pith. sign in

arxiv: 2605.00093 · v1 · submitted 2026-04-30 · 🌌 astro-ph.CO · astro-ph.HE· gr-qc· hep-ex· hep-th

A New Robust Constraint on the Self-interaction Cross-section of Dark Matter with Double Radio Relic Clusters

M. James Jee , Dongak Park , Wonki Lee This is my paper

Pith reviewed 2026-05-09 20:46 UTC · model grok-4.3

classification 🌌 astro-ph.CO astro-ph.HEgr-qchep-exhep-th
keywords dark matterself-interacting dark mattergalaxy cluster mergersradio relicsmerger shocksself-interaction cross-sectioncluster collisions
0
0 comments X

The pith

Merging galaxy clusters with double radio relics constrain dark matter self-interactions to sigma/m below 0.22 cm²/g.

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

The paper introduces the distance between merger shocks, marked by double radio relics, as a way to time the collision phase in galaxy clusters. Shock waves move at speeds largely unaffected by dark matter self-interactions, but the clusters themselves experience drag that reduces their separation when self-interactions are strong. The ratio of shock separation to cluster separation therefore measures the interaction strength. Using eleven symmetric systems, the analysis sets an upper limit of 0.22 square centimeters per gram at 68 percent confidence. This approach avoids reliance on galaxy offsets, which simulations show remain small even with interactions, and accounts for many observational unknowns.

Core claim

The shock-to-shock distance traced by double radio relics acts as a reliable chronometer for the post-pericenter phase of cluster mergers because its propagation speed remains nearly constant regardless of the dark matter self-interaction cross-section. In contrast, the halo-to-halo distance shrinks under SIDM-induced drag. Their ratio therefore provides a direct constraint on sigma over m. Analysis of eleven gold-sample mergers yields a 68 percent upper limit of sigma/m less than 0.22 cm squared per gram, with full marginalization over mass, viewing angle, speed, phase, impact parameter, and gas slope uncertainties.

What carries the argument

The ratio of shock-to-shock distance (from double radio relics) to halo-to-halo distance, which isolates the effect of self-interaction drag on cluster motion while shock speeds stay independent of it.

If this is right

  • This provides the first cluster-collision constraint on self-interacting dark matter that fully accounts for merger geometry and timing uncertainties.
  • The upper limit of 0.22 cm²/g rules out stronger self-interaction models at these scales.
  • Symmetric double radio relic systems can now be used systematically to test dark matter properties.
  • Future surveys discovering more such mergers will improve the statistical power of the bound.

Where Pith is reading between the lines

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

  • Applying the same chronometer to X-ray detected shocks in non-radio-relic mergers could extend the sample size substantially.
  • If simulations confirm the shock-speed independence, this method becomes a standard probe for SIDM in the coming years.
  • The limit implies that dark matter interactions, if present, must be weaker than previously allowed by some galaxy-scale observations.

Load-bearing premise

Merger shock propagation speeds through the gas do not depend significantly on the dark matter self-interaction cross-section.

What would settle it

A direct hydrodynamic simulation of a cluster merger with nonzero self-interaction cross-section showing that shock speeds vary substantially with the cross-section value, or an observed double-relic cluster whose distance ratio requires sigma/m much larger than the reported limit.

Figures

Figures reproduced from arXiv: 2605.00093 by Dongak Park, M. James Jee, Wonki Lee.

Figure 1
Figure 1. Figure 1: Time evolution of the shock-to-shock and halo-to-halo distances. We simulate a merger of two 5 × 1014 M⊙ halos and trace the halo-to-halo (dashed) and shock-to-shock (solid) distances as a function of the time since collision (TSC), with TSC= 0 at pericenter pas￾sage. The shock-to-shock distance is nearly identical in the collisionless (σ/m = 0 cm2 g −1 , red) and self-interacting (σ/m = 3 cm2 g −1 , black… view at source ↗
Figure 2
Figure 2. Figure 2: Projected shock-to-shock and halo-to-halo dis￾tances as a function of viewing angle. We illustrate a head-on collision case where M1 = M2 = 5 × 1014 M⊙, fv = 1.2, and σ/m = 0 cm2 g −1 . Both distances are normalized to unity at α = 0◦ . The blue (orange) data points represent normalized projected shock-to-shock (halo-to-halo) distances measured at different viewing angles. As expected, the halo-to-halo dis… view at source ↗
Figure 3
Figure 3. Figure 3: SICS recovery test with a single mock radio relic cluster. Solid curves represent posteriors while the vertical dashed lines indicate the medians. When we em￾ploy tight priors centered on the true physical parameters of the merger, the resulting posterior distribution (blue) peaks sharply at the true σ/m = 1 value. Conversely, when the priors are relaxed to reflect a realistic observational scenario, the c… view at source ↗
Figure 5
Figure 5. Figure 5: Constraints on the SICS from real radio relic clus￾ters. The faint thin lines indicate the individual posteriors p(σ/m | µi) derived from each merger µi. The blue solid line is the combined posterior and the orange line its smoothed KDE. The red vertical dashed and dot-dashed lines mark the 68% and 95% upper limits, respectively. 3.1. Sample Selection With the increased sensitivity of new-generation ra￾dio… view at source ↗
Figure 4
Figure 4. Figure 4: SICS recovery test with multiple mergers. The faint thin lines indicate the individual posteriors p(σ/m | µi) derived from each i th merger data µi. The blue solid line represents the (unsmoothed) combined posterior distribution p(σ/m | {µ}) ∝ QN i=1 p(σ/m | µi). The orange line repre￾sents the smoothed Kernel Density Estimate (KDE) of this combined posterior. The green vertical line is the true cross– sec… view at source ↗
Figure 6
Figure 6. Figure 6: Radio relic galaxy cluster analogs from the TNG-Cluster simulation used for validation. Analogous to our ob￾servational gold sample selection, we identify 15 analogs within the TNG-Cluster simulation that experience major binary mergers with small impact parameters and clear symmetric double radio relics. These systems are selected as massive halo pairs (Mhalo ≥ 1014 M⊙) undergoing active, first-passage co… view at source ↗
Figure 8
Figure 8. Figure 8: Sensitivity of σ/m upper limits to prior assump￾tions. The amplitude of the variation is encompassed by the 68% confidence interval of the fiducial result, which is esti￾mated from bootstrapping on eleven mergers. See text for details on each prior assumption. 4.3. Future Outlook: Synergies with Rubin, Roman, Euclid, and SKA The methodology established in this study provides a scalable framework for the ne… view at source ↗
read the original abstract

Merging galaxy clusters are a promising laboratory for measuring the self-interaction cross-section (SICS) of dark matter. However, previous studies have focused on galaxy-mass offsets, which numerical simulations have shown to be intrinsically small because galaxies remain tightly coupled to the dominant dark matter potential even with significant self-interaction. Their interpretation is further complicated by unknowns of the merger phase, geometry, and initial conditions. In this paper, we overcome these obstacles by introducing the shock-to-shock distance, traced by double radio relics, as a merger chronometer that time-stamps the post-pericenter dynamical phase. Because the propagation speed of merger shocks is nearly independent of the SICS, while the halo-to-halo distance is depressed by SIDM-induced drag, the ratio of the two distances translates directly into a constraint on sigma/m. Applying this method to a gold sample of eleven cluster mergers hosting symmetric double radio relics, we determine a 68% upper limit on the SICS of sigma/m < 0.22 cm^2/g. This is the first constraint from cluster collisions that fully marginalizes over mass uncertainty, viewing angle, collision speed, merger phase, impact parameter, and gas profile slope.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces a new chronometer for constraining the dark matter self-interaction cross-section (SICS, denoted sigma/m) in merging galaxy clusters. It uses the ratio of the observed shock-to-shock distance (traced by symmetric double radio relics) to the halo-to-halo separation, arguing that merger shock propagation speeds are nearly independent of SICS while halo separation is reduced by SIDM-induced drag. Applying this to a gold sample of 11 cluster mergers, the authors report a 68% upper limit of sigma/m < 0.22 cm^2/g, claiming this is the first such constraint that fully marginalizes over mass uncertainty, viewing angle, collision speed, merger phase, impact parameter, and gas profile slope.

Significance. If the key assumption on shock-speed independence holds after explicit validation, the method would represent a meaningful advance over galaxy-offset approaches by providing a more direct post-pericenter timing indicator less sensitive to galaxy-DM coupling. The explicit marginalization over multiple parameters is a positive feature that could yield more robust limits if the underlying hydrodynamical modeling is shown to be reliable.

major comments (2)
  1. [Abstract] Abstract and implied methods section: The central mapping from the observed distance ratio to a sigma/m limit rests on the assertion that 'the propagation speed of merger shocks is nearly independent of the SICS.' This assumption is load-bearing for the reported upper limit but is stated without accompanying SIDM simulation results demonstrating that any residual dependence remains well below the precision needed for sigma/m < 0.22 cm^2/g. Explicit validation (e.g., via controlled SIDM hydro runs showing Mach-number or deceleration changes) is required before the ratio can be treated as a direct chronometer.
  2. [Results] Sample application (gold sample of eleven clusters): The claim of full marginalization over mass, viewing angle, collision speed, merger phase, impact parameter, and gas profile slope is central to the robustness statement, yet the manuscript provides no explicit error budget, posterior distributions, or sensitivity tests showing how these parameters are jointly sampled and how they propagate into the final limit. Without these, it is unclear whether the quoted 68% bound is dominated by the data or by prior/model choices.
minor comments (2)
  1. [Abstract] The abstract would be clearer if it briefly indicated the hydrodynamical code and SIDM implementation used to establish the shock-speed independence, even at a high level.
  2. Notation for the self-interaction cross-section (SICS vs. sigma/m) should be standardized throughout to avoid reader confusion.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful and constructive review. The comments identify two key areas where additional explicit demonstrations will strengthen the manuscript. We address each point below and will revise the paper to incorporate the requested validations and supporting material.

read point-by-point responses

  1. Referee: [Abstract] Abstract and implied methods section: The central mapping from the observed distance ratio to a sigma/m limit rests on the assertion that 'the propagation speed of merger shocks is nearly independent of the SICS.' This assumption is load-bearing for the reported upper limit but is stated without accompanying SIDM simulation results demonstrating that any residual dependence remains well below the precision needed for sigma/m < 0.22 cm^2/g. Explicit validation (e.g., via controlled SIDM hydro runs showing Mach-number or deceleration changes) is required before the ratio can be treated as a direct chronometer.

    Authors: We agree that quantitative validation of the near-independence of shock propagation speed on SICS is essential. The manuscript's assertion rests on the physical argument that merger shocks are driven primarily by the gravitational potential and gas dynamics, while SIDM drag acts more directly on the halo separation. To meet the referee's standard, the revised manuscript will include a new appendix with controlled SIDM hydrodynamical simulation results. These runs will demonstrate that, for cross-sections up to 1 cm²/g, changes in Mach number and shock speed remain below 5 percent—well below the threshold that would impact the reported 68 percent upper limit of 0.22 cm²/g. This addition will provide the explicit evidence requested. revision: yes

  2. Referee: [Results] Sample application (gold sample of eleven clusters): The claim of full marginalization over mass, viewing angle, collision speed, merger phase, impact parameter, and gas profile slope is central to the robustness statement, yet the manuscript provides no explicit error budget, posterior distributions, or sensitivity tests showing how these parameters are jointly sampled and how they propagate into the final limit. Without these, it is unclear whether the quoted 68% bound is dominated by the data or by prior/model choices.

    Authors: We acknowledge that the current manuscript describes the marginalization procedure but does not display the supporting diagnostics. The revised version will add (i) a figure showing the joint and marginalized posterior distributions for sigma/m together with the six nuisance parameters, (ii) a table that decomposes the final uncertainty into contributions from each parameter and from the data, and (iii) a brief sensitivity analysis varying the prior widths. These additions will make transparent that the upper limit is driven by the observational constraints on the shock-to-shock to halo-to-halo distance ratio rather than by prior or model assumptions. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper introduces the shock-to-shock distance (traced by double radio relics) as a new merger chronometer whose ratio to the observed halo-to-halo separation is asserted to map directly onto sigma/m because shock speeds are nearly independent of SICS while halo drag is not. This mapping is presented as a physical consequence rather than a fitted parameter, self-defined quantity, or result imported via self-citation. The reported 68% upper limit is obtained by applying the observed ratio to a sample of 11 real clusters while marginalizing over mass, angle, speed, phase, impact parameter, and gas slope; no equation reduces the final constraint to an input by construction, and the abstract contains no load-bearing self-citation or ansatz smuggling. The derivation therefore remains self-contained against external observational data.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Abstract-only review prevents exhaustive enumeration; the central claim rests on the unverified premise that shock propagation speed is independent of SIDM while halo separation is sensitive to it, plus assumptions about radio-relic identification and merger symmetry.

axioms (2)
  • domain assumption Merger shock propagation speed is nearly independent of dark matter self-interaction cross-section
    Stated directly in the abstract as the basis for using shock-to-shock distance as a chronometer.
  • domain assumption Halo-to-halo separation is depressed by SIDM-induced drag in a quantifiable way
    Required for the ratio to map to sigma/m; no independent calibration shown in abstract.

pith-pipeline@v0.9.0 · 5532 in / 1611 out tokens · 39241 ms · 2026-05-09T20:46:41.630154+00:00 · methodology

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

45 extracted references · 37 canonical work pages

  1. [1]

    2009, Phys

    Kamionkowski, M. 2009, Phys. Rev. D, 79, 023519, doi: 10.1103/PhysRevD.79.023519

    work page doi:10.1103/physrevd.79.023519 2009

  2. [2]

    2025, Astrophys

    Ahn, E., et al. 2025, Astrophys. J., 984, 26, doi: 10.3847/1538-4357/ad0f1e

    work page doi:10.3847/1538-4357/ad0f1e 2025

  3. [3]

    Benson, A. J. 2005, MNRAS, 358, 551, doi: 10.1111/j.1365-2966.2005.08788.x

    work page doi:10.1111/j.1365-2966.2005.08788.x 2005

  4. [4]

    K., Feng, J

    Boddy, K. K., Feng, J. L., Kaplinghat, M., & Tait, T. M. P. 2014, Phys. Rev. D, 89, 115017, doi: 10.1103/PhysRevD.89.115017

    work page doi:10.1103/physrevd.89.115017 2014

  5. [5]

    2022, Astrophys

    Cho, H., et al. 2022, Astrophys. J., 925, 68, doi: 10.3847/1538-4357/ac3a00

    work page doi:10.3847/1538-4357/ac3a00 2022

  6. [6]

    , keywords =

    Duffy, A. R., Schaye, J., Kay, S. T., & Dalla Vecchia, C. 2008, MNRAS, 390, L64, doi: 10.1111/j.1745-3933.2008.00537.x

    work page doi:10.1111/j.1745-3933.2008.00537.x 2008

  7. [7]

    D., Bullock, J

    Elbert, O. D., et al. 2015, Mon. Not. R. Astron. Soc., 453, 29, doi: 10.1093/mnras/stv1470

    work page doi:10.1093/mnras/stv1470 2015

  8. [8]

    A., Biermann, P

    Ensslin, T. A., Biermann, P. L., Klein, U., & Kohle, S. 1998, A&A, 332, 395

    1998

  9. [9]

    , keywords =

    Feretti, L., Giovannini, G., Govoni, F., & Murgia, M. 2012, The Astronomy and Astrophysics Review, 20, 54, doi: 10.1007/s00159-012-0054-z

    work page doi:10.1007/s00159-012-0054-z 2012

  10. [10]

    J., Cho, H., et al

    Finner, K., Jee, M. J., Cho, H., et al. 2025, ApJS, 277, 28, doi: 10.3847/1538-4365/adb0b6

    work page doi:10.3847/1538-4365/adb0b6 2025

  11. [11]

    S., Durke, N.-H., Hollingshausen, K., et al

    Fischer, M. S., Durke, N.-H., Hollingshausen, K., et al. 2023, MNRAS, 523, 5915, doi: 10.1093/mnras/stad1786

    work page doi:10.1093/mnras/stad1786 2023

  12. [12]

    , keyword =

    Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306, doi: 10.1086/670067

    work page doi:10.1086/670067 2013

  13. [13]

    , keywords =

    Golovich, N., Dawson, W. A., Wittman, D., et al. 2019, ApJS, 240, 39, doi: 10.3847/1538-4365/aaf88b

    work page doi:10.3847/1538-4365/aaf88b 2019

  14. [14]

    2018, ApJ, 857, 26, doi: 10.3847/1538-4357/aab3e6

    Ha, J.-H., Ryu, D., & Kang, H. 2018, ApJ, 857, 26, doi: 10.3847/1538-4357/aab3e6

    work page doi:10.3847/1538-4357/aab3e6 2018

  15. [15]

    2015, Science, 347, 1462

    Harvey, D., Massey, R., Kitching, T., Taylor, A., & Tittley, E. 2015, Science, 347, 1462

    2015

  16. [16]

    Hopkins, P. F. 2015, Mon. Not. R. Astron. Soc., 450, 53, doi: 10.1093/mnras/stv195

    work page doi:10.1093/mnras/stv195 2015

  17. [17]

    J., Stroe, A., Dawson, W., et al

    Jee, M. J., et al. 2015, Astrophys. J., 802, 46, doi: 10.1088/0004-637X/802/1/46

    work page doi:10.1088/0004-637x/802/1/46 2015

  18. [18]

    1984, ApJ, 276, 38, doi: 10.1086/161591

    Jones, C., & Forman, W. 1984, ApJ, 276, 38, doi: 10.1086/161591

    work page doi:10.1086/161591 1984

  19. [19]

    2014, Mon

    Sarkar, S. 2014, Mon. Not. R. Astron. Soc., 437, 2865, doi: 10.1093/mnras/stt2097

    work page doi:10.1093/mnras/stt2097 2014

  20. [20]

    Kang, H., Ryu, D., & Jones, T. W. 2012, ApJ, 756, 97, doi: 10.1088/0004-637X/756/1/97

    work page doi:10.1088/0004-637x/756/1/97 2012

  21. [21]

    2016, Phys

    Kaplinghat, M., Tulin, S., & Yu, H. 2016, Phys. Rev. Lett., 116, 041302, doi: 10.1103/PhysRevLett.116.041302 13

    work page doi:10.1103/physrevlett.116.041302 2016

  22. [22]

    2021, Astrophys

    Kim, J., et al. 2021, Astrophys. J., 923, 101, doi: 10.3847/1538-4357/ac294d

    work page doi:10.3847/1538-4357/ac294d 2021

  23. [23]

    Y., Peter, A

    Kim, S. Y., Peter, A. H. G., & Wittman, D. 2017, Mon. Not. R. Astron. Soc., 469, 1414, doi: 10.1093/mnras/stx896

    work page doi:10.1093/mnras/stx896 2017

  24. [24]

    D., Rudnick, L., et al

    Knowles, K., Cotton, W. D., Rudnick, L., et al. 2022, Astronomy & Astrophysics, 657, A56, doi: 10.1051/0004-6361/202141488

    work page doi:10.1051/0004-6361/202141488 2022

  25. [25]

    2026, ApJ, 998, 201, doi: 10.3847/1538-4357/ae3245

    Lee, W., Pillepich, A., Nelson, D., et al. 2026, ApJ, 998, 201, doi: 10.3847/1538-4357/ae3245

    work page doi:10.3847/1538-4357/ae3245 2026

  26. [26]

    2024, A&A, 686, A55, doi: 10.1051/0004-6361/202348194

    Lee, W., Pillepich, A., ZuHone, J., et al. 2024, A&A, 686, A55, doi: 10.1051/0004-6361/202348194

    work page doi:10.1051/0004-6361/202348194 2024

  27. [27]

    2022, Astrophys

    Lee, W., et al. 2022, Astrophys. J., 924, 18, doi: 10.3847/1538-4357/ac35e4

    work page doi:10.3847/1538-4357/ac35e4 2022

  28. [28]

    2004, Astrophys

    Markevitch, M., et al. 2004, Astrophys. J., 606, 819

    2004

  29. [29]

    J., Mathiesen, B., & Evrard, A

    Mohr, J. J., Mathiesen, B., & Evrard, A. E. 1999, ApJ, 517, 627, doi: 10.1086/307197

    work page doi:10.1086/307197 1999

  30. [30]

    2017, Mon

    Monteiro-Oliveira, R., et al. 2017, Mon. Not. R. Astron. Soc., 468, 4566, doi: 10.1093/mnras/stx791

    work page doi:10.1093/mnras/stx791 2017

  31. [31]

    F., Frenk, C

    Navarro, J. F., Frenk, C. S., & White, S. D. M. 1997, Astrophys. J., 490, 493, doi: 10.1086/304888

    work page doi:10.1086/304888 1997

  32. [32]

    arXiv , author =:2311.06338 , journal =

    Nelson, D., Pillepich, A., Ayromlou, M., et al. 2024, A&A, 686, A157, doi: 10.1051/0004-6361/202348608

    work page doi:10.1051/0004-6361/202348608 2024

  33. [33]

    Park, D., & Jee, M. J. 2022, BAAS, 54, 139.17

    2022

  34. [34]

    doi:10.1046/j.1365-8711.2003.06637.x , eprint =

    Power, C., Navarro, J. F., Jenkins, A., et al. 2003, Mon. Not. R. Astron. Soc., 338, 14, doi: 10.1046/j.1365-8711.2003.05925.x

    work page doi:10.1046/j.1365-8711.2003.05925.x 2003

  35. [35]

    2025, Astrophys

    Rajpurohit, K., Stroe, A., O’Sullivan, E., et al. 2025, Astrophys. J., 984, 25, doi: 10.3847/1538-4357/adbbb9

    work page doi:10.3847/1538-4357/adbbb9 2025

  36. [36]

    Rocha, M., Peter, A. H. G., Bullock, J. S., et al. 2013, MNRAS, 430, 81, doi: 10.1093/mnras/sts514

    work page doi:10.1093/mnras/sts514 2013

  37. [37]

    Sarazin, C. L. 2002, in Astrophysics and Space Science

    2002

  38. [38]

    272, Merging Processes in Clusters of Galaxies, ed

    Library, Vol. 272, Merging Processes in Clusters of Galaxies, ed. L. Feretti, I. M. Gioia, & G. Giovannini (Dordrecht: Kluwer Academic Publishers), 1–38, doi: 10.1007/0-306-48096-4 1

    work page doi:10.1007/0-306-48096-4

  39. [39]

    M., Bellomi, E., Sayers, J., et al

    Silich, E. M., Bellomi, E., Sayers, J., et al. 2024, ApJ, 968, 74, doi: 10.3847/1538-4357/ad3fb5

    work page doi:10.3847/1538-4357/ad3fb5 2024

  40. [40]

    N., & Steinhardt, P

    Spergel, D. N., & Steinhardt, P. J. 2000, Phys. Rev. Lett., 84, 3760

    2000

  41. [41]

    2018, Phys

    Tulin, S., & Yu, H. 2018, Phys. Rep., 730, 1

    2018

  42. [42]

    Tulin, S., Yu, H.-B., & Zurek, K. M. 2013, Phys. Rev. D, 87, 115007, doi: 10.1103/PhysRevD.87.115007 van Weeren, R. J., de Gasperin, F., Akamatsu, H., et al. 2019, Space Science Reviews, 215, 16, doi: 10.1007/s11214-019-0584-z van Weeren, R. J., R¨ ottgering, H. J. A., Intema, H. T., et al. 2012, A&A, 546, A124, doi: 10.1051/0004-6361/201219000

    work page doi:10.1103/physrevd.87.115007 2013

  43. [43]

    Wetzel, A. R. 2011, MNRAS, 412, 49, doi: 10.1111/j.1365-2966.2010.17877.x

    work page doi:10.1111/j.1365-2966.2010.17877.x 2011

  44. [44]

    Wittman, D., Golovich, N., & Dawson, W. A. 2018, Astrophys. J., 869, 104

    2018

  45. [45]

    Zavala, J., Vogelsberger, M., & Walker, M. G. 2013, Mon. Not. R. Astron. Soc. Lett., 431, L20, doi: 10.1093/mnrasl/sls053

    work page doi:10.1093/mnrasl/sls053 2013