Pith. sign in

REVIEW 3 major objections 5 minor 55 references

Cosmic-Ray Feedback from Supernovae in a Parker-Unstable Medium

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Putting 10% of supernova energy into cosmic rays, instead of all into heat and motion, drives a faster, hotter, and more massive outflow from a galactic disk.

desk verdict Solid, honest simulation comparison, but the headline CR-vs-thermal outflow contrast rests on one run per setup, so the numbers are plausible, not proven. read the letter →

arxiv 2412.12249 v2 pith:T3KVXI5Y submitted 2024-12-16 astro-ph.GA astro-ph.HE

Roark Habegger , Ellen G. Zweibel This is my paper · ORCID
classification astro-ph.GAastro-ph.HE
keywords cosmicraysParkerinstabilitygalacticwindssupernovafeedbackmagnetohydrodynamicsimulationsinterstellarmediumcalorimetricfractiongamma-rayluminosity
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

Supernovae are thought to put about 10% of their energy into cosmic rays, yet many feedback models put all of it into heat and gas motion. This paper asks what difference that 10% makes and argues the difference is large: cosmic-ray injections drive a faster, hotter, and more massive outflow that persists long after individual injection events. The same simulations show that the Parker instability, the magnetic buoyancy instability of a stratified, magnetized gas, builds cold clouds, decorrelates cosmic-ray pressure from gas density, and reorients the magnetic field vertically so cosmic rays escape and the galaxy produces fewer gamma rays. A sympathetic reader would take away that the non-thermal 10% is not a small correction but a main driver of how gas leaves the disk.

What carries the argument

The load-bearing setup is a galactic-patch magnetohydrodynamic simulation with a separate cosmic-ray fluid that diffuses and streams along the resolved magnetic field. The central physical object is the Parker instability, the magnetic buoyancy instability of a stratified, magnetized atmosphere, which converts the initial horizontal field into vertical loops, plumes, and cold valleys. The argument is carried by the cosmic-ray pressure gradient: after the instability reorients the field, cosmic rays stream outward at the Alfvén speed, and their pressure gradient becomes the dominant vertical force at $|z| \gtrsim 1$ kpc, accelerating gas out of the disk.

What would settle it

A repetition of the setup with vertical boundaries that admit galactic-halo inflow would settle the central claim: if the long-lived cosmic-ray-driven outflow disappears or reverses when gas falls back in, the headline result fails.

Watch

Extended reading notes

Core claim

The central claim is that depositing 10% of each supernova's energy into a cosmic-ray fluid, rather than into thermal and kinetic energy alone, changes the long-term behavior of the interstellar medium. In two otherwise identical simulations of a stratified galactic patch, the run with cosmic-ray injection develops a steady outflow that begins at $|z| \simeq 0.5$ kpc, is faster and hotter, and is still present at late times, while the run without injection shows a weaker, fluctuating flow. In the steady state the cosmic-ray pressure gradient dominates vertical acceleration at $|z| \gtrsim 1$ kpc. This happens because the Parker instability overturns the gas and makes the magnetic field predominantly vertical outside the midplane; cosmic rays stream along those field lines, heat the diffuse gas, and push it outward. The same geometry lets cosmic rays escape before they collide with gas, lowering the calorimetric fraction, provided the magnetic field is resolved well enough to avoid numerical reconnection.

Load-bearing premise

The vertical boundaries let gas leave the simulated patch but never fall back in, so any real inflow from the halo that would oppose or dilute the wind is excluded.

Editorial extensions

If this is right

  • A cosmic-ray energy share as small as 10% cannot be dropped from models of the disk-halo interface; simulations that inject only thermal and kinetic energy will understate the speed and mass loss of outflows.
  • Resolved vertical magnetic field connecting the midplane to the halo acts as an escape route for cosmic rays, so galaxy-scale simulations may overpredict gamma-ray luminosity unless they resolve the magnetic structure of diffuse gas.
  • The Parker instability can produce a two-component vertical structure and cold clouds containing more than half the gas mass, with mass-loading factors above $10^3$ even at a low supernova rate.
  • A cosmic-ray-driven outflow can persist for more than 100 million years and reach down to $|z| \simeq 0.5$ kpc, forming the base of a large-scale galactic wind.
  • Where streaming dominates, cosmic-ray transport behaves like advection rather than diffusion, which steepens pressure gradients and changes the effective polytropic index of the cosmic-ray fluid.

Reading between the lines

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

  • If vertical field lines are the main escape highways, then a galaxy's gamma-ray and wind properties may be controlled by magnetic topology rather than by the cosmic-ray diffusion coefficient; this could be tested by relating observed gamma-ray luminosity to radio-polarization maps of halo fields.
  • The diode boundary condition likely makes the sustained outflow an upper limit; allowing halo gas to fall back in could weaken or shorten the cosmic-ray-driven phase.
  • The resolution dependence suggests coarser galaxy simulations need a subgrid prescription for field-aligned cosmic-ray escape, not just an increased diffusion coefficient.
  • Because the Parker instability puts cold clouds in low cosmic-ray-pressure valleys, star-forming gas may be shielded from cosmic-ray pressure; this predicts a spatial anti-correlation between dense gas and cosmic-ray pressure that synchrotron and gamma-ray observations could check.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper presents two Athena++ MHD+CR simulations of a stratified galactic patch (1 kpc × 1 kpc × 4.8 kpc, 5 pc resolution) that are identical except for the partitioning of supernova energy: in CRInj, 10% of each SN's energy is injected as cosmic-ray energy, while in CRBkg all injected energy is thermal and kinetic. The authors report that cosmic-ray injections produce a faster, hotter, and more massive outflow that persists for over 100 Myr after the initial transients, with cosmic-ray pressure gradients dominating the vertical acceleration at |z| ≳ 1 kpc during the steady state. They also analyze the Parker instability, cold-cloud formation, the decorrelation of cosmic-ray pressure from gas density, and the resulting reduction in the hadronic calorimetric fraction, which they compare to the starburst estimates of Lacki et al. (2011). The central claim is comparative: the two simulations differ only in the injection prescription, and the late-time outflow contrast is attributed to cosmic-ray pressure gradients.

Significance. If the result holds, it is an important contribution to the debate on cosmic-ray feedback in the multiphase ISM. The study's design is strong: identical initial conditions and injection rates, standard transport coefficients (diffusion coefficient from the B/C ratio, streaming at the Alfvén speed), a linear-theory check of the Parker instability growth rate, and an external observational benchmark for the calorimetric fraction. The 5 pc resolution and the inclusion of both diffusion and streaming are notable strengths. The main weaknesses are statistical power (one realization per setup), a likely typographical error in the printed cooling function, and an unacknowledged boundary-condition effect on the persistence of the outflow.

major comments (3)
  1. [§3.3, Figs. 10–13; §2.3] The headline comparative claim—that cosmic-ray injections drive a faster, hotter, and more massive outflow long after injection—rests on a single realization of each setup. The supernova injection schedule is Poisson-random (Section 2.3), and Figure 4 shows a quasi-periodic fountain cycle with a ~30 Myr period, so the t = 260–400 Myr steady state contains only about four to five independent cycles. Figures 10 and 13 show substantial time variation within this window, yet no estimate of realization-to-realization variance is provided. If the two runs use the same random seed, the comparison is still not a controlled pairing after the first supernova, because the local gas state into which each SN injects differs between the runs. The reported factor-of-three acceleration enhancement and the ≥95% cosmic-ray-gradient dominance could therefore be altered by a different seed. The authors should add either multiple realizations with different seeds or a quantitative bootstrap over independent time windows to demonstrate that the CR-vs-thermal contrast is not noise.
  2. [§2.2, Eq. (18)] As printed, the cooling function Λ(T) has prefactors 7.3×10^21 and 7.9×10^27 erg cm^3 s^-1. Substituting these into Equation (17) gives equilibrium densities many orders of magnitude below the range shown in Figure 2; for example, at T = 10^4 K the second term alone is ~10^28 erg cm^3 s^-1, which is physically impossible for the ISM. The standard Inoue et al. (2006) form uses negative exponents (10^-21 and 10^-27). This appears to be a sign typo, but it is load-bearing because the printed equation is inconsistent with the equilibrium curve in Figure 2 and would prevent reproduction of the simulations. The authors must correct the coefficients and confirm that the simulations used the correct form.
  3. [§2.1, §4.3, §5 conclusion item 4] The z-boundary 'diode' condition allows outflow but no inflow, so the simulation cannot represent gas falling back from the halo or cosmological accretion. Since one of the key conclusions is that the cosmic-ray-driven outflow survives for more than 100 Myr, the absence of any return flow may artificially enhance both the persistence and the mass of the wind. The paper should state this limitation explicitly and discuss how realistic inflows could alter the result; at minimum, the caveat belongs in Section 4.3 alongside the other acknowledged simplifications.
minor comments (5)
  1. [§3.2, Fig. 8] The measured exponential growth rate of vertical kinetic and magnetic energy (τ ≈ 18 Myr) is compared directly with the linear-theory Parker growth rate (τ ≈ 32 Myr), but the measured quantity includes ongoing supernova driving and nonlinear effects; the comparison should be phrased more cautiously or the fitting interval and decomposition should be justified.
  2. [§3.3, Figs. 11–13] The acceleration decomposition omits magnetic tension, with the text noting that the xy-plane averaged tension is predominantly zero. It would be helpful to state the magnitude of the tension variation (about ±0.2 km s^-1 Myr^-1) in the figure captions or in the main text where the decomposition is introduced.
  3. [Table 1 and Fig. 3] The units in Table 1 are inconsistent: Pc is listed in eV cm^-3, while Figure 3 plots pressures in units of 10^4 K cm^-3. Please unify the pressure units across the paper.
  4. [§4.1, Fig. 19] The resolution comparison for the calorimetric fraction compares only 5 pc and 10 pc runs; the text concludes that resolution is 'the reason' earlier work overpredicted γ-ray luminosity, but a two-point comparison cannot establish convergence. A more cautious statement, or an additional intermediate resolution, would be appropriate.
  5. [Various] There are several typographical errors: 'exgtended' in the Figure 1 caption, 'T able 1' in the Table 1 caption, 'simuation' in the Figure 13 caption, and 'Fcall' in Section 3.4. These should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the CRInj/CRBkg comparison is a controlled numerical experiment, transport coefficients and benchmarks are external, and the self-citations are non-load-bearing.

full rationale

The paper's derivation chain is a controlled numerical experiment: two otherwise identical MHD+CR simulations differ only in whether 10% of each supernova's energy is injected as cosmic-ray energy (CRInj) or all as thermal energy (CRBkg). The headline result—faster, hotter, and more massive outflow with cosmic-ray pressure gradients dominating at |z| ≥ 1 kpc—is measured from simulation outputs (Figures 10-13), not recovered from a fitted parameter or from a definition. The transport coefficients (κ∥ = 3×10^28 cm^2/s, negligible κ⊥, Alfvén-speed streaming) are taken from external literature (Jones et al. 2001; Jiang & Oh 2018) and are not tuned to force the target result. The Parker-instability check uses the analytic criterion and dispersion relations of Heintz & Zweibel (2018); although Zweibel is a co-author, this is an independently published linear-theory result, and the paper explicitly compares its measured growth rate (18-25 Myr) against the predicted value (~33 Myr) rather than imposing it. The F_cal comparison uses the external observational benchmark of Lacki et al. (2011), with the net calorimetric fraction obtained by background subtraction, not by fitting. Self-citations (Habegger et al. 2023 for setup provenance and V_m convergence; Habegger et al. 2024 for diffusion-coefficient discussion) are supporting references and are not load-bearing for the central comparison. The paper's own caveats (diode boundary, constant heating rate, incomplete star-formation feedback loop, numerical reconnection) concern external validity and robustness, not circularity; no equation in the paper reduces by construction to its own inputs, and no fitted parameter is renamed as a prediction.

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

The central claims rest on standard MHD plus CR transport assumptions, a Parker-unstable initial condition, and a simplified boundary and gravity setup. No new physical entities are introduced. The free parameters are choices of energy partitioning, temperature floor, supernova rate, and numerical speed of light, none of which are fitted to the simulation's output.

free parameters (4)
  • f_CR = 0.1
    Fraction of supernova energy injected as cosmic rays, set to match the observed cosmic-ray energy density in the Milky Way ISM (Section 2). Central to the CRInj versus CRBkg comparison.
  • T_floor = 30 K
    Temperature floor imposed to avoid numerical artifacts and molecular gas formation; it caps the cold gas phase and thus affects the reported cold-gas mass fraction (Section 2.2).
  • supernova_rate = 1 Myr^-1
    Constant rate of supernova injection, chosen to correspond to a low star formation rate of 10^-4 Msun yr^-1 kpc^-2; the mass loading factor depends inversely on this choice (Section 2.3).
  • V_m = 0.1c
    Modified speed of light that limits cosmic-ray transport speed; a numerical parameter with a convergence study cited from prior work, but it sets the MHD timestep and can affect transport on short timescales (Section 2).
assumptions (6)
  • domain assumption Cosmic-ray transport is modeled with anisotropic diffusion along the local magnetic field at kappa_parallel = 3e28 cm^2/s plus streaming at the Alfven speed (Eqs. 5 through 11).
    Assumed closure for CR transport from Jiang and Oh 2018; the result that CRs escape via vertical field lines depends on this transport model.
  • domain assumption The initial atmosphere is linearly unstable to the Parker instability, using the criterion of Newcomb 1961 and Heintz and Zweibel 2018 (Eq. 21).
    The central role of the Parker instability is a consequence of choosing a setup that satisfies this instability criterion with beta = 1 and beta_cr = 1.
  • domain assumption The z boundaries are one-way 'diode' boundaries that allow outflow but no inflow (Section 2.1).
    This prevents infall from the halo and may inflate the sustained outflow, directly affecting the claim of a faster, more massive CR-driven outflow.
  • domain assumption Gravity is from an infinite isothermal stellar sheet with Sigma_star = 50 Msun/pc^2 and H_star = 100 pc (Eq. 12), neglecting dark matter and finite disk effects.
    Simplified gravity profile is used for the initial hydrostatic equilibrium; the authors note it is unrealistic at large |z| and would reduce the outflow rate if corrected (Section 4.3).
  • domain assumption Supernovae are injected at random midplane positions at a constant rate, not tied to gas density or star formation, and the molecular weight is constant mu = 1 (Sections 2.2 and 2.3).
    This decouples feedback from the simulated gas state; the paper acknowledges the feedback loop is not closed and that dense-gas-tied injection would change the midplane CR transport.
  • domain assumption Streaming uses the full gas density instead of ion density in the Alfven speed (Eq. 9).
    The authors state this simplification only impacts cold gas, but it alters the streaming speed and heating in the cold, partially ionized clouds that play a role in the decorrelation and Fcal results.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Cosmic-Ray Feedback from Supernovae in a Parker-Unstable Medium." pith.science (2026). https://pith.science/paper/T3KVXI5Y

@misc{pith2026241212249,
  author       = {Pith},
  title        = {Pith review of: Cosmic-Ray Feedback from Supernovae in a Parker-Unstable Medium},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T3KVXI5Y}},
  note         = {Machine review of arXiv:2412.12249}
}
abstract

Supernova energy drives interstellar medium (ISM) turbulence and can help launch galactic winds. What difference does it make if $10\%$ of the energy is initially deposited into cosmic rays? To answer this question and study cosmic-ray feedback, we perform galactic patch simulations of a stratified ISM. We compare two magnetohydrodynamic and cosmic ray (MHD+CR) simulations, which are identical except for how each supernova's energy is injected. In one, $10\%$ of the energy is injected as cosmic-ray energy. In the other case, energy injection is strictly thermal and kinetic. We find that cosmic-ray injections drive a faster, hotter, and more massive outflow long after the injections occur. Both simulations show the formation of cold clouds (with a total mass fraction $>50\%$) through the Parker instability and thermal instability. The Parker instability simultaneously produces high mass loading factors $\eta > 10^3$ as it requires few supernovae. We also show how the Parker instability naturally leads to a decorrelation of cosmic-ray pressure and gas density. This decorrelation leads to a significant decrease in the calorimetric fraction for injected cosmic rays, but it depends on having a highly resolved magnetic field.

Figures

Figures reproduced from arXiv: 2412.12249 by the authors.

Figure 1
Figure 1. 3D snapshots of each simulation at t = 200 Myr (see [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Top: Our combined radiative heating and cool￾ing rate L in Equation 17 gives equilibrium states shown as a black line in this gas pressure-gas density phase diagram. We plot our temperature floor as a dashed line in the bot￾tom right of the plot. The contours show the net heating rate in red contours and net cooling rates in blue contours, with darker colors being faster rates. Bottom: We use the radiative cooling r… view at source ↗
Figure 3
Figure 3. The time evolution of simulations CRInj (left column) and CRBkg (right column), with each quantity averaged over the entire simulation volume. The top row shows the average velocity divided into its directional components (ˆx, y, ˆ zˆ), and the middle row shows the same for magnetic field strength. Compared to the simulation with only an initial background of cosmic rays, additional cosmic ray injections amplify the… view at source ↗
Figures from the paper (16 more)
Figure 4
Figure 4. Figure 4: Top: Mass outflow and inflow rates for regions above and below the midplane (|z| > 1 kpc). Solid lines show the CRInj simulation and dashed lines show the CRBkg simulation. The outflows and inflows are out of phase, sug￾gesting the simulations produce a fountain flow. …
Figure 5
Figure 5. Figure 5: Left: Kinetic energy spectrum ˜v(k) (in units of km s−1 ) for the CRInj simulation at each 1 Myr time dump. The color map shows the amplitude of velocity fluctuations at each wavelength L = 2π/k and the green line traces the peak wavelength of the spectrum. The spectru…
Figure 6
Figure 6. Figure 6: Left: Vertical velocity colormap with gas density contours and magnetic field lines overlaid. The large plume centered at x ∼ 0.1kpc is moving upward at a fast rate (vz ∼ 50 − 100 km s−1 ). Right: Colormap of cosmic-ray pressure with gas density contours and magnetic f…
Figure 7
Figure 7. Figure 7: A zoom-in on dense gas from [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Top: Growth of vertical kinetic and magnetic energy density in the CRInj simulation is exponential with a characteristic time of τ ∼ 18 Myr. Middle: Column density colormap and average cosmic-ray pressure contours at t = 90 Myr. Bottom: Same as middle plot, but at t = …
Figure 9
Figure 9. Figure 9: The average vertical density profile over the pe￾riod t = [260, 400] Myr. The solid green line is from the CRInj simulation and dashed purple line is from the CRBkg simu￾lation. We calculated the profile at every 1 Myr time dump by averaging over each xy-plane and then…
Figure 10
Figure 10. Figure 10: xy-plane averaged outflow/inflow velocities from both simulations. The averaged velocity profile over the steady state time frame (t ∈ [260, 400]Myr) is shown for each simulation. The solid green line is from the CRInj simula￾tion and dashed purple line is from the CR…
Figure 11
Figure 11. Figure 11: Spacetime diagrams of the dominant acceleration components in the CRInj simulation (top figure) and the CRBkg simulation (bottom figure). The lower quarter of the colorbar (purple) is for regions where the xy-plane averaged vertical acceleration is predominantly drive…
Figure 12
Figure 12. Figure 12: Vertical acceleration by various forces in the CRInj simulation (left plot) and in the CRBkg simulation (right plot). The solid black line shows the gravitational acceleration from the background profile, the solid orange line shows the acceleration from the vertical …
Figure 14
Figure 14. Figure 14: xy-plane averaged fraction of the magnetic field directed vertically. The lines show the average verti￾cal magnetic field fraction for the steady state time period (t ∈ [260, 400]Myr). The solid green line is from the CRInj simulation and dashed purple line is from th…
Figure 13
Figure 13. Figure 13: Bottom: Net acceleration in each simulation, calculated by adding up the forces in [PITH_FULL_IMAGE:figures/full_fig_p017_13.png]
Figure 15
Figure 15. Figure 15: Average phase diagrams for the CRInj simulation (left column) and the CRBkg simulation over the steady state time frame (t > 260 Myr). Top Row: cosmic-ray pressure-gas number density phase diagram, which show predominatly diffusive transport in the high density midpla…
Figure 16
Figure 16. Figure 16: shows the average vertical profiles and our median fit to those profiles during the saturated, steady state. The averaged data are plotted within a shaded region, which shows the total variation over the satu￾rated time frame. We removed the cold and thermally unstabl…
Figure 17
Figure 17. Figure 17: Gas scale heights in CRInj simulation and CRBkg simulation. Black lines show the inner warm gas density scale heights, and red lines show the outer warm gas density scale heights. The green lines show the scale height for the magnetic energy density, and blue lines sh…
Figure 18
Figure 18. Figure 18: Bottom: Fraction of cosmic-ray energy injected which would be lost to hadronic interactions, following the definition of Fcal in Lacki et al. (2011). We plot the results for both simulations, with solid lines showing the CRInj sim￾ulation and dashed lines showing the …
Figure 19
Figure 19. Figure 19: Top plot: Net Fcal as calculated in [PITH_FULL_IMAGE:figures/full_fig_p022_19.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 9 canonical work pages

  1. [1]

    C., & Jiang, Y.-F

    Armillotta, L., Ostriker, E. C., & Jiang, Y.-F. 2021, ApJ, 922, 11, doi: 10.3847/1538-4357/ac1db2

    doi:10.3847/1538-4357/ac1db2 2021

  2. [2]

    C., Kim, C.-G., & Jiang, Y.-F

    Armillotta, L., Ostriker, E. C., Kim, C.-G., & Jiang, Y.-F. 2024, ApJ, 964, 99, doi: 10.3847/1538-4357/ad1e5c Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. 2013, A&A, 558, A33, doi: 10.1051/0004-6361/201322068 Astropy Collaboration, Price-Whelan, A. M., Sip˝ ocz, B. M., et al. 2018, AJ, 156, 123, doi: 10.3847/1538-3881/aabc4f

    doi:10.3847/1538-4357/ad1e5c 2024

  3. [3]

    1934, Physical Review, 46, 76, doi: 10.1103/PhysRev.46.76.2

    Baade, W., & Zwicky, F. 1934, Physical Review, 46, 76, doi: 10.1103/PhysRev.46.76.2

    doi:10.1103/physrev.46.76.2 1934

  4. [4]

    D., & Ostriker, J

    Blandford, R. D., & Ostriker, J. P. 1978, ApJL, 221, L29, doi: 10.1086/182658

    doi:10.1086/182658 1978

  5. [5]

    F., & Voelk, H

    Breitschwerdt, D., McKenzie, J. F., & Voelk, H. J. 1991, A&A, 245, 79

    work page 1991

  6. [6]

    Bustard, C., & Zweibel, E. G. 2021, ApJ, 913, 106, doi: 10.3847/1538-4357/abf64c

    doi:10.3847/1538-4357/abf64c 2021

  7. [7]

    2014, ApJ, 783, 91, doi: 10.1088/0004-637X/783/2/91

    Caprioli, D., & Spitkovsky, A. 2014, ApJ, 783, 91, doi: 10.1088/0004-637X/783/2/91

    doi:10.1088/0004-637x/783/2/91 2014

  8. [8]

    K., Kereˇ s, D., Gurvich, A

    Chan, T. K., Kereˇ s, D., Gurvich, A. B., et al. 2022, MNRAS, 517, 597, doi: 10.1093/mnras/stac2236

    doi:10.1093/mnras/stac2236 2022

Show all 55 references
  1. [9]

    K., Kereˇ s, D., Hopkins, P

    Chan, T. K., Kereˇ s, D., Hopkins, P. F., et al. 2019, MNRAS, 488, 3716, doi: 10.1093/mnras/stz1895

    2019 doi

  2. [10]

    Desiati, P., & Zweibel, E. G. 2014, ApJ, 791, 51, doi: 10.1088/0004-637X/791/1/51

    2014 doi

  3. [11]

    2018, PhRvL, 121, 091101, doi: 10.1103/PhysRevLett.121.091101 26

    Diesing, R., & Caprioli, D. 2018, PhRvL, 121, 091101, doi: 10.1103/PhysRevLett.121.091101 26

    2018 doi

  4. [12]

    Draine, B. T. 2011, Physics of the Interstellar and Intergalactic Medium

    2011

  5. [13]

    2022, ApJ, 941, 162, doi: 10.3847/1538-4357/aca27d

    Elia, D., Molinari, S., Schisano, E., et al. 2022, ApJ, 941, 162, doi: 10.3847/1538-4357/aca27d

    2022 doi

  6. [14]

    2019, PhRvD, 99, 103023, doi: 10.1103/PhysRevD.99.103023

    Evoli, C., Aloisio, R., & Blasi, P. 2019, PhRvD, 99, 103023, doi: 10.1103/PhysRevD.99.103023

    2019 doi

  7. [15]

    2020, PhRvD, 101, 023013, doi: 10.1103/PhysRevD.101.023013

    Evoli, C., Morlino, G., Blasi, P., & Aloisio, R. 2020, PhRvD, 101, 023013, doi: 10.1103/PhysRevD.101.023013

    2020 doi

  8. [16]

    Farber, R., Ruszkowski, M., Yang, H. Y. K., & Zweibel, E. G. 2018, ApJ, 856, 112, doi: 10.3847/1538-4357/aab26d Ferri` ere, K. M. 2001, Reviews of Modern Physics, 73, 1031, doi: 10.1103/RevModPhys.73.1031

    2018 doi

  9. [17]

    T., & Shu, F

    Giz, A. T., & Shu, F. H. 1993, ApJ, 404, 185, doi: 10.1086/172267

    1993 doi

  10. [18]

    Guo, F., & Oh, S. P. 2008, MNRAS, 384, 251, doi: 10.1111/j.1365-2966.2007.12692.x

    2008

  11. [19]

    W., Yuen, K

    Habegger, R., Ho, K. W., Yuen, K. H., & Zweibel, E. G. 2024, ApJ, 974, 17, doi: 10.3847/1538-4357/ad67da

    2024 doi

  12. [20]

    G., & Wong, S

    Habegger, R., Zweibel, E. G., & Wong, S. 2023, ApJ, 951, 99, doi: 10.3847/1538-4357/accf8e

    2023 doi

  13. [21]

    R., Millman, K

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

    2020 doi

  14. [22]

    Heintz, E., Bustard, C., & Zweibel, E. G. 2020, ApJ, 891, 157, doi: 10.3847/1538-4357/ab7453

    2020 doi

  15. [23]

    Heintz, E., & Zweibel, E. G. 2018, ApJ, 860, 97, doi: 10.3847/1538-4357/aac208

    2018 doi

  16. [24]

    W., Yuen, K

    Ho, K. W., Yuen, K. H., & Lazarian, A. 2024, arXiv e-prints, arXiv:2407.14199, doi: 10.48550/arXiv.2407.14199

    2024 doi arXiv

  17. [25]

    N., Wasserman, D., & Cowley, S

    Hosking, D. N., Wasserman, D., & Cowley, S. C. 2024, arXiv e-prints, arXiv:2401.01336, doi: 10.48550/arXiv.2401.01336

    2024 doi arXiv

  18. [26]

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

    2007 doi

  19. [27]

    2006, ApJ, 652, 1331, doi: 10.1086/508334

    Inoue, T., Inutsuka, S.-i., & Koyama, H. 2006, ApJ, 652, 1331, doi: 10.1086/508334

    2006 doi

  20. [28]

    Jiang, Y.-F., & Oh, S. P. 2018, ApJ, 854, 5, doi: 10.3847/1538-4357/aaa6ce

    2018 doi

  21. [29]

    C., Lukasiak, A., Ptuskin, V., & Webber, W

    Jones, F. C., Lukasiak, A., Ptuskin, V., & Webber, W. 2001, ApJ, 547, 264, doi: 10.1086/318358

    2001 doi

  22. [30]

    Kim, C.-G., Kim, J.-G., Gong, M., & Ostriker, E. C. 2023, ApJ, 946, 3, doi: 10.3847/1538-4357/acbd3a

    2023 doi

  23. [31]

    Kim, C.-G., & Ostriker, E. C. 2015, ApJ, 802, 99, doi: 10.1088/0004-637X/802/2/99

    2015 doi

  24. [32]

    2011, ApJ, 734, 107, doi: 10.1088/0004-637X/734/2/107

    Waxman, E. 2011, ApJ, 734, 107, doi: 10.1088/0004-637X/734/2/107

    2011 doi

  25. [33]

    McLeod, A. F. 2025, arXiv e-prints, arXiv:2506.17689, doi: 10.48550/arXiv.2506.17689

    2025 doi arXiv

  26. [34]

    M., Stanimirovi´ c, S., & Rybarczyk, D

    McClure-Griffiths, N. M., Stanimirovi´ c, S., & Rybarczyk, D. R. 2023, ARA&A, 61, 19, doi: 10.1146/annurev-astro-052920-104851

    2023 doi

  27. [35]

    F., & Voelk, H

    McKenzie, J. F., & Voelk, H. J. 1982, A&A, 116, 191

    1982

  28. [36]

    Newcomb, W. A. 1961, Physics of Fluids, 4, 391, doi: 10.1063/1.1706342 O’Neill, T. J., Zucker, C., Goodman, A. A., & Edenhofer, G. 2024, ApJ, 973, 136, doi: 10.3847/1538-4357/ad61de

    1961 doi

  29. [37]

    Mitchell, A. M. W. 2023, Galaxies, 11, 86, doi: 10.3390/galaxies11040086

    2023 doi

  30. [38]

    Parker, E. N. 1966, ApJ, 145, 811, doi: 10.1086/148828

    1966 doi

  31. [39]

    2021, MNRAS, 504, 1039, doi: 10.1093/mnras/stab900

    Rathjen, T.-E., Naab, T., Girichidis, P., et al. 2021, MNRAS, 504, 1039, doi: 10.1093/mnras/stab900

    2021 doi

  32. [40]

    Rosen, A., & Bregman, J. N. 1995, ApJ, 440, 634, doi: 10.1086/175303

    1995 doi

  33. [41]

    2023, A&A Rv, 31, 4, doi: 10.1007/s00159-023-00149-2

    Ruszkowski, M., & Pfrommer, C. 2023, A&A Rv, 31, 4, doi: 10.1007/s00159-023-00149-2

    2023 doi

  34. [42]

    R., Wenger, T

    Rybarczyk, D. R., Wenger, T. V., & Stanimirovi´ c, S. 2024, ApJ, 975, 167, doi: 10.3847/1538-4357/ad79f7

    2024 doi

  35. [43]

    1959, ApJ, 129, 243, doi: 10.1086/146614

    Schmidt, M. 1959, ApJ, 129, 243, doi: 10.1086/146614

    1959 doi

  36. [44]

    A., Kravtsov, A

    Semenov, V. A., Kravtsov, A. V., & Caprioli, D. 2021, ApJ, 910, 126, doi: 10.3847/1538-4357/abe2a6

    2021 doi

  37. [45]

    2022, ApJ, 926, 8, doi: 10.3847/1538-4357/ac4110

    Shimoda, J., & Inutsuka, S.-i. 2022, ApJ, 926, 8, doi: 10.3847/1538-4357/ac4110

    2022 doi

  38. [46]

    M., Tomida, K., White, C

    Stone, J. M., Tomida, K., White, C. J., & Felker, K. G. 2020, The Astrophysical Journal Supplement Series, 249, 4, doi: 10.3847/1538-4365/ab929b

    2020 doi

  39. [47]

    A., et al

    Tharakkal, D., Shukurov, A., Gent, F. A., et al. 2023, MNRAS, 525, 5597, doi: 10.1093/mnras/stad2610

    2023 doi

  40. [48]

    2024, arXiv e-prints, arXiv:2405.13121, doi: 10.48550/arXiv.2405.13121

    Thomas, T., Pfrommer, C., & Pakmor, R. 2024, arXiv e-prints, arXiv:2405.13121, doi: 10.48550/arXiv.2405.13121

    2024 doi arXiv

  41. [49]

    Townsend, R. H. D. 2009, ApJS, 181, 391, doi: 10.1088/0067-0049/181/2/391

    2009 doi

  42. [50]

    Tsung, T. H. N., Oh, S. P., & Bustard, C. 2023, MNRAS, 526, 3301, doi: 10.1093/mnras/stad2720

    2023 doi

  43. [51]

    Tsung, T. H. N., Oh, S. P., & Jiang, Y.-F. 2022, MNRAS, 513, 4464, doi: 10.1093/mnras/stac1123 van der Walt, S., Colbert, S. C., & Varoquaux, G. 2011, Computing in Science and Engineering, 13, 22, doi: 10.1109/MCSE.2011.37

    2022 doi

  44. [52]

    2015, MNRAS, 454, 238, doi: 10.1093/mnras/stv1975

    Walch, S., Girichidis, P., Naab, T., et al. 2015, MNRAS, 454, 238, doi: 10.1093/mnras/stv1975

    2015 doi

  45. [53]

    P., & Zweibel, E

    Wiener, J., Oh, S. P., & Zweibel, E. G. 2017, MNRAS, 467, 646, doi: 10.1093/mnras/stx109 27

    2017 doi

  46. [54]

    G., & Ruszkowski, M

    Wiener, J., Zweibel, E. G., & Ruszkowski, M. 2019, MNRAS, 489, 205, doi: 10.1093/mnras/stz2007 Wolfram Research Inc. 2023, Mathematica, Version 13.3

    2019 doi

  47. [55]

    Zweibel, E. G. 2017, Physics of Plasmas, 24, 055402, doi: 10.1063/1.4984017

    2017 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.