. B., Moulavi, D., & Sander, J. 2013, in Advances in Knowledge Discovery and Data Mining, ed. J. Pei, V. S. Tseng, L. Cao, H. Motoda, & G. Xu (Berlin, Heidelberg: Springer Berlin Heidelberg), 160--172

Show all 56 references
  1. [9]

    A., et al

    Caselli , P., Keto , E., Bergin , E. A., et al. 2012, , 759, L37

  2. [10]

    E., Sipil \"a , O., et al

    Caselli , P., Pineda , J. E., Sipil \"a , O., et al. 2022, , 929, 13

  3. [11]

    M., Zucconi , A., et al

    Caselli , P., Walmsley , C. M., Zucconi , A., et al. 2002, , 565, 331

  4. [12]

    2019, , 622, A141

    Chac \'o n-Tanarro , A., Caselli , P., Bizzocchi , L., et al. 2019, , 622, A141

  5. [13]

    I., Bergin , E

    Cleeves , L. I., Bergin , E. A., Alexander , C. M. O. D., et al. 2014, Science, 345, 1590

  6. [14]

    2015, , 454, 2067

    Colombo , D., Rosolowsky , E., Ginsburg , A., Duarte-Cabral , A., & Hughes , A. 2015, , 454, 2067

  7. [15]

    M., et al

    Crapsi , A., Caselli , P., Walmsley , C. M., et al. 2005, , 619, 379

  8. [16]

    C., & Tafalla , M

    Crapsi , A., Caselli , P., Walmsley , M. C., & Tafalla , M. 2007, , 470, 221

  9. [17]

    N., Coudert , L

    Drozdovskaya , M. N., Coudert , L. H., Margul \`e s , L., et al. 2022, , 659, A69

  10. [18]

    N., Schroeder I , I

    Drozdovskaya , M. N., Schroeder I , I. R. H. G., Rubin , M., et al. 2021, , 500, 4901

  11. [19]

    1996, Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, KDD-96 (AAAI Press), 226

    Ester , M., Kriegel , H.-P., Sander , J., & Xu , X. 1996, Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, KDD-96 (AAAI Press), 226

  12. [20]

    2024, Astronomy and Computing, 48, 100851

    Fotopoulou , S. 2024, Astronomy and Computing, 48, 100851

  13. [21]

    Galli , P. A. B., Loinard , L., Bouy , H., et al. 2019, , 630, A137

  14. [22]

    Galli , P. A. B., Loinard , L., Ortiz-L \'e on , G. N., et al. 2018, , 859, 33

  15. [23]

    2019, radio-astro-tools/spectral-cube: v0.4.4

    Ginsburg , A., Koch , E., Robitaille , T., et al. 2019, radio-astro-tools/spectral-cube: v0.4.4

  16. [24]

    2002, , 565, 359

    Hirota , T., Ito , T., & Yamamoto , S. 2002, , 565, 359

  17. [25]

    & Caselli , P

    Keto , E. & Caselli , P. 2008, , 683, 238

  18. [26]

    2015, , 446, 3731

    Keto , E., Caselli , P., & Rawlings , J. 2015, , 446, 3731

  19. [27]

    J., Bergin , E

    Lada , C. J., Bergin , E. A., Alves , J. F., & Huard , T. L. 2003, , 586, 286

  20. [28]

    W., Myers , P

    Lee , C. W., Myers , P. C., & Tafalla , M. 2001, , 136, 703

  21. [29]

    2022, , 665, A131

    Lin , Y., Spezzano , S., Sipil \"a , O., Vasyunin , A., & Caselli , P. 2022, , 665, A131

  22. [30]

    Mangum , J. G. & Shirley , Y. L. 2015, , 127, 266

  23. [31]

    2017, The Journal of Open Source Software, 2

    McInnes, L., Healy, J., & Astels, S. 2017, The Journal of Open Source Software, 2

  24. [32]

    A., Campello, R

    Moulavi, D., Jaskowiak, P. A., Campello, R. J. G. B., Zimek, A., & Sander, J. 2014, Density-Based Clustering Validation, 839--847

  25. [33]

    M \"u ller , H. S. P., Thorwirth , S., Roth , D. A., & Winnewisser , G. 2001, , 370, L49

  26. [34]

    2019, , 630, A136

    Nagy , Z., Spezzano , S., Caselli , P., et al. 2019, , 630, A136

  27. [35]

    W., Wilner , D

    Ohashi , N., Lee , S. W., Wilner , D. J., & Hayashi , M. 1999, , 518, L41

  28. [36]

    2021, , 923, 168

    Okoda , Y., Oya , Y., Abe , S., et al. 2021, , 923, 168

  29. [37]

    2020, , 900, 40

    Okoda , Y., Oya , Y., Sakai , N., Watanabe , Y., & Yamamoto , S. 2020, , 900, 40

  30. [38]

    2011, Journal of Machine Learning Research, 12, 2825

    Pedregosa, F., Varoquaux, G., Gramfort, A., et al. 2011, Journal of Machine Learning Research, 12, 2825

  31. [39]

    2005, in SF2A-2005: Semaine de l'Astrophysique Francaise, ed

    Pety , J. 2005, in SF2A-2005: Semaine de l'Astrophysique Francaise, ed. F. Casoli , T. Contini , J. M. Hameury , & L. Pagani , 721

  32. [40]

    2018, , 855, 112

    Punanova , A., Caselli , P., Feng , S., et al. 2018, , 855, 112

  33. [41]

    2021, , 656, A109

    Redaelli , E., Sipil \"a , O., Padovani , M., et al. 2021, , 656, A109

  34. [42]

    2010, , 522, A42

    Semenov , D., Hersant , F., Wakelam , V., et al. 2010, , 522, A42

  35. [43]

    2015, , 578, A55

    Sipil \"a , O., Caselli , P., & Harju , J. 2015, , 578, A55

  36. [44]

    D., Hennebelle , P., Martin , P

    Soler , J. D., Hennebelle , P., Martin , P. G., et al. 2013, , 774, 128

  37. [45]

    2016, , 592, L11

    Spezzano , S., Bizzocchi , L., Caselli , P., Harju , J., & Br \"u nken , S. 2016, , 592, L11

  38. [46]

    M., & Lattanzi , V

    Spezzano , S., Caselli , P., Bizzocchi , L., Giuliano , B. M., & Lattanzi , V. 2017, , 606, A82

  39. [47]

    E., et al

    Spezzano , S., Caselli , P., Pineda , J. E., et al. 2020, , 643, A60

  40. [48]

    & Santiago , J

    Tafalla , M. & Santiago , J. 2004, , 414, L53

  41. [49]

    M., & Gottlieb , C

    Thaddeus , P., Vrtilek , J. M., & Gottlieb , C. A. 1985, , 299, L63

  42. [50]

    T., Pineda , J

    Valdivia-Mena , M. T., Pineda , J. E., Segura-Cox , D. M., et al. 2023, , 677, A92

  43. [51]

    C., Herbst , E., et al

    Wakelam , V., Loison , J. C., Herbst , E., et al. 2015, , 217, 20

  44. [52]

    2010, in P roceedings of the 9th P ython in S cience C onference, ed

    W es M c K inney. 2010, in P roceedings of the 9th P ython in S cience C onference, ed. S t\'efan van der W alt & J arrod M illman, 56 -- 61

  45. [53]

    P., Myers , P

    Williams , J. P., Myers , P. C., Wilner , D. J., & Di Francesco , J. 1999, , 513, L61

  46. [54]

    & Lovas , F

    Xu , L.-H. & Lovas , F. 1997, J. Phys. Chem. Ref. Data, 26, 17

  47. [55]

    2022, , 164, 55

    Yan , Q.-Z., Yang , J., Su , Y., et al. 2022, , 164, 55

  48. [56]

    & Lee , J.-E

    Yun , H.-S. & Lee , J.-E. 2023, , 958, 113

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.