Kglobal Model for Macroscale Reconnection

• Kglobal model is a computational framework that blends an MHD-like fluid backbone with guiding-center macro-particles to study large-scale magnetic reconnection and non-thermal particle energization.
• It eliminates kinetic scales to focus on Fermi reflection and parallel electric fields, making it ideal for modeling reconnection in environments such as solar flares and magnetotail current sheets.
• The model employs energy-conserving closures and pressure-anisotropy feedback to capture phenomena like flux-rope mergers, shock formation, and simultaneous ion–electron energization.

Searching arXiv for papers on the kglobal model in magnetic reconnection. The kglobal model is a computational model for macroscale magnetic reconnection that combines an MHD-like fluid backbone with guiding-center macro-particles to study non-thermal particle energization without resolving kinetic scales such as the ion and electron gyro-radii, inertial lengths, or Debye length. It was introduced to address regimes in which fully kinetic PIC simulations are computationally prohibitive, while pure MHD lacks self-consistent particle acceleration. Across its published development, kglobal has been used to study energetic electron production, large-scale parallel electric fields and return currents, simultaneous ion and electron energization, upstream slow shocks, and system-size control of maximum particle energy during reconnection (Arnold et al., 2019, Yin et al., 2024, Arnold et al., 2021, Yin et al., 15 Dec 2025).

1. Definition and physical scope

kglobal is described as a macroscale reconnection model for energetic electron production, and later as a framework extended to include particle ions as well as electrons. Its basic premise is that in large reconnection systems the dominant energization channels are controlled by large-scale magnetic geometry, especially Fermi reflection in large-scale magnetic fields, rather than by kinetic-scale boundary layers localized near x-lines and separatrices (Arnold et al., 2019, Yin et al., 2024).

The model therefore eliminates kinetic scales and evolves the plasma self-consistently on fluid scales while advancing energetic particles as guiding-center macro-particles. In this formulation, particles move across the magnetic field via $\mathbf{E}\times\mathbf{B}$ drift and along the field at their parallel velocity, while their gyromotion is not resolved. This design makes the model suitable for systems such as solar flares, large magnetotail reconnection regions, and heliospheric current sheets, where the scale separation between the global energy-release region and kinetic plasma scales is extreme (Arnold et al., 2019, Yin et al., 31 Jul 2025, Desai et al., 2024).

A central consequence of this ordering is that kglobal is not intended as a full kinetic first-principles description of microscopic processes. Instead, it is intended to retain the reconnection dynamics most relevant to large-scale particle acceleration, including flux-rope formation, island mergers, pressure anisotropy feedback, and parallel electric-field effects that remain important after kinetic layers are removed (Arnold et al., 2019, Yin et al., 2024).

2. Core formulation and dynamical variables

The original kglobal model contains three plasma components: ion fluid, electron fluid, and particle electrons. In the later ion-enabled formulation, a fourth component, particle ions, is added. No species conversion is allowed after initialization (Yin et al., 2024).

In the electron-only formulation, particle electrons are advanced in the guiding-center limit with conserved magnetic moment

$$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$

and parallel momentum evolution

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$

The terms represent curvature/Fermi acceleration, mirror force, and parallel electric-field acceleration. The same guiding-center structure is later used for particle ions, with

$\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$

and

$$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$

This establishes a unified macroscale representation of both ion and electron energization (Yin et al., 2024).

The ion fluid obeys a continuity equation and an MHD-like momentum equation coupled to particle stresses and the electromagnetic field. In the original model the momentum equation includes ion-fluid stress, electron-fluid stress, particle-electron stress, the Lorentz force, and the field-aligned electric force. The magnetic field evolves through Faraday’s law using the perpendicular electric field,

$$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$

The parallel electric field is determined from electron pressure and stress terms rather than from kinetic boundary-layer dynamics (Yin et al., 2024).

The model uses quasi-neutrality and a zero-parallel-current closure. In the original formulation,

$n_{ef}=n_i-n_{ep},$

and

$$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$

with the ordering

$$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$

In the ion-extended version, these relations become

$n_{ef}=n_{if}+n_{ip}-n_{ep},$

and

$$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$0

These closures encode the macroscale assumption that current-drift scales have been ordered out, while field-aligned flows remain self-consistent (Yin et al., 2024).

3. Large-scale parallel electric field and energy-conserving closure

A major development of kglobal was the explicit inclusion of a large-scale parallel electric field produced by magnetic-field-aligned gradients in electron pressure. This extension distinguishes between two different classes of field-aligned electric field: localized kinetic-scale parallel electric fields near the x-line and separatrices, and a macroscale $$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$1 that acts over the reconnection exhaust and is dynamically important for transport, early-stage heating, and return-current closure (Arnold et al., 2019).

In the upgraded formulation, the large-scale parallel electric field is

$$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$2

where $$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$3, $$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$4 and $$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$5 are cold and hot electron densities, $$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$6 is the cold-electron parallel flow, $$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$7 is the cold-electron scalar pressure, $$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$8, and $$\mu_{ep} = \frac{p_{ep,\perp}^2}{2B} = \text{const.}$$9 is the hot-electron gyrotropic stress tensor. The latter is written as

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$0

with

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$1

The normalization of this field is

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$2

and

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$3

Accordingly, $$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$4 is retained for motion along the field but neglected in Faraday’s law for the evolution of the magnetic field (Arnold et al., 2019).

This extension also introduces cold-electron return currents in open systems. The cold-electron parallel speed is constrained by

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$5

so that escaping hot-electron current is balanced by a compensating cold-electron flow. The model therefore represents large-scale current closure and suppression of hot-electron escape without reintroducing kinetic boundary layers (Arnold et al., 2019).

A defining feature of this formulation is that the added $$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$6 is derived in an energy-conserving manner. The resulting conservation law is

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$7

with cold-electron energy

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$8

A later ion-inclusive formulation extends this conservation structure to

$$\frac{d p_{ep,\parallel}}{dt} = p_{ep,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\frac{\mu_e}{\gamma_e}\mathbf{b}\cdot\nabla B - eE_{\parallel}.$$9

up to small errors associated with formally neglected terms. This places energy conservation at the center of the model’s reduced-physics design (Arnold et al., 2019, Yin et al., 2024).

4. Extension from electron-only to ion-electron energization

The paper “A Computational Model for Ion and Electron Energization during Macroscale Magnetic Reconnection” extends kglobal from an electron-only kinetic-fluid hybrid to a model with both ions and electrons as particle species (Yin et al., 2024). The principal structural change is the inclusion of particle-ion inertia in the fluid momentum equation.

In the original model, electron inertia could be neglected in the ion-fluid momentum balance. This is no longer possible once particle ions are added. The upgraded fluid-ion momentum equation becomes

$\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$0

with the correction term

$\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$1

This term accounts for the difference in parallel inertia between fluid and particle ions (Yin et al., 2024).

The particle-ion stress tensor is gyrotropic,

$\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$2

with

$\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$3

The electron-fluid and ion-fluid pressures use adiabatic closures,

$\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$4

with $\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$5 (Yin et al., 2024).

This extension broadens the model’s scope from energetic electron production to the simultaneous non-thermal energization of ions and electrons. A plausible implication is that kglobal becomes suitable not only for electron-acceleration problems, but also for energy partition studies in large reconnection systems where proton and electron heating evolve differently.

5. Dominant acceleration channels and species dependence

Across the kglobal literature, the dominant acceleration mechanism is repeatedly identified as Fermi reflection associated with contracting magnetic structures, reconnection exhausts, and especially flux-rope or magnetic-island mergers. In the model’s physical picture, particles reflect from moving magnetic structures whose motion is tied to Alfvénic reconnection outflows, so the energization rate can be proportional to particle energy (Arnold et al., 2019, Yin et al., 15 Dec 2025, Yin et al., 31 Jul 2025).

In the study of maximum particle energy, one merger of two equal flux ropes of radius $\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$6 is described by the field-line shortening

$\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$7

which yields a $\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$8 increase in $\mu_{ip}=\frac{p_{ip,\perp}^2}{2B}=\text{const.}$9 and therefore approximately

$$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$0

After $$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$1 comparable mergers,

$$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$2

The number of mergers is linked to the effective system size through hyper-resistive control of the smallest island width,

$$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$3

with

$$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$4

so that when the largest island reaches system scale,

$$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$5

and therefore

$$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$6

The simulations vary $$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$7 from $$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$8 to $$\frac{d p_{ip,\parallel}}{dt} = p_{ip,\parallel}\mathbf{v}_E\cdot \boldsymbol{\kappa} -\mu_i\,\mathbf{b}\cdot\nabla B +eE_{\parallel}.$$9 and find that the inferred merger count rises from about $$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$0 to $$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$1, while the high-energy cutoff shifts to higher energies as $$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$2 increases (Yin et al., 15 Dec 2025).

The model has also been used to explain why proton heating and energization exceed those of electrons when the upstream temperatures of the two species are equal. In that study, particles entering a reconnection exhaust gain a velocity increment of order

$$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$3

so that for a single reflection

$$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$4

For protons, the initial gain scales like

$$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$5

For electrons, the initial energy gain is smaller and scales as

$$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$6

The simulations report late-time temperature increments nearly insensitive to mass ratio,

$$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$7

for $$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$8, respectively. This is interpreted as evidence that the relative heating is controlled more by the early injection energy and the large-scale Fermi process than by the artificial mass ratio (Yin et al., 31 Jul 2025).

Guide-field strength modifies the efficiency of this process. A strong guide field increases the curvature radius of reconnected field lines and suppresses Fermi reflection, reducing energy gain. In the maximum-energy study the guide field is held fixed so that the principal scaling remains the system-size dependence $$\frac{\partial \mathbf{B}}{\partial t} = -c\,\boldsymbol{\nabla\times}\mathbf{E}_{\perp}, \qquad \mathbf{E}_{\perp}=-\frac{1}{c}\mathbf{v}_i\times\mathbf{B}.$$9 (Yin et al., 15 Dec 2025).

6. Reconnection structures, shocks, and benchmarks

kglobal has been used not only for non-thermal tails, but also for shock formation and anisotropic MHD-like wave physics. In “Slow Shock Formation Upstream of Reconnecting Current Sheets,” slow shocks are documented in the upstream region of simulations with the kglobal kinetic macroscale simulation model. During multi-island reconnection, the formation and merging of flux ropes drives plasma flows and pressure disturbances in the upstream region. These disturbances steepen into slow shocks that propagate along the reconnecting component of the magnetic field and satisfy the expected Rankine-Hugoniot jump conditions. Plasma heating arises from both compression across the shock and the parallel electric field that develops to maintain charge neutrality in a kinetic system. The shocks are weaker at lower plasma $n_{ef}=n_i-n_{ep},$0, where shock steepening is slow, and their contribution to electron heating is described as relatively minor compared with Fermi reflection and the parallel electric fields that bound the reconnection outflow (Arnold et al., 2021).

The model has also been benchmarked against wave phenomena that probe its anisotropic closures. In the large-scale $n_{ef}=n_i-n_{ep},$1 paper, kglobal accurately captures the damping of electron acoustic modes in a plasma with hot kinetic electrons and cold fluid electrons. For the perturbed distribution,

$n_{ef}=n_i-n_{ep},$2

the perturbed parallel electric field is

$n_{ef}=n_i-n_{ep},$3

and the dispersion relation is

$n_{ef}=n_i-n_{ep},$4

The reported damping rates agree very well with the linear theory. The same paper benchmarks the suppression of hot-electron transport by comparing kglobal with the PIC code p3d, finding that kglobal with $n_{ef}=n_i-n_{ep},$5 matches p3d well over most of the domain, whereas kglobal without $n_{ef}=n_i-n_{ep},$6 spreads much more rapidly (Arnold et al., 2019).

The ion-inclusive formulation is benchmarked using a circularly polarized Alfvén wave in anisotropic plasma and the firehose instability. The predicted phase speed is

$n_{ef}=n_i-n_{ep},$7

and the firehose growth rate is

$n_{ef}=n_i-n_{ep},$8

with $n_{ef}=n_i-n_{ep},$9 and a test case using $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$0. The measured phase speeds and growth rates are reported to agree well with theory, and no spurious short-wavelength instabilities appear (Yin et al., 2024).

7. Applications, observational interpretation, and limitations

The kglobal model has been used as an interpretive framework for direct spacecraft observations. In a Parker Solar Probe study of the near-Sun heliospheric current sheet, kglobal is described as a global reconnection modeling framework upgraded to handle ion energization as well as electron energization, with all kinetic scales eliminated. Simulations of a Harris-like equilibrium use upstream parameters based directly on observations: $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$1, $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$2, $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$3, $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$4, proton temperature $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$5, electron temperature $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$6, and guide fields $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$7 and $$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$8 (Desai et al., 2024).

The hybrid Alfvén speed is

$$n_{ef}v_{ef,\parallel}=n_iv_{i,\parallel}-n_{ep}v_{ep,\parallel},$$9

with

$$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$0

Despite this modest magnetic energy per particle, the simulations produce protons up to $$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$1 for $$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$2 and electrons up to $$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$3, while the observed proton spectrum between about $$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$4 and $$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$5 is fitted by

$$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$6

The $$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$7 simulation gives a softer spectrum with power-law index $$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$8 and maximum proton energy $$\frac{V_d}{C_A}=\frac{J}{neC_A}\sim \frac{d_i}{L}\ll 1.$$9, and is identified as the best match to the observed slope. The study argues that a guide field near $n_{ef}=n_{if}+n_{ip}-n_{ep},$0 would likely match both slope and cutoff more closely (Desai et al., 2024).

These applications reinforce a recurring interpretive theme: kglobal attributes extended power-law tails and large maximum energies to repeated Fermi acceleration in merging magnetic islands or flux ropes, while allowing pressure anisotropy and firehose feedback to regulate the reconnection dynamics. This suggests that the model is especially effective when the dominant physics is macroscale island-driven acceleration rather than kinetic boundary-layer processes or strong turbulence.

The model’s limitations are stated explicitly in the literature. kglobal is 2D in several published applications, removes kinetic scales, uses guiding-center dynamics, employs hyper-resistivity and, in some studies, an artificial proton-to-electron mass ratio of 25. The upper energy cutoff can also be controlled by the size of the simulation domain, so simulated maxima are not always pure physics cutoffs. The model does not support plasma waves that require violation of charge neutrality, and it is not designed to capture fine kinetic boundary layers or short-wavelength plasma oscillations (Arnold et al., 2019, Desai et al., 2024).

A common misconception is that kglobal is simply a cheaper PIC surrogate. The published descriptions do not support that characterization. Instead, kglobal is a reduced-kinetic macroscale model built around a specific claim: that the high-energy particle population in large reconnection systems is controlled mainly by large-scale Fermi processes, parallel electric fields from pressure gradients, and pressure-anisotropy feedback, rather than by directly resolving kinetic boundary layers (Arnold et al., 2019, Yin et al., 2024).