Accelerating star formation of dense clumps

Xunchuan Liu

REVIEW 5 major objections 5 minor 2 cited by

Accelerating star formation of dense clumps

T0 review · 5 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read Dense clumps gain star-forming mass exponentially, and the most massive embedded cores grow on the same accelerating clock.

desk verdict The paper's clock is built by requiring Mvir to grow exponentially, so Eq. 2 is a calibration, not a discovery; still, the CMF convolution idea is neat and worth a referee's time. read the letter →

arxiv 2510.25436 v2 pith:QB4452SD submitted 2025-10-29 astro-ph.GA

Xunchuan Liu This is my paper · ORCID

classification astro-ph.GA

keywords starformationdenseclumpsvirialmassdusttemperaturecorefunctionprotoclustersexponentialgrowthlaw

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper aims to establish that star formation inside dense molecular clumps is an accelerating process: both the total mass of star-forming gas in a protocluster and the mass of its most massive embedded core grow exponentially in time, with similar e-folding timescales. The author constructs a statistical 'clock' from the cumulative distribution of dust temperatures across a large unbiased sample of clumps, then shows that virial mass grows as 200 e^t solar masses and maximum core mass as 0.9 e^t solar masses. If correct, this single accelerating framework reproduces observed luminosity evolution, the top-heavy shape of the core mass function, and the linear dense-gas star formation law, linking stellar and cluster scales.

What carries the argument

The machinery is a statistical evolutionary clock: the cumulative distribution function (CDF) of dust temperature, mapped to normalized time t = gamma CDF(T_dust) t0 (Eq. 1), with gamma calibrated so that virial mass rises as e^t. On this clock, the exponential relations M_vir = 200 e^t and M_max_core = 0.9 e^t are established, and the assumption that every core accretes at a rate proportional to its own mass (Mdot_core = M_core) converts exponential growth into a convolution that preserves an M^-1 high-mass tail in the CMF.

What would settle it

Find an independent age indicator—for instance, chemical abundances or outflow momentum—and measure clump virial mass and maximum core mass for the same objects. If M_vir and M_max_core do not grow together on a single e-folding timescale, the central claim is falsified. Alternatively, a sample with a different T_dust distribution but the same true ages would break the universal clock.

Watch

Extended reading notes

Core claim

The central claim is that protoclusters and their most massive cores undergo exponentially accelerating mass assembly with comparable timescales. Using the cumulative distribution of dust temperature to define a normalized evolutionary time t, the author finds M_vir = 200 e^t M_sun for star-forming clumps and M_max_core = 0.9 e^t M_sun for the embedded cores, so both scales grow as e^t. This self-similar growth naturally yields a core mass function whose high-mass slope approaches M^-1 regardless of the initial core mass function, and it reproduces the observed luminosity evolution and the linear dense-gas star formation law.

Load-bearing premise

The whole result rests on treating the cumulative dust-temperature distribution as a linear clock, assuming T_dust rises monotonically with age across an unbiased clump sample; if that mapping is wrong, the exponential growth is an artifact of the time axis.

Editorial extensions

If this is right

Star-forming gas in a clump doubles on a characteristic timescale t0 (~10^5 yr, uncalibrated), so protocluster assembly accelerates rather than proceeding at a steady rate.
The most massive core tracks the protocluster mass, so the growth timescales of protoclusters and their most massive protostars are comparable, linking stellar and cluster scales.
The core mass function becomes progressively top-heavy with time, with an asymptotic high-mass slope near M^-1 independent of the initial core mass function.
Bolometric luminosity should rise gently at first (accretion-dominated, e^t) and then steeply (stellar-dominated, e^{3.5t}), matching observed clump luminosity.
The dense-gas star formation law SFR ∝ Mdense follows from Mdot_SF ∝ M_SF, extending the clump-scale result to cloud and galactic scales.

Reading between the lines

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

If the e-folding timescale t0 is universal, it should also appear in other samples spanning a wider temperature range; measuring t0 directly (e.g., via outflow kinematics or chemical clocks) would test the framework.
The clock assumes an unbiased, complete sample; evolutionary selection effects or non-monotonic dust temperature evolution could distort the t-axis and mimic exponential growth, so an independent age tracer is needed.
The framework predicts a tight correlation between virial mass and maximum core mass with a specific power law; this could be checked in high-resolution ALMA surveys beyond ALMAGAL.
Because the CMF slope is claimed to be independent of the initial CMF, this can be tested by varying mass-selection thresholds in observations of low-mass vs high-mass star-forming regions.

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.

Desk Editor's Note

read the letter

Quick take: the central claim 'Mvir ∝ e^t' is real by construction. Section 3.1 defines t = γ CDF(Tdust) t0 and then says 'We use this relation to calibrate γ such that Mvir ∝ e^t.' So Eq. 2 is a chosen normalization, not a measured law. The rest of the paper reuses the same t-axis for Mmax_core and Lbol, which means those correlations cannot independently confirm the acceleration. The stress-test note is right, and the reader's circularity burden score of 8 is fair.

That said, the paper is not sloppy. It is clearly written, honest about the calibration (it says exactly what it does), and the underlying idea is interesting: if you believe Tdust CDF is a steady-state clock, then the method should linearize any monotonic evolution, and you could test it by checking whether different properties collapse onto simple curves. The author applies the CDF-linearization trick from disk studies to clumps and gets Mmax_core ~ e^t and Lbol ~ e^t + e^{3.5t}. Those are consistent with the calibrated t, but they are not independent checks.

The CMF argument in Sect. 4.1 is the most novel piece: under exponential accretion (Mdot_core = M_core) and constant ICMF, the convolution of shifted ICMFs produces a top-heavy CMF with slope ~ -1, independent of ICMF shape. That's a clean mathematical point and it makes a falsifiable prediction: the CMF should flatten with time, which is what Coletta et al. (2025) report qualitatively. The comparison is scaled by hand, though, so it's suggestive rather than quantitative.

Soft spots, in order: (1) the t-axis assumes the ATLASGAL sample is an unbiased steady-state draw and Tdust is a single-valued monotonic age tracer. The paper asserts this but doesn't demonstrate it; the 10% NH3 subsample could be biased. (2) t0 is left uncalibrated, so absolute timescales (10^5 yr) are just an order-of-magnitude guess. (3) The luminosity model has two fitted coefficients (a, b) and ignores feedback, so Eq. 14 is not a strong test. (4) Eq. 6 normalization (0.9 M_sun at t=0) seems suspiciously low, but that's a minor point.

Who should read it: anyone working on clump evolution or the CMF. It's a provocative framework, and the convolution result is worth discussing even if you don't buy the clock. I'd give it a serious referee — the flaws are structural but the author might be able to reframe the paper as a testable model rather than a claim of discovery. As written, I'd reject for publication because the headline is fitted; but it's close enough to warrant a chance for major revision.

Referee Report

5 major / 5 minor

Summary. The paper proposes a statistical framework in which the cumulative distribution function of dust temperature in ATLASGAL clumps is used as a linear evolutionary clock. The virial mass Mvir is claimed to grow as 200 e^t M_sun; the maximum core mass in ALMAGAL clumps is claimed to grow as 0.9 e^t M_sun, implying exponentially accelerating mass assembly on comparable timescales. The framework is then used to reproduce the observed bolometric luminosity evolution, a top-heavy core mass function with high-mass slope alpha=1, and the dense-gas star formation law, and it is extended to a discussion of the CMZ. The paper is written in a clear and direct style and uses large public surveys.

Significance. If the central exponential-growth claim were independently established, the paper would offer a unified quantitative framework linking clump-scale and core-scale star formation, and would naturally explain the shape of the CMF and the dense-gas star formation law. The manuscript is explicitly argued and makes its assumptions visible, which is a strength. However, the main evidence is circular: the time axis is defined so that Mvir becomes exponential, and every subsequent result inherits this constructed correlation. The luminosity model has fitted coefficients, the CMF comparison is scaled by hand, and the key convolution lemma is only cited to an unpublished preprint. The paper therefore does not currently provide the empirical support needed for its central claim, although the framework may become useful if calibrated against an independent age indicator in future work.

major comments (5)

[Section 3.1, Eqs. (1) and (2)] The evolutionary time is defined as t = gamma * CDF(T_dust) * t0, and the factor gamma is then calibrated so that Mvir ∝ e^t, yielding Eq. (2). This is a rank transformation: for any monotonic relation between Mvir and T_dust, gamma can be chosen to make the Mvir–t relation approximately exponential. Thus Eq. (2) is fitted by construction, not a measured exponential growth. In addition, using the CDF as an age proxy assumes that the ATLASGAL sample is an unbiased, complete, steady-state representation of the evolutionary sequence and that T_dust is a single-valued, monotonic function of age. The paper asserts that the T_dust distribution is continuous and smooth but does not test these assumptions; the subsample with virial masses is only about 10% of the full sample and may be subject to selection effects. The central claim therefore lacks an independent calibration.
[Section 3.2, Eq. (6) and Figure 2] The ALMAGAL sample is assigned the same t derived from the ATLASGAL T_dust CDF. The correlation between log10(Mmax_core) and t is therefore not independent evidence; it is the same rank-ordering applied to a different, biased sample. The statement that cluster mass and maximum core mass have 'comparable timescales' compares two exponential fits on the same constructed t-axis, so it does not demonstrate a physical correspondence between the two growth processes. The model also assumes Mmax_core ∝ Mcluster via the CMF (Eq. 5), which itself depends on an adopted CMF slope, so the loop from assumption to conclusion is not broken.
[Section 3.3, Eqs. (13)–(15)] The luminosity model contains two free coefficients, a and b, which are fitted to the observed Lbol–t relation. Since t is constructed from T_dust and Lbol is independently known to increase with T_dust, the fit only shows that a two-parameter exponential sum can describe the data; it is not a test of the accelerating scenario. Moreover, the model assumes Lacc ∝ Mdot_SF ∝ e^t, which is exactly the exponential trend that the paper claims to discover. The good agreement in Figure 2 is therefore a consequence of the model’s flexibility, not an independent confirmation.
[Section 4.1, Eqs. (16)–(19) and Figure 3] The derivation of a top-heavy CMF with slope alpha=1 relies on the lemma that the shallowest exponential tail is preserved under convolution, cited to an arXiv preprint (Liu 2025), and on the assumptions of Eq. (8) and Eq. (9): Mdot_core = Mcore and Ncore ∝ e^t. Both assumptions are motivated by the circularly-derived exponential growth of Mvir, so the resulting CMF slope is not an independent prediction. In addition, the comparison with ALMAGAL data in Figure 3 is made only after linearly scaling the x-axis values to match observations; no quantitative goodness-of-fit or uncertainty estimate is provided. Thus the CMF test does not provide strong support for the model.
[Section 4.2, Eq. (21)] The derived star formation law SFR ∝ Mvir ∝ e^t uses the same constructed exponential relation, so it cannot serve as validation. The discussion of the CMZ invokes an assumed virial parameter alpha_vir ~ 0.1 to reconcile the low SFR, but this is a post hoc adjustment rather than a test. At this point the framework has not been compared against any independent chronological tracer or any quantitative prediction that was not already built into the time axis.

minor comments (5)

[Figure 1 caption] In panel (f) the caption says 'the exponential fit between Mvir and t (bla)'; 'bla' appears to be a typo and should probably read 'black'. This wording also makes the circular calibration explicit, which should be acknowledged as a limitation in the main text.
[Section 3.1] The statement 't0 (approximately 10^5 years)' is presented without derivation or justification. Since t0 is explicitly left uncalibrated, the physical interpretation of the e-folding timescale should be clearly labeled as a speculative estimate.
[Section 3.2, Eq. (7)] The symbol M_SF is used in Eq. (7) and the surrounding text but is not defined at first use. It is later identified with Mvir, but the identification should be made explicit in Eq. (7) or earlier.
[Section 4.1, Eq. (16)] The integral in Eq. (16) is written in a compact form that is not immediately transparent. The role of the function f(y;t), the definition of ICMF_ln[y], and the meaning of the convolution with respect to ln Mcore should be spelled out.
[References] The key mathematical result used in Eq. (19) (preservation of the shallowest exponential tail under convolution) is cited to an arXiv preprint (Liu 2025). Since the result is load-bearing, the derivation should be either included in the appendices or replaced by a peer-reviewed reference.

Circularity Check

1 steps flagged · score 6.0 of 10

Eq. (2)'s exponential virial-mass growth is imposed by calibrating γ on Mvir; downstream 'accelerating' relations use the same fitted clock.

fitted input called prediction [Sec. 3.1, Eqs. (1)-(2) and Fig. 1 caption]
"t=γCDF(Tdust)t0 ... We use this relation to calibrate γ such that Mvir∝e^t, which for the best fit (black line in panel f) corresponds to Mvir=200e^t M⊙."

The normalized age t is defined from the CDF of Tdust, and γ is then adjusted so that Mvir∝e^t on that axis. Thus Eq. (2) is the calibration relation itself, not an independent discovery: the exponential growth of the virial mass is put in by choosing the horizontal scale. Since t0 is explicitly left uncalibrated ('the absolute timescale t0 ... remains undetermined'), no independent clock anchors the e-folding time. Every later use of the same t-axis (Eqs. 6, 13, 14, 16, 21) inherits this fitted normalization, so the cross-scale comparison cannot validate the exponential scenario.

full rationale

The paper's central quantitative claim—that virial mass grows as Mvir ∝ e^t—reduces to the way the time axis is built. Eq. (1) defines t as a CDF rank of Tdust multiplied by γ, and Sect. 3.1 states that γ is calibrated so that Mvir ∝ e^t. Therefore Eq. (2) is not a derived prediction but a restatement of the calibration target. This is a direct case of a fitted input being presented as a discovered relation. The later maximum-core-mass relation (Eq. 6) uses the same t-axis and is an independent correlation in the sense that its slope is not fixed by the Mvir fit; however, it cannot independently validate the exponential timescale because the clock itself was set using Mvir. The luminosity fit (Eq. 14) and the CMF construction (Eq. 16) similarly operate on the same calibrated axis. I do not count the self-citation to Liu (2025) for the convolution tail-preservation lemma as circular: it is a simple mathematical property that can be checked independently, and the paper's main circular step is the gamma calibration, not that lemma. Overall, one central prediction is forced by construction, but the paper also contains separate fits to other data (Mmax_core, Lbol) that are not themselves derived from Eq. (2). Hence the score is 6 rather than 8: partial circularity, with the central exponential result being definitional in origin.

Assumptions & free parameters 7 free parameters · 7 assumptions · 0 invented entities

The central claims rest on a calibrated time axis (γ, t0), several fitted coefficients, and assumptions including monotonic T_dust, unbiased sampling, an exponential accretion pattern, and a self-cited convolution lemma. No new physical entities are introduced.

free parameters (7)

gamma (γ) = not quoted (set so Mvir=200 e^t)
Scaling factor in t=γ CDF(T_dust); calibrated to make the virial-mass trend exponential, so it directly embeds the central result.
characteristic timescale t0 = ~10^5 yr (undetermined)
Absolute normalization of t is left free and described as calibratable in future; all exponential statements are in units of t0.
Mvir normalization = 200 M_sun
Fitted constant in Eq. 2.
Mmax_core normalization = 0.9 M_sun
Fitted constant in Eq. 6.
luminosity coefficient a = 40 L_sun
Fitted in Eq. 14-15 for the e^t accretion component of Lbol.
luminosity coefficient b = 10 L_sun
Fitted in Eq. 14-15 for the e^{3.5t} stellar component.
CMF high-mass slope α = 1
Adopted fiducial from ALMA-IMF literature and used to relate Mmax_core to cluster mass; later 'reproduced' by the model.

assumptions (7)

domain assumption ATLASGAL sample is an unbiased, complete draw from one evolutionary sequence; T_dust is monotonic with age
Loaded into Eq. 1; if false the 'time' axis is just a temperature rank with selection bias.
ad hoc to paper Convolution preserves the shallowest exponential tail (Liu 2025)
Central CMF derivation (Eq. 16-19) relies on a self-cited arXiv result with no proof or formalization in this paper.
ad hoc to paper All cores accrete as Mdot_core = M_core and new cores form at an exponentially growing rate
Eq. 8-9; this is the exponential scenario already being tested, used to 'derive' the CMF.
domain assumption Constant ICMF for nascent cores
Needed for the CMF convolution; low-mass end noted as poorly constrained.
domain assumption ZAMS Lstar ∝ M^3.5; luminosity dominated by the most massive star; accretion luminosity ∝ M_SF
Eq. 10-13; ignores radius inflation and feedback; acknowledged in text.
domain assumption Star-forming mass M_SF can be identified with virial mass M_vir
Needed to map Eq. 2 to the SF law in Eq. 21; virial parameter variations are ignored.
domain assumption Stream width w ∝ M_SF^0.5 and mass flux ∝ M_SF
Eq. 22-23; plausibility argument imported from Jog 2013 / Li 2025, not directly measured here.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Accelerating star formation of dense clumps." pith.science (2026). https://pith.science/paper/QB4452SD

@misc{pith2026251025436,
  author       = {Pith},
  title        = {Pith review of: Accelerating star formation of dense clumps},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QB4452SD}},
  note         = {Machine review of arXiv:2510.25436}
}

read the original abstract

We present a statistical framework that establishes an accelerating star formation scenario for dense clumps using ATLASGAL and ALMAGAL samples. By employing the cumulative distribution function of dust temperature as a monotonic evolutionary indicator, we linearize clump evolution into a normalized timescale, enabling direct comparison across different samples. The virial mass of clumps increases exponentially with this normalized time, revealing an accelerating buildup of star-forming gas within protoclusters. The evolution of the maximum core mass further shows that the growth timescales of protoclusters and their embedded most massive protostars are comparable, implying a self-similar acceleration of star formation from the stellar to the protocluster scale. This unified framework naturally reproduces the observed evolution of luminosity, the core mass function, the mass growth of the most massive protostars, and the dense gas star formation law on clump scales, establishing a coherent picture of accelerating star formation across scales.

Figures

Figures reproduced from arXiv: 2510.25436 by the authors.

**Figure 1.** Distribution of ATLASGAL clumps (Urquhart et al. 2018). (a) Probability density function (PDF) of dust temperature (Tdust). (b) Cumulative distribution function (CDF) of Tdust. The two y-axes indicate the linear mapping between the CDF of Tdust and the evolutionary age (t). (c) Distribution of clump mass (Mcl) versus Tdust (blue dots); orange dots represent clumps with virial mass measurements. (d) Same as panel (c… view at source ↗

**Figure 2.** Upper: Distribution of log10(Mmax core ) as a function of t for the ALMAGAL sample. Black error bars indicate the mean and standard deviation of log10(Mmax core ) within each time bin, while blue error bars represent the standard error of the mean (i.e., stderr = std/ √ N, with N being the number of sources in each bin). The red line shows the exponential fit between Mmax core and t. Lower: Distribution of Lbol as a… view at source ↗

**Figure 3.** Upper: A schematic illustration showing how shifted ICMFs (colored curves) combine to produce a top-heavy CMF (black curve). The upper shifted ICMFs represent newly formed cores, while the accretion process shifts the mass function of older cores toward higher masses, forming right-shifted ICMFs. Lower: Cumulative functions of the observed CMFs for three time bins, together with the corresponding modeled CMFs (see … view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

Shared star formation in the Milky Way and Magellanic Clouds
astro-ph.GA 2026-07 conditional novelty 6.0 of 10

Milky Way and Magellanic Cloud dense clumps are physical analogs sharing a ~1 pc parent scale and hierarchical layout, yielding SFRs of ~0.4 and ~0.1 M⊙ yr⁻¹ for LMC and SMC.
How Should We Understand the Core Mass Function? A memo of the CMF2IMF conference at ESO Garching
astro-ph.GA 2026-07 conditional novelty 5.5 of 10

The high-mass slope of the core mass function depends strongly on the minimum fitting mass: completeness-based fits look top-heavy, while KS-selected tail fits move toward Salpeter, and the early-stage ASHES sample ap...

Reference graph

Works this paper leans on

24 extracted references · 1 linked inside Pith · cited by 2 Pith papers

[1]

A., Bate, M

Bonnell, I. A., Bate, M. R., Clarke, C. J., & Pringle, J. E. 2001, MNRAS, 323, 785

2001
[2]

A., Vine, S

Bonnell, I. A., Vine, S. G., & Bate, M. R. 2004, MNRAS, 349, 735

2004
[3]

C., Liu, M., et al

Cheng, Y ., Tan, J. C., Liu, M., et al. 2018, ApJ, 853, 160

2018
[4]

2025, A&A, 696, A151

Coletta, A., Molinari, S., Schisano, E., et al. 2025, A&A, 696, A151

2025
[5]

& Solomon, P

Gao, Y . & Solomon, P. M. 2004, ApJ, 606, 271 Article number, page 5 of 6 A&A proofs:manuscript no. AcceleratingSF

2004
[6]

S., et al

Heyer, M., Gutermuth, R., Urquhart, J. S., et al. 2016, A&A, 588, A29

2016
[7]

W., & Omukai, K

Hosokawa, T., Yorke, H. W., & Omukai, K. 2010, ApJ, 721, 478

2010
[8]

Jog, C. J. 2013, Monthly Notices of the Royal Astronomical Society: Letters, 434, L56

2013

Show all 24 references

[9]

Kennicutt, Jr., R. C. 1998, ApJ, 498, 541

1998
[10]

2025, arXiv e-prints, arXiv:2509.18495

Li, G.-X. 2025, arXiv e-prints, arXiv:2509.18495

2025
[11]

2025, arXiv e-prints, arXiv:2502.20458

Liu, X. 2025, arXiv e-prints, arXiv:2502.20458

2025 arXiv
[12]

2025, Research in Astronomy and Astrophysics, 25, 025020

Liu, X., Liu, T., Mai, X., et al. 2025, Research in Astronomy and Astrophysics, 25, 025020

2025
[13]

N., Bally, J., Testi, L., et al

Longmore, S. N., Bally, J., Testi, L., et al. 2012, Monthly Notices of the Royal Astronomical Society, 429, 987

2012
[14]

2024, A&A, 690, A33

Louvet, F., Sanhueza, P., Stutz, A., et al. 2024, A&A, 690, A33

2024
[15]

McKee, C. F. & Tan, J. C. 2003, ApJ, 585, 850

2003
[16]

2025, A&A, 696, A149

Molinari, S., Schilke, P., Battersby, C., et al. 2025, A&A, 696, A149

2025
[17]

1998, A&A, 336, 150

Motte, F., Andre, P., & Neri, R. 1998, A&A, 336, 150

1998
[18]

& Stahler, S

Palla, F. & Stahler, S. W. 1991, ApJ, 375, 288

1991
[19]

2022, A&A, 664, A26

Pouteau, Y ., Motte, F., Nony, T., et al. 2022, A&A, 664, A26

2022
[20]

Salpeter, E. E. 1955, ApJ, 121, 161

1955
[21]

M., Contreras, Y ., et al

Schuller, F., Menten, K. M., Contreras, Y ., et al. 2009, A&A, 504, 415

2009
[22]

2017, A&A, 604, A74 Suárez, G., Galván-Madrid, R., Aguilar, L., et al

Shimajiri, Y ., André, P., Braine, J., et al. 2017, A&A, 604, A74 Suárez, G., Galván-Madrid, R., Aguilar, L., et al. 2021, ApJ, 921, 48

2017
[23]

S., König, C., Giannetti, A., et al

Urquhart, J. S., König, C., Giannetti, A., et al. 2018, MNRAS, 473, 1059

2018
[24]

S., Wells, M

Urquhart, J. S., Wells, M. R. A., Pillai, T., et al. 2022, MNRAS, 510, 3389 Article number, page 6 of 6

2022

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [1]

A., Bate, M

Bonnell, I. A., Bate, M. R., Clarke, C. J., & Pringle, J. E. 2001, MNRAS, 323, 785

2001

[2] [2]

A., Vine, S

Bonnell, I. A., Vine, S. G., & Bate, M. R. 2004, MNRAS, 349, 735

2004

[3] [3]

C., Liu, M., et al

Cheng, Y ., Tan, J. C., Liu, M., et al. 2018, ApJ, 853, 160

2018

[4] [4]

2025, A&A, 696, A151

Coletta, A., Molinari, S., Schisano, E., et al. 2025, A&A, 696, A151

2025

[5] [5]

& Solomon, P

Gao, Y . & Solomon, P. M. 2004, ApJ, 606, 271 Article number, page 5 of 6 A&A proofs:manuscript no. AcceleratingSF

2004

[6] [6]

S., et al

Heyer, M., Gutermuth, R., Urquhart, J. S., et al. 2016, A&A, 588, A29

2016

[7] [7]

W., & Omukai, K

Hosokawa, T., Yorke, H. W., & Omukai, K. 2010, ApJ, 721, 478

2010

[8] [8]

Jog, C. J. 2013, Monthly Notices of the Royal Astronomical Society: Letters, 434, L56

2013