Pith. sign in

REVIEW 4 major objections 8 minor 99 references

Sutra unifies filament spine detection and beam-scale physics in one automated pipeline for the interstellar medium.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-11 13:22 UTC pith:5LYBVQ2K

load-bearing objection Useful integrated crest+characterization pipeline for HGBS-style maps; novelty is the packaging, not new physics, and the “extra low-contrast filaments are real” claim is only weakly independent of the DisPerSE/getsf labels. the 4 major comments →

arxiv 2607.04797 v1 pith:5LYBVQ2K submitted 2026-07-06 astro-ph.GA astro-ph.IM

Sutra : An integrated framework for identification and characterization of filaments in the interstellar medium

Shivam Kumaram , Ushasi Bhowmick , Vipin Kumar , Manish Chauhan , Munn V Shukla , Mehul R Pnadya This is my paper
classification astro-ph.GA astro-ph.IM
keywords star formationinterstellar mediuminterstellar filamentscolumn density mapsU-NetPlummer profilesHerschel Gould Belt Survey
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Star-forming gas in molecular clouds is organized into long, thread-like filaments whose widths, masses, and stability set the early conditions of star formation. Existing extractors disagree on which ridges count as filaments and usually separate detection from physical measurement, so large statistical samples remain inconsistent. Sutra trains a U-Net to predict crest-likelihood maps from column-density images using the union of two complementary classical skeletons, then immediately measures beam-scale radial profiles and keeps only segments that fit a cylindrical Plummer law. On Aquila, Orion, and Polaris cutouts the pipeline recovers longer total filament length, especially at low contrast, while the retained structures still show the familiar ~0.1 pc width and shallow p-index. The result is a largely parameter-light, end-to-end tool intended for both local hub-filament studies and survey-scale censuses of early-stage filaments.

Core claim

A crest-focused U-Net trained on the union of DisPerSE and getsf skeletons, followed by beam-scale Plummer filtering, recovers filamentary structures that are consistent with cylindrical profiles even in relatively low-intensity and low-contrast environments, and does so inside a single automated pipeline that also produces local physical property maps.

What carries the argument

The dis:gsf consensus crest label (union of DisPerSE and getsf single-pixel skeletons) together with physics-guided skeleton refinement: each beam-sized segment is kept only if its radial profile yields an acceptable Plummer fit and adequate contrast.

Load-bearing premise

The union skeleton of the two classical methods is treated as a sufficiently complete and unbiased ground truth for supervised crest learning, so that extra ridges the network finds and that pass the Plummer filter are genuine physical filaments rather than inherited method bias.

What would settle it

On a controlled synthetic or high-resolution observational field with an independent ground-truth spine, measure whether the extra low-contrast segments accepted by Sutra (absent from both parent methods) systematically fail kinematic or multi-wavelength filament criteria while the shared segments succeed.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Large Herschel-scale catalogs can be produced with consistent crest definitions and beam-resolved width, line-mass, and contrast maps without per-field threshold retuning.
  • Sub-critical, low-contrast filaments become routinely measurable, allowing statistical tracking of filaments from early formation to fragmentation.
  • Hub-filament systems can be mapped with continuous property gradients along individual crests at beam resolution.
  • The modular design supports transfer learning or new crest models for more distant surveys once maps are reprojected or fine-tuned.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the extra low-contrast population is real, previous width and mass statistics based on high-contrast selections may have systematically under-sampled the sub-critical end of the filament mass function.
  • Crest-likelihood maps that remain stable under background amplification could serve as a common intermediate product that lets different classical extractors be compared on the same probability field rather than on hard masks.
  • Beam-level line-mass maps open a direct route to classify supercritical segments for follow-up molecular-line or magnetic-field observations without a second detection pipeline.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 8 minor

Summary. The manuscript presents Sutra, a U-Net-based pipeline that predicts crest-likelihood maps of interstellar filaments from Herschel column-density maps and then performs beam-scale radial-profile characterization (Plummer fits, line mass, contrast, width). Training labels are single-pixel consensus skeletons formed from the morphological union of DisPerSE and getsf on five nearby HGBS clouds (dis:gsf). After thresholding and medial-axis skeletonization, beam-sized segments are retained or rejected using Plummer-fit quality and contrast, then reconnected by MST. The authors report high recovery of the training skeletons, PCA/KL similarity of radial profiles to DisPerSE/getsf, larger total filament length (especially at low contrast) on Aquila/Orion/Polaris cutouts, and stable precision/recall under synthetic fBm backgrounds. They position Sutra as a largely parameter-light, integrated tool for region-scale and survey-scale filament statistics.

Significance. If the methodological claims hold after clarification, Sutra is a practically useful contribution: it unifies crest-focused identification with beam-resolved physical maps in one automated workflow, which is more directly usable for fragmentation and hub-filament studies than sequential classical extractors. Training on skeleton crests rather than broad intensity masks, using a union of complementary extractors, and releasing a modular package/portal are genuine strengths. The synthetic robustness tests (Appendix A) and resolution/pixel-scale checks (Appendix C) are valuable engineering evidence. The work does not introduce a novel network architecture; its value is the integrated, reproducible pipeline and the beam-level property products. Those products would be of clear interest to the star-formation and ISM community provided the validation of the expanded low-contrast population is framed and tested more carefully.

major comments (4)
  1. §5.1–§5.2, Table 2, Fig. 8–10: The central claim that Sutra recovers additional genuine low-contrast cylindrical filaments rests largely on circular grounding. Labels are the DisPerSE∪getsf union (dis:gsf, §3.1); validation then emphasizes recovery of dis:gsf (Fig. 5), low KL divergence to dis:gsf profiles (Table 2), and that the set-subtraction population N has ⟨p⟩≈2.03, ⟨2R_flat⟩≈0.09 pc and mean reduced χ²≈1.81 after the same Plummer filter used for retention (§3.4, §4.2). That shows morphological consistency with the training prior, not independent physical truth. Please either (i) add an independent check (e.g., expert visual audit with inter-rater rates; comparison to a method not used in labels such as FilFinder/Hessian/template matching on the same cutouts; or velocity-coherence/line-width tests where available), or (ii) reframe claims to “extrapolates the dis:gsf crest morpholog
  2. §5.2 and Table 1: The main comparative science demonstration uses Aquila, Orion, and Polaris cutouts, all of which are among the five training fields. Chunk-level held-out IoU (§3.2.3) does not substitute for field-level generalization. Please add at least one fully held-out HGBS (or other) field never used in label construction/training, report the same Table 3-style statistics there, and discuss any degradation. Without that, claims of suitability for large-scale statistical analyses remain under-supported.
  3. Abstract, §1, §6: The repeated “parameter-free” claim is not accurate. Free or user-set choices include St (§3.3), WBCE weights w1=1, w0=0.1 (§3.2.3), UN64 patch size and 95% overlap inference (§3.2.1), contrast cut C>0.3 (§5.2, Appendix B), DisPerSE persistence/robustness and getsf max length used to build labels (Table 1), flattening factor f=95th percentile (Eq. 2), radial max distance, and segment length Δl. Please replace “parameter-free” with “parameter-light / no per-field retuning at inference” (or similar) and list the fixed defaults and any user knobs in one place (e.g., a short table or Appendix D).
  4. §3.2.3, Eq. (6), Fig. 3: Reported test IoU ≈0.09–0.1 for the selected UN64 model is extremely low for a segmentation metric and is only weakly discussed. For single-pixel crests this may be expected after dilation mismatch, but readers will read IoU as failure unless you (i) justify why IoU is a poor primary metric here, (ii) report precision/recall/F1 or distance-to-crest metrics on held-out chunks (as you already do for synthetics in Appendix A), and (iii) show that the chosen St and physical filter, not IoU alone, define the delivered catalog quality.
minor comments (8)
  1. Throughout: Normalize the product name (Sutra / S¯utra / S ¯utra) and fix spacing artifacts from the LaTeX conversion (e.g., “S ¯utra”, “dis:gsf”, missing spaces after periods).
  2. Eq. (1): Local normalization uses σ(CD_i)^2 in the denominator; confirm whether this is intentional (variance) or a typesetting error for σ(CD_i). If intentional, justify.
  3. Table 3 header and caption: N_fil_H2 units are written inconsistently (cm−1 vs cm−2 in places); fix units and the “×10^21 cm−2” notation.
  4. Fig. 9: Filaments are “thickened for visualization”; state the display dilation explicitly so readers do not confuse display width with measured 2R_flat.
  5. §3.1 / Fig. 1: Briefly quantify the fractional length unique to DisPerSE vs getsf vs the union (beyond total lengths in Table 1), so the benefit of the consensus label is measurable.
  6. Appendix A: State clearly that synthetic spines are analytically cylindrical, so these tests probe background robustness, not correctness of the cylindrical assumption used in filtering.
  7. §5.2.2 / software: The GitHub link is given; if the portal/package is not yet public at acceptance, state what will be released (weights, training scripts, example notebooks) and any license.
  8. References: Some entries appear duplicated (e.g., André et al. 2010; Griffin et al. 2010; Hacar et al. 2023; Ostriker 1964). Clean the bibliography.

Circularity Check

3 steps flagged

High recovery of dis:gsf and Plummer-similarity of the extra set N largely reconfirm reproduction of the training labels plus the cylindrical filter, rather than independent external validation of the low-contrast population.

specific steps
  1. fitted input called prediction [§3.1, §3.3, Fig. 5, §5.1, Table 2]
    "using consensus skeletons constructed from the union of filaments identified by DisPerSE and getsf. ... At the standard classification threshold St = 0.5, the U-Net recovers more than 98% of the dis:gsf skeleton. ... The low value of KL divergence shows that the filaments extracted by the three algorithms have similar radial profiles. ... The similarity of Sutra profiles is higher with dis:gsf skeleton compared to other two methods. This reflects that Sutra has learnt to identify filament crest similar to the dis:gsf skeleton."

    The network is trained to match the single-pixel dis:gsf labels; high recovery of those same labels and low KL divergence of the resulting radial-profile distribution to dis:gsf are therefore statistically forced once the model converges, not an independent test that the learned crests are physical filaments.

  2. self definitional [§3.4, §4.2, §5.2, Fig. 10]
    "the quality of the Plummer fit and the filament contrast are used as physical diagnostics to determine whether the segment is consistent with a cylindrical filament structure. Segments that do not satisfy these criteria are rejected ... we isolate the portion of the Sutra skeleton that is absent from the dis:gsf skeleton. Formally, the set of these additional filaments, denoted here as N ... we perform Plummer fits on N, obtaining a mean reduced χ² of 1.81. ... physical properties p−index and Rflat follow a similar distribution with ⟨p−index⟩=2.03±0.25 and ⟨2Rflat⟩=0.09±0.02 pc, suggesting tha"

    N is defined as the set-subtraction of skeletons that already survived the Plummer-quality and contrast filter of §3.4; reporting that those retained segments have good reduced χ² and p≈2 / 2R_flat≈0.1 pc simply restates the acceptance criteria rather than providing external evidence that the extra low-contrast ridges are genuine filaments.

  3. other [Appendix A, Fig. 12–14]
    "we first define the skeleton of a filament as a 3D Bezier curve ... The radial profiles are randomly modulated ... To complement the qualitative comparison, we perform a quantitative evaluation using the Bezier spine, C(t) projected on z-axis, as ground truth reference spine."

    Synthetic ground-truth spines are constructed to be analytically cylindrical (Plummer-like by design); recovery metrics and Plummer-filter success on these maps therefore cannot independently corroborate that real low-contrast ridges missed by DisPerSE/getsf are physical filaments.

full rationale

Sutra is a methods paper whose core pipeline is supervised crest learning on the union skeleton dis:gsf (DisPerSE ∪ getsf) followed by beam-scale Plummer filtering. Recovery fractions of dis:gsf (Fig. 5, St = 0.5 recovers >98 %), PCA/KL similarity of radial profiles (Fig. 8, Table 2), and the claim that the set-subtraction population N consists of “physically meaningful” filaments (mean reduced χ² ≈ 1.81, ⟨p⟩ ≈ 2.03, ⟨2R_flat⟩ ≈ 0.09 pc) are therefore expected by construction once the network generalizes the labels and the filter retains only good Plummer fits. Synthetic tests embed analytically cylindrical spines, so they cannot break the loop. The paper still has independent engineering content (crest-likelihood formulation, modular beam-level maps, parameter-light automation, qualitative robustness under fBm backgrounds), so the circularity is partial rather than total; the load-bearing scientific claim that the extra low-contrast ridges are genuine filaments beyond the classical extractors remains circularly grounded. No self-citation uniqueness theorems or ansatz smuggling appear.

Axiom & Free-Parameter Ledger

6 free parameters · 4 axioms · 1 invented entities

The central claim rests on supervised learning from two classical extractors, a cylindrical Plummer prior for acceptance, and several operational thresholds. No new physical entity is postulated; the free parameters are algorithmic and the main domain axioms are standard in Herschel filament work. The ledger is therefore mostly tooling and community conventions rather than new physics.

free parameters (6)
  • Skeletonization threshold St
    Converts continuous crest-likelihood to binary skeleton; recovery and precision trade off with St (Fig. 5). Default discussion centers near 0.5–0.7.
  • WBCE class weights (w1=1, w0=0.1)
    Chosen by non-exhaustive hyper-parameter tuning (§3.2.3) to handle extreme class imbalance; directly shapes what the network treats as crest.
  • Patch size UN64 and 95% overlap inference
    Selected after comparing UN32/64/128/256 IoU curves (Fig. 3); sets receptive field and effective feature scale.
  • Contrast cut C>0.3 (and related filtering)
    Used for tabulated comparisons and optional filtering following Arzoumanian et al. 2019; changes which population enters “robust” statistics.
  • DisPerSE persistence/robustness and getsf max length used to build labels
    Table 1 parameters taken from prior HGBS work define the training skeleton union; different choices would change the supervised target.
  • Flattening factor f = 95th percentile of normalized chunk
    Contrast transform in preprocessing (§3.2.1) is a hand-chosen percentile that alters local dynamic range before training/inference.
axioms (4)
  • domain assumption Observed dense filaments are well described as approximately cylindrical structures whose projected column-density profiles follow a Plummer-like law, so fit quality and contrast can accept/reject skeleton segments.
    Invoked in §3.4 and §4.2 with citations to Ostriker, André, Arzoumanian, Federrath; load-bearing for the physics-guided filter.
  • ad hoc to paper The morphological union of DisPerSE and getsf skeletons (after dilation, medial-axis, and length cut <3 HPBW) is a suitable consensus training label for “true” filament crests.
    §3.1 constructs dis:gsf this way; the supervised objective and much of the validation are defined relative to that choice.
  • domain assumption Herschel HGBS column-density maps at SPIRE 500 μm resolution (HPBW 36.3″) adequately trace the filamentary ISM for nearby (<500 pc) clouds.
    Data section §2; standard in the field but limits claimed generality until transfer learning.
  • ad hoc to paper Local chunk normalization plus high-overlap averaging yields a coherent full-map crest probability without destroying filament connectivity.
    §3.2.1 preprocessing/inference strategy; necessary for the end-to-end map product.
invented entities (1)
  • Sutra integrated crest-likelihood + beam-scale characterization pipeline independent evidence
    purpose: Unify supervised ridge detection with Plummer-based segment filtering and property maps in one modular tool.
    Software/method entity, not a new physical particle or force; independent evidence is the public tool and multi-cloud demos rather than a new observable.

pith-pipeline@v1.1.0-grok45 · 28943 in / 3917 out tokens · 34596 ms · 2026-07-11T13:22:00.114717+00:00 · methodology

0 comments
read the original abstract

Observations of the interstellar medium (ISM) at Far-infrared(FIR) and sub-millimetre (sub-mm) wavelengths reveal a complex filamentary structure of dust and gas, which plays a pivotal role in both low and high mass star formation. Large scale identification and characterization of filaments is crucial to establish a link between the ISM and the star formation. We present Sutra, a machine learning based framework that unifies filament identification and beam-scale physical characterization within a single automated pipeline. The framework employs a U-Net architecture to perform supervised segmentation on column density maps and is trained on five nearby (<500pc) molecular clouds from the Herschel Gould Belt Survey (HGBS), using consensus skeletons constructed from the union of filaments identified by DisPerSE and getsf. Rather than reproducing broad intensity-based masks, Sutra predicts crest-likelihood maps focused on filament spines. Beyond identification, Sutra characterizes the filaments at the beam resolution by extracting radial profiles perpendicular to the crest and deriving local structural parameters. The framework provides a parameter-free, computationally efficient approach for consistent filaments identification and systematic investigation of their local properties and shows stable behaviour across varying background conditions in controlled synthetic tests. We demonstrate its application on selected regions from Aquila, Orion and Polaris molecular clouds, and compare the derived filament characteristics with those obtained using existing algorithms. Sutra robustly recovers filamentary structures consistent with cylindrical profiles, including in relatively low-intensity and low-contrast environments, making it well suited for both region-specific studies and large-scale statistical analyses of early-stage star formation and ISM structure.

Figures

Figures reproduced from arXiv: 2607.04797 by Manish Chauhan, Mehul R Pnadya, Munn V Shukla, Shivam Kumaram, Ushasi Bhowmick, Vipin Kumar.

Figure 1
Figure 1. Figure 1: Comparison of filaments identified using Dis￾PerSE and getsf . The top panel compares filaments iden￾tified by DisPerSE (blue) and getsf (red) in an IC5146 cloud cutout. The bottom panel shows the combined skele￾ton (dis:gsf skeleton) used for U-Net training. Grayscale regions are training chunks for the ML model; red regions are excluded. sive, often taking days (e.g., 3-4 days see [PITH_FULL_IMAGE:figur… view at source ↗
Figure 2
Figure 2. Figure 2: Filament Identification workflow (left). Inputs for the algorithm is CD map and the skeleton map, which is passed through the preprocessing module §3.2.1 followed by input to the ML algorithm. Architecture of U-Net model with image subdivision size 64 × 64 (§3.2.1). The size of each layer is indicated in the figure. The dotted orange lines represent the skip-connections between the encoder and decoder laye… view at source ↗
Figure 3
Figure 3. Figure 3: The loss on the validation dataset with training epoch for different models considering different chunk sizes from 32 pixels to 256 pixels for loss functions : WBCE (top) and DICE loss (bottom) apply transfer learning by fine-tuning the final network layers on resampled data. Appendix C demonstrates a controlled test applying the trained model to Hi-GAL maps at native and resampled scales. Appendix C also … view at source ↗
Figure 5
Figure 5. Figure 5: Fraction of filament in the test-dataset recovered by the U-Net model after thresholding and skeletonization. The solid red curve shows the median of the relation over all the chunks in test-dataset and the vertical errorbars show median absolute deviation. The solid blue line shoes the fraction of predicted skeleton which also overlaps with the dis:gsf skeleton. We examine the recovery of the dis:gsf skel… view at source ↗
Figure 4
Figure 4. Figure 4: Illustrations for algorithm adopted for evaluation of recovering capability of the U-Net model in Sutra ¯ . The panels are (a): Chunk of CD map; (b): dis:gsf skeleton cor￾responding to this chunk (§3.1); (c): output of the U-Net model with values lying between 0-1; (d) and (e) : the model output is converted to threshold mask (0.1 for (d) and 0.7 for (e) panel), skeletonised and then dilated to the beam si… view at source ↗
Figure 6
Figure 6. Figure 6: Schematic illustration of crest detection and physics-guided skeleton refinement in Sutra ¯ (a) Input column density (CD) map. (b) Corresponding crest-likelihood map Pcrest(x, y) produced by the trained U-Net ridge-detection model, where high-probability values trace filament spines. (c) Binary skeleton obtained after thresholding the crest-likelihood map and medial-axis refinement. Beam-sized segments (nu… view at source ↗
Figure 7
Figure 7. Figure 7: Illustration of workflow for local characterization of individual filament starting from the output of U-Net identifier after skeletonization: (a) Skeleton obtained from the U-Net identifier. (b) Zooming in to one filament, with the smoothened skeleton(shown in yellow curve) and lines along which radial profile is extracted (green lines). (c) radial profiles for the selected filament. (d) The radial profil… view at source ↗
Figure 8
Figure 8. Figure 8: Distribution Components (C1, C2) obtained by PCA of radial profiles of filaments identified by different al￾gorithms getsf , DisPerSE , Sutra ¯ . 5. RESULT AND DISCUSSION 5.1. Validation: Similarity of filaments via radial profile Implementation of Sutra ¯ on the fields used in train￾ing, (selected chunks containing dis:gsf skeleton) leads to identification of a large number of new filaments even in the ch… view at source ↗
Figure 9
Figure 9. Figure 9: Comparison of Sutra ¯ with DisPerSE and getsf on cutouts in Aquila, Orion and Polaris clouds (Column-wise). The image plot in top two rows shows comparison of DisPerSE +getsf filaments(row-1) with the Sutra ¯ filaments(row-2). The filaments are thickened for better visualization. In Aquila both getsf and DisPerSE filaments mostly overlaps. Orion cutout is mostly dominated by DisPerSE filaments whereas the … view at source ↗
Figure 10
Figure 10. Figure 10: Histogram showing the distribution of p−index and the filament width 2 × Rf lat for all the filaments ex￾tracted using Sutra ¯ on three cutouts of Aquila, Orion and Polaris (filaments in the bottom panel of [PITH_FULL_IMAGE:figures/full_fig_p015_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Example of filament property extraction us￾ing Sutra ¯ on the Mon-R2 region. Filament skeleton map is shown over the CD map. Variation of filament density variation is shown with ’red’ colormap. The radius of beam elements shows the width of filament (R + bg + R − bg) . 5.2.1. Sutra ¯ as a tool to study star-formation Sutra ¯ has significant implications for studying the initial conditions of star formati… view at source ↗
Figure 12
Figure 12. Figure 12: Synthetic data construction used for controlled testing. Left: intrinsic filament density distribution (F) generated from analytically defined 3D spines. Second panel: turbulent background density field (B) generated using fBm model with a power-law power spectrum. Third panel: projected column density map of the combined field F + B. Right panel: higher-back￾ground realization F + 2 × B illustrating redu… view at source ↗
Figure 13
Figure 13. Figure 13: Comparison of structures extracted by different algorithms, tested on a synthetic filament. The top and bottom row shows the result on F + B and F + 2 × B respectively. The panels in Columns shows:( from left to right ):(1) Crest probability map from Sutra ¯ , (2) Eigen values of Hessian matrix, (3) mask generated by FilFinder , (4) Skeleton traced by DisPerSE and (5) Filament and skeleton obtained by get… view at source ↗
Figure 14
Figure 14. Figure 14: Comparison of Sutra ¯ , FilFinder and DisPerSE using precision, recall and F1 score (panels from left to right) on synthetic data with varying background intensity. The error-bar indicates the standard deviation in metric computed on 16 different generated cloud with fixed filament crest for give K on x-axis. nearest point on the spine C(t). we generate three such filaments with central densities as 10,25… view at source ↗
Figure 15
Figure 15. Figure 15: Examples of edge-cases identified by Sutra ¯ . The top row shows a single filament from the Aquila molecular cloud. The middle row shows a single beam element from the filament. The bottom row. Panel (A) shows a beam element corresponding to a low-contrast filament. Panel (B) shows a beam element with overlapping filaments, which lead to kinks in the central maxima. Panel (C) shows a filament with overlap… view at source ↗
Figure 16
Figure 16. Figure 16: The effect of pixel scale on filament identification by Sutra ¯ . The top row shows a CD map at scale 3 pixel/beam and the corresponding probability map identified by Sutra ¯ . Note: both the CD maps are convolved to a beam resolution of 36′′before applying Sutra ¯ . is more complex than the Plummer profile assumed. Examples from three major occurrences are shown in [PITH_FULL_IMAGE:figures/full_fig_p022… view at source ↗
Figure 17
Figure 17. Figure 17: Resolution degradation test using effective distance scaling. Left: Median beam-level filament width across as a function of distance (scaled relative to 260 pc), with error bars indicating dispersion across the sample. Right: Example column density maps and corresponding filament skeletons extracted by Sutra ¯ at three representative distances (260 pc, 520 pc, and 780 pc). The results show that the deriv… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

99 extracted references · 30 canonical work pages · 23 internal anchors

  1. [1]

    Encoding the position to optimize the detection of filaments over a wide range of column density and contrast

    Supervised machine learning on Galactic filaments: II. Encoding the position to optimize the detection of filaments over a wide range of column density and contrast. , keywords =. doi:10.1051/0004-6361/202450828 , adsurl =

    work page doi:10.1051/0004-6361/202450828

  2. [2]

    Probing the multi-scale interplay between gravity and turbulence - Power-law like gravitational energy spectra of the Orion Complex

    Probing the multiscale interplay between gravity and turbulence - power-law-like gravitational energy spectra of the Orion Complex. , keywords =. doi:10.1093/mnras/stw2504 , archivePrefix =. 1603.05417 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stw2504

  3. [3]

    Statistical model for filamentary structures of molecular clouds -- The modified multiplicative random cascade model and its multifractal nature

    Statistical model for filamentary structures of molecular clouds. The modified multiplicative random cascade model and its multifractal nature. , keywords =. doi:10.1051/0004-6361/201937085 , archivePrefix =. 2007.08206 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/201937085 2007

  4. [4]

    Large-scale Velocity-coherent Filaments in the SEDIGISM Survey: Association with Spiral Arms and Fraction of Dense Gas

    Large-scale velocity-coherent filaments in the SEDIGISM survey: Association with spiral arms and the fraction of dense gas. , keywords =. doi:10.1051/0004-6361/202245784 , archivePrefix =. 2305.07353 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202245784

  5. [5]

    Unified Focal loss: Generalising Dice and cross entropy-based losses to handle class imbalanced medical image segmentation , journal =

    Michael Yeung and Evis Sala and Carola-Bibiane Schönlieb and Leonardo Rundo , keywords =. Unified Focal loss: Generalising Dice and cross entropy-based losses to handle class imbalanced medical image segmentation , journal =. 2022 , issn =. doi:https://doi.org/10.1016/j.compmedimag.2021.102026 , url =

    doi:10.1016/j.compmedimag.2021.102026 2022

  6. [6]

    PeerJ , volume=

    scikit-image: image processing in Python , author=. PeerJ , volume=. 2014 , publisher=

    2014

  7. [7]

    and Varoquaux, G

    Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V. and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P. and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E. , journal=. Scikit-learn: Machine Learning in

  8. [8]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Probabilistic principal component analysis , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 1999 , publisher=

    1999

  9. [9]

    Communications of the ACM , volume=

    A fast parallel algorithm for thinning digital patterns , author=. Communications of the ACM , volume=. 1984 , publisher=

    1984

  10. [10]

    BMC medical research methodology , volume=

    Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range , author=. BMC medical research methodology , volume=. 2014 , publisher=

    2014

  11. [11]

    Comparison of Herschel and ArT\'eMiS observations of massive filaments

    Comparison of Herschel and ArT \'e MiS observations of massive filaments. , keywords =. doi:10.1051/0004-6361/202346425 , archivePrefix =. 2501.11507 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202346425

  12. [12]

    , keywords =

    Template matching method for the analysis of interstellar cloud structure. , keywords =. doi:10.1051/0004-6361/201628727 , archivePrefix =. 1607.01931 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201628727

  13. [13]

    , keywords =

    The Astropy Project: Sustaining and Growing a Community-oriented Open-source Project and the Latest Major Release (v5.0) of the Core Package. , keywords =. doi:10.3847/1538-4357/ac7c74 , archivePrefix =. 2206.14220 , primaryClass =

    Pith/arXiv arXiv doi:10.3847/1538-4357/ac7c74

  14. [14]

    and Haberland, Matt and Reddy, Tyler and Cournapeau, David and Burovski, Evgeni and Peterson, Pearu and Weckesser, Warren and Bright, Jonathan and

    Virtanen, Pauli and Gommers, Ralf and Oliphant, Travis E. and Haberland, Matt and Reddy, Tyler and Cournapeau, David and Burovski, Evgeni and Peterson, Pearu and Weckesser, Warren and Bright, Jonathan and. Nature Methods , year =

  15. [15]

    L1495 Revisited: A PPMAP View of a Star-Forming Filament

    L1495 revisited: a PPMAP view of a star-forming filament. , keywords =. doi:10.1093/mnras/stz2234 , archivePrefix =. 1908.02295 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stz2234 1908

  16. [16]

    On the 3D Curvature and Dynamics of the Musca filament

    On the 3D Curvature and Dynamics of the Musca Filament. , keywords =. doi:10.3847/1538-4357/acc462 , archivePrefix =. 2303.09049 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.3847/1538-4357/acc462

  17. [17]

    CUTEX: CUrvature Thresholding EXtractor

  18. [18]

    Filamentary structure as seen in C ^ 18 O emission

    The CARMA-NRO Orion Survey. Filamentary structure as seen in C ^ 18 O emission. , keywords =. doi:10.1051/0004-6361/201834049 , archivePrefix =. 1901.00176 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201834049 1901

  19. [19]

    RadFil: a Python Package for Building and Fitting Radial Profiles for Interstellar Filaments

    RadFil: A Python Package for Building and Fitting Radial Profiles for Interstellar Filaments. , keywords =. doi:10.3847/1538-4357/aad3b5 , archivePrefix =. 1807.06567 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.3847/1538-4357/aad3b5

  20. [20]

    , keywords =

    The JCMT Gould Belt Survey: properties of star-forming filaments in Orion A North. , keywords =. doi:10.1093/mnras/stv369 , adsurl =

    work page doi:10.1093/mnras/stv369

  21. [21]

    Fiber networks in Orion

    Emergence of high-mass stars in complex fiber networks (EMERGE): III. Fiber networks in Orion. , keywords =. doi:10.1051/0004-6361/202449316 , archivePrefix =. 2409.01321 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/202449316

  22. [22]

    The ATLASGAL survey: a catalog of dust condensations in the Galactic plane

    The ATLASGAL survey: a catalog of dust condensations in the Galactic plane. , keywords =. doi:10.1051/0004-6361/201322434 , archivePrefix =. 1312.0937 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/201322434

  23. [23]

    First data release for the inner Milky Way: +68 l -70

    Hi-GAL, the Herschel infrared Galactic Plane Survey: photometric maps and compact source catalogues. First data release for the inner Milky Way: +68 l -70. , keywords =. doi:10.1051/0004-6361/201526380 , archivePrefix =. 1604.05911 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201526380

  24. [24]

    Filament coalescence and hub structure in MonR2: Implications to massive star and cluster formation

    Filament coalescence and hub structure in Mon R2. Implications for massive star and cluster formation. , keywords =. doi:10.1051/0004-6361/202140363 , archivePrefix =. 2112.06803 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202140363

  25. [25]

    , keywords =

    Kinematics and star formation of hub-filament systems in W49A. , keywords =. doi:10.1051/0004-6361/202348580 , archivePrefix =. 2406.08906 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/202348580

  26. [26]

    , keywords =

    G321.93-0.01: A Rare Site of Multiple Hub-filament Systems with Evidence of Collision and Merging of Filaments. , keywords =. doi:10.3847/1538-3881/ad98ff , archivePrefix =. 2411.13870 , primaryClass =

    Pith/arXiv arXiv doi:10.3847/1538-3881/ad98ff

  27. [27]

    Massive star-formation in the hub-filament system of RCW 117

    Massive star formation in the hub-filament system of RCW 117. , keywords =. doi:10.1093/mnras/stad3385 , archivePrefix =. 2311.00477 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stad3385

  28. [28]

    The Impact of Expanding HII Regions on Filament G37:Curved Magnetic Field and Multiple Direction Material Flows

    The impact of expanding HII regions on filament G37: Curved magnetic field and multiple direction material flows. , keywords =. doi:10.1051/0004-6361/202450965 , archivePrefix =. 2503.03219 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202450965

  29. [29]

    Observational signatures of end-dominated collapse in the S242 filamentary structure

    Observational Signatures of End-dominated Collapse in the S242 Filamentary Structure. , keywords =. doi:10.3847/1538-4357/ab1aa6 , archivePrefix =. 1904.07639 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.3847/1538-4357/ab1aa6 1904

  30. [30]

    Filament fragmentation: Density gradients suppress end dominated collapse

    Filament fragmentation: density gradients suppress end-dominated collapse. , keywords =. doi:10.1093/mnras/stad2517 , archivePrefix =. 2307.11162 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stad2517

  31. [31]

    Velocity Structure and Molecular Formation in Polaris Molecular Cloud

    Velocity Structure and Molecular Formation in the Polaris Molecular Cloud. , keywords =. doi:10.3847/1538-4357/adb418 , archivePrefix =. 2502.10668 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.3847/1538-4357/adb418

  32. [32]

    , keywords =

    On the universality of interstellar filaments: theory meets simulations and observations. , keywords =. doi:10.1093/mnras/stv2880 , archivePrefix =. 1510.05654 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/stv2880

  33. [33]

    On the typical width of Herschel filaments

    The typical width of Herschel filaments. , keywords =. doi:10.1051/0004-6361/202244541 , archivePrefix =. 2210.04736 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202244541

  34. [34]

    Zavagno and F.-X

    A. Zavagno and F.-X. Duper and S. Bemsaid and others , title =. Astronomy and Astrophysics (A and A) , volume =

  35. [35]

    , year = 1964, month = oct, volume =

    The Equilibrium of Polytropic and Isothermal Cylinders. , year = 1964, month = oct, volume =. doi:10.1086/148005 , adsurl =

    doi:10.1086/148005 1964

  36. [36]

    , keywords =

    Theory of Star Formation. , keywords =. doi:10.1146/annurev.astro.45.051806.110602 , archivePrefix =. 0707.3514 , primaryClass =

    Pith/arXiv arXiv doi:10.1146/annurev.astro.45.051806.110602

  37. [37]

    A Photographic Atlas of Selected Regions of the Milky Way

  38. [38]

    , year = 1962, month = may, volume =

    Catalogue of Dark Nebulae. , year = 1962, month = may, volume =. doi:10.1086/190072 , adsurl =

    doi:10.1086/190072 1962

  39. [39]

    , keywords =

    A catalog of dark globular filaments. , keywords =. doi:10.1086/190609 , adsurl =

    doi:10.1086/190609

  40. [40]

    , keywords =

    Evolution of dust properties in an interstellar filament. , keywords =. doi:10.1051/0004-6361:20021309 , adsurl =

    doi:10.1051/0004-6361:20021309

  41. [41]

    , keywords =

    Filamentary Structure of Star-forming Complexes. , keywords =. doi:10.1088/0004-637X/700/2/1609 , archivePrefix =. 0906.2005 , primaryClass =

    Pith/arXiv arXiv doi:10.1088/0004-637x/700/2/1609 2005

  42. [42]

    Flows, Fragmentation, and Star Formation. I. Low-Mass Stars in Taurus. , keywords =. doi:10.1086/342657 , archivePrefix =. astro-ph/0207216 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/342657

  43. [43]

    , keywords =

    Large-Scale Structure of the Molecular Gas in Taurus Revealed by High Linear Dynamic Range Spectral Line Mapping. , keywords =. doi:10.1086/587166 , archivePrefix =. 0802.2206 , primaryClass =

    Pith/arXiv arXiv doi:10.1086/587166

  44. [44]

    , year = 1907, month = apr, volume =

    On a nebulous groundwork in the constellation Taurus. , year = 1907, month = apr, volume =. doi:10.1086/141434 , adsurl =

    doi:10.1086/141434 1907

  45. [45]

    An ESA facility for far-infrared and submillimetre astronomy

    Herschel Space Observatory. An ESA facility for far-infrared and submillimetre astronomy. , keywords =. doi:10.1051/0004-6361/201014759 , archivePrefix =. 1005.5331 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201014759

  46. [46]

    , keywords =

    From filamentary clouds to prestellar cores to the stellar IMF: Initial highlights from the Herschel Gould Belt Survey. , keywords =. doi:10.1051/0004-6361/201014666 , archivePrefix =. 1005.2618 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201014666

  47. [47]

    , keywords =

    Initial highlights of the HOBYS key program, the Herschel imaging survey of OB young stellar objects. , keywords =. doi:10.1051/0004-6361/201014690 , adsurl =

    doi:10.1051/0004-6361/201014690

  48. [48]

    , keywords =

    Clouds, filaments, and protostars: The Herschel Hi-GAL Milky Way. , keywords =. doi:10.1051/0004-6361/201014659 , archivePrefix =. 1005.3317 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201014659

  49. [49]

    Protostars and Planets VI , year = 2014, editor =

    From Filamentary Networks to Dense Cores in Molecular Clouds: Toward a New Paradigm for Star Formation. Protostars and Planets VI , year = 2014, editor =. doi:10.2458/azu_uapress_9780816531240-ch002 , archivePrefix =. 1312.6232 , primaryClass =

    Pith/arXiv arXiv doi:10.2458/azu_uapress_9780816531240-ch002 2014

  50. [50]

    Filament morphologies

    On the nature of star-forming filaments - I. Filament morphologies. , keywords =. doi:10.1093/mnras/stu1915 , archivePrefix =. 1407.6716 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/stu1915

  51. [51]

    , keywords =

    A census of dense cores in the Aquila cloud complex: SPIRE/PACS observations from the Herschel Gould Belt survey. , keywords =. doi:10.1051/0004-6361/201525861 , archivePrefix =. 1507.05926 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201525861

  52. [52]

    Evidence that widespread star formation may be underway in G0.253+016, "The Brick"

    Evidence that widespread star formation may be underway in G0.253+0.016, `The Brick'. , keywords =. doi:10.1093/mnrasl/slw080 , archivePrefix =. 1604.07609 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnrasl/slw080

  53. [53]

    , keywords =

    Herschel view of the Taurus B211/3 filament and striations: evidence of filamentary growth?. , keywords =. doi:10.1051/0004-6361/201220500 , archivePrefix =. 1211.6360 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201220500

  54. [54]

    , keywords =

    Filamentary Accretion Flows in the Embedded Serpens South Protocluster. , keywords =. doi:10.1088/0004-637X/766/2/115 , archivePrefix =. 1301.6792 , primaryClass =

    Pith/arXiv arXiv doi:10.1088/0004-637x/766/2/115

  55. [55]

    Part III

    On the Gravitational Instability of Some Magneto-Hydrodynamical Systems of Astrophysical Interest. Part III. , keywords =

  56. [56]

    The large-scale structure and dynamics of filamentary molecular clouds

    Magnetized interstellar molecular clouds - II. The large-scale structure and dynamics of filamentary molecular clouds. , keywords =. doi:10.1093/mnras/stz653 , archivePrefix =. 1901.04593 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/stz653 1901

  57. [57]

    Progress of Theoretical Physics , year = 1987, month = mar, volume =

    Gravitational Instability of the Isothermal Gas Cylinder with an Axial magnetic Field. Progress of Theoretical Physics , year = 1987, month = mar, volume =. doi:10.1143/PTP.77.635 , adsurl =

    doi:10.1143/ptp.77.635 1987

  58. [58]

    , keywords =

    Optical Polarization Maps of Star-forming Regions in Perseus, Taurus, and Ophiuchus. , keywords =. doi:10.1086/169070 , adsurl =

    work page doi:10.1086/169070

  59. [59]

    Planck 2015 results. XIX. Constraints on primordial magnetic fields. , keywords =. doi:10.1051/0004-6361/201525821 , archivePrefix =. 1502.01594 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201525821 2015

  60. [60]

    Planck intermediate results. XXXIII. Signature of the magnetic field geometry of interstellar filaments in dust polarization maps. , keywords =. doi:10.1051/0004-6361/201425305 , archivePrefix =. 1411.2271 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201425305

  61. [61]

    , keywords =

    An Imprint of Molecular Cloud Magnetization in the Morphology of the Dust Polarized Emission. , keywords =. doi:10.1088/0004-637X/774/2/128 , archivePrefix =. 1303.1830 , primaryClass =

    Pith/arXiv arXiv doi:10.1088/0004-637x/774/2/128

  62. [62]

    Comptes Rendus Geoscience , keywords =

    Interstellar filaments and star formation. Comptes Rendus Geoscience , keywords =. doi:10.1016/j.crte.2017.07.002 , archivePrefix =. 1710.01030 , primaryClass =

    Pith/arXiv arXiv doi:10.1016/j.crte.2017.07.002 2017

  63. [63]

    From Interstellar Clouds to Star-Forming Galaxies: Universal Processes? , year = 2016, editor =

    Properties of interstellar filaments as derived from Herschel, Planck, and molecular line observations. From Interstellar Clouds to Star-Forming Galaxies: Universal Processes? , year = 2016, editor =. doi:10.1017/S1743921316007250 , adsurl =

    work page doi:10.1017/s1743921316007250 2016

  64. [64]

    Protostars and Planets VII , year = 2023, editor =

    Initial Conditions for Star Formation: a Physical Description of the Filamentary ISM. Protostars and Planets VII , year = 2023, editor =. doi:10.48550/arXiv.2203.09562 , archivePrefix =. 2203.09562 , primaryClass =

    Pith/arXiv arXiv doi:10.48550/arxiv.2203.09562 2023

  65. [65]

    , keywords =

    Characterizing interstellar filaments with Herschel in IC 5146. , keywords =. doi:10.1051/0004-6361/201116596 , archivePrefix =. 1103.0201 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201116596

  66. [66]

    Removing visual bias in filament identification: a new goodness-of-fit measure

    Removing Visual Bias in Filament Identification: A New Goodness-of-fit Measure. , keywords =. doi:10.3847/2041-8213/aa6e50 , archivePrefix =. 1704.06377 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.3847/2041-8213/aa6e50 2041

  67. [67]

    , year = 1916, month = jan, volume =

    Some of the dark markings on the sky and what they suggest. , year = 1916, month = jan, volume =. doi:10.1086/142225 , adsurl =

    work page doi:10.1086/142225 1916

  68. [68]

    arXiv e-prints , keywords =

    V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation. arXiv e-prints , keywords =. doi:10.48550/arXiv.1606.04797 , archivePrefix =. 1606.04797 , primaryClass =

    Pith/arXiv arXiv doi:10.48550/arxiv.1606.04797

  69. [69]

    , keywords =

    Filament identification through mathematical morphology. , keywords =. doi:10.1093/mnras/stv1521 , archivePrefix =. 1507.02289 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/stv1521

  70. [70]

    2009 , publisher=

    The Elements of Statistical Learning: Data Mining, Inference, and Prediction , author=. 2009 , publisher=

    2009

  71. [71]

    Proceedings of the 19th annual symposium on Computational Geometry , pages=

    Morse-Smale complexes for piecewise linear 3-manifolds , author=. Proceedings of the 19th annual symposium on Computational Geometry , pages=. 2003 , publisher=

    2003

  72. [72]

    , keywords =

    The Atacama Pathfinder EXperiment (APEX) - a new submillimeter facility for southern skies -. , keywords =. doi:10.1051/0004-6361:20065420 , adsurl =

    doi:10.1051/0004-6361:20065420

  73. [73]

    Astronomy & Astrophysics , volume=

    The ATLASGAL survey: A catalog of dense clumps in the 330° < ℓ < 21° Galactic plane , author=. Astronomy & Astrophysics , volume=. 2014 , publisher=

    2014

  74. [74]

    , keywords =

    ATLASGAL: A Galaxy-wide sample of dense filamentary structures. , keywords =. doi:10.1051/0004-6361/201527468 , archivePrefix =. 1604.00544 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201527468

  75. [75]

    Astronomy & Astrophysics , volume=

    Initial highlights of the HOBYS key program, the Herschel imaging survey of OB young stellar objects , author=. Astronomy & Astrophysics , volume=. 2010 , publisher=

    2010

  76. [76]

    An ESA facility for far-infrared and submillimetre astronomy , author=

    Herschel Space Observatory. An ESA facility for far-infrared and submillimetre astronomy , author=. Astronomy & Astrophysics , volume=. 2010 , publisher=

    2010

  77. [77]

    Astronomy & Astrophysics , volume=

    From filamentary clouds to prestellar cores to the stellar IMF: Initial highlights from the Herschel Gould Belt Survey , author=. Astronomy & Astrophysics , volume=. 2010 , publisher=

    2010

  78. [78]

    arXiv e-prints , keywords =

    U-Net: Convolutional Networks for Biomedical Image Segmentation. arXiv e-prints , keywords =. doi:10.48550/arXiv.1505.04597 , archivePrefix =. 1505.04597 , primaryClass =

    Pith/arXiv arXiv doi:10.48550/arxiv.1505.04597

  79. [79]

    Theory and implementation

    The persistent cosmic web and its filamentary structure - I. Theory and implementation. , keywords =. doi:10.1111/j.1365-2966.2011.18394.x , archivePrefix =. 1009.4015 , primaryClass =

    doi:10.1111/j.1365-2966.2011.18394.x 2011

  80. [80]

    , keywords =

    Characterizing the properties of nearby molecular filaments observed with Herschel. , keywords =. doi:10.1051/0004-6361/201832725 , archivePrefix =. 1810.00721 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201832725

Showing first 80 references.