MInIT: MHD Instability-Induced Turbulence

• MHD-Instability-Induced Turbulence (MInIT) is a framework describing how magnetic instabilities generate and regulate turbulence by destabilizing current sheets and coherent structures.
• It encompasses various mechanisms—including tearing, plasmoid, MRI, and KHI—that modify spectral transfer and energy dissipation in both astrophysical and laboratory plasmas.
• The concept also informs sub-grid mean-field models that reconstruct unresolved Maxwell, Reynolds, and Faraday stresses to improve simulation fidelity.

MHD-Instability-Induced Turbulence (MInIT) denotes a class of magnetohydrodynamical processes in which an MHD instability either generates turbulence directly or destabilizes structures produced by an existing cascade, so that the instability feeds back on transport, dissipation, spectral transfer, or coherent-field generation. In the cited literature, the term is used in two closely connected senses: as a physical picture for turbulence self-organization by tearing, plasmoid, MRI, KHI, or statistical-state instabilities, and as a mean-field sub-grid model that evolves unresolved turbulent energies and reconstructs Maxwell, Reynolds, and Faraday stresses (Walker et al., 2018, Miravet-Tenés et al., 2022, Miravet-Tenés et al., 2023).

1. Conceptual basis and scope

The central MInIT idea is recursive: turbulence or large-scale MHD motion produces thin current sheets, shear layers, or coherent mean structures; those structures then become linearly or statistically unstable; the resulting instability modifies the subsequent turbulent evolution rather than acting only as a terminal dissipation mechanism. In the tearing-mediated formulation, turbulence generates strongly anisotropic, sheet-like eddies/current sheets, and below a critical scale $\lambda_c$ those eddies can become tearing-unstable before being destroyed by their own nonlinear turnover (Walker et al., 2018). In the sub-grid formulation, unresolved instability-driven turbulence is summarized by a set of turbulent energy densities whose effect on the resolved flow is encoded through closure relations for the turbulent stresses (Miravet-Tenés et al., 2022).

This broad usage includes several distinct mechanisms. One class involves current-sheet disruption by tearing or plasmoid instability, which changes the inertial-range cascade itself. A second class involves instability-regulated transport, as in the magnetic-Prandtl-number-dependent MRI turbulence of accretion discs. A third class treats turbulence as arising from an instability of the statistical state, rather than from a direct linear instability of a laminar base flow. Related developments extend the concept further: the nonlinear self-deformation of unidirectionally propagating waves in structured plasmas was termed “uniturbulence” and was presented as conceptually aligned with instability/structure-driven routes to turbulent transfer (Magyar et al., 2019).

A recurrent diagnostic language across these works is the decomposition of fluctuations into stress correlations. In the mean-field MInIT model, the unresolved dynamics are described by the Maxwell, Reynolds, and Faraday stress tensors,

$$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$

which are then related to instability energies through calibrated coefficients (Miravet-Tenés et al., 2022). This stress-centric formulation makes MInIT simultaneously a physical interpretation of instability-driven turbulence and a practical closure strategy.

2. Current-sheet disruption: tearing- and plasmoid-mediated turbulence

In the tearing-mediated picture, the key competition is between nonlinear eddy turnover and tearing growth. For the anisotropic eddies studied numerically, the nonlinear rate is

$\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$

while the tearing rate for the sine-profile current sheet used in the simulations is

$$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$

The onset criterion $\gamma_t\sim\gamma$ yields

$S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$

so sufficiently elongated eddies become tearing-unstable before ordinary nonlinear destruction (Walker et al., 2018). In the broader cascade model, this defines a transition scale $\lambda_c$, with

$\lambda_c \sim L S_L^{-4/7}$

for a Harris/tanh-like sheet and

$\lambda_c \sim L S_L^{-6/11}$

for the sine-like profile relevant to that study (Walker et al., 2018).

The numerical evidence supports a genuine regime change. At $S=64000$, the tearing time is longer than the eddy turnover time, $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$0 and $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$1, nonlinear shearing dominates, and the spectrum remains close to $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$2. At $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$3, tearing becomes faster than the eddy dynamics, with $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$4, so the eddy is disrupted more quickly by tearing. The intermediate case $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$5 is the transition regime: the spectrum broadens and approaches $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$6, the slope predicted for tearing-mediated turbulence in the Boldyrev scenario (Walker et al., 2018). The significance of this result is that tearing is not merely a reconnection event in a laminar current sheet; it can modify inertial-range dynamics above the dissipation scale.

The plasmoid-mediated extension emphasizes intermittency and dynamic alignment. In two-dimensional, decaying, visco-resistive MHD turbulence at sufficiently large magnetic Reynolds number, especially $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$7, intense intermittent current sheets undergo copious plasmoid formation. The scale-dependent alignment diagnostic

$$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$8

gives $$M_{ij} = B'_i B'_j,\qquad R_{ij} = v'_i v'_j,\qquad F_{ij} = v'_i B'_j - v'_j B'_i,$$9, but the mean alignment alone is insufficient to explain plasmoid onset; the decisive ingredient is the PDF-level intermittency of local alignment and field increments, which produces rare, highly anisotropic structures that reach the critical aspect ratio (Dong et al., 2018). In that plasmoid-mediated regime the magnetic energy spectrum steepens from an inertial-range index close to $\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$0 to an index near $\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$1, and the paper interprets the spectral break as direct evidence that plasmoid disruption facilitates the cascade towards small scales (Dong et al., 2018).

A related, but more explicitly closure-based, formulation is turbulent plasmoid reconnection. There the filtered electromotive force is modeled as

$\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$2

with $\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$3, $\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$4, and $\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$5 determined by turbulent energy, cross-helicity, and helicity. In anti-parallel current sheets, turbulent resistivity enhances reconnection. With guide field, symmetry breaking produces turbulent helicity, which reduces the effective influence of the apparent turbulent resistivity and lowers the reconnection rate relative to the anti-parallel case (Widmer et al., 2016). This establishes that instability-generated turbulence does not universally accelerate reconnection; the feedback is regime-dependent.

3. Instability-driven turbulence beyond tearing

MRI-driven MInIT appears in several forms. In accretion-disc shearing-box simulations with explicit viscosity and resistivity, the Shakura–Sunyaev stress parameter depends on magnetic Prandtl number,

$\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$6

with previous work suggesting $\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$7, and the instability criterion in the analytic framework cited there is $\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$8 (Potter et al., 2017). The stress is measured by

$\gamma \sim \tau_A^{-1} \sim \frac{b_x}{L_x},$9

When turbulent heating, radiative cooling, and a temperature- and density-dependent $$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$0 are included, the local thermal equilibrium,

$$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$1

develops an S-shaped curve with a hot, high-$$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$2, high-$$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$3 branch, a cool, low-$$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$4, low-$$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$5 branch, and an intermediate unstable branch, yielding an unstable limit cycle (Potter et al., 2017). In this formulation, MRI turbulence is self-modulated by microphysical transport coefficients.

Kelvin-Helmholtz-instability-induced turbulence provides a different MInIT pathway. In oscillating coronal loops, shear flows produced by kink oscillations trigger KHI at the loop boundary; the instability broadens the layer into turbulence and transfers wave energy to small scales. The analytical estimate for the dissipation rate gives

$$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$6

and because the velocity shear is proportional to the oscillation amplitude, the result is expressed as $$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$7 (Hillier et al., 2020). Under the steady-state assumption, balancing footpoint energy injection with turbulent dissipation implies a cube-root amplitude law. Applying that estimate to observed decayless kink waves leads to the conclusion that the observed amplitudes are probably insufficient to turbulently heat the solar corona through this mechanism alone (Hillier et al., 2020).

In relativistic pair-plasma turbulence, collisionless tearing supplies the microphysical disruption mechanism. The study of guide-field current sheets with a power-law distribution $$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$8 distinguishes pressure-supported and force-free regimes. In pressure-supported sheets, spectral hardening with $$\gamma_t \sim \frac{b_x}{L_y} S^{-3/7},\qquad S=\frac{b_x L_y}{\eta}.$$9 suppresses tearing strongly; in the force-free limit relevant for $\gamma_t\sim\gamma$0, the growth rate becomes essentially spectrum-independent. The resulting disruption scales for critically balanced relativistic turbulence are

$\gamma_t\sim\gamma$1

for elongated eddies and

$\gamma_t\sim\gamma$2

for intermittent current sheets, while charge starvation imposes a critical magnetization threshold above which the plasma cannot supply the required current (Demidov et al., 2024). This both supports and constrains relativistic MInIT.

4. Statistical-state and modulational formulations

A distinct line of work interprets MInIT as an instability of the statistical state. In incompressible conducting Couette flow, the statistical state dynamics (SSD) / S3T formulation splits velocity and magnetic field into streamwise mean and streamwise fluctuation,

$\gamma_t\sim\gamma$3

and evolves the coupled mean state and covariance through

$\gamma_t\sim\gamma$4

The resulting composite velocity–magnetic roll-streak structure is an unstable mode of the SSD, produced by a mean–covariance feedback in which small mean perturbations reorganize the fluctuation covariance and the resulting Reynolds and Maxwell stresses reinforce the mean structure (Kim et al., 22 Oct 2025). The instability equilibrates either to a fixed point or to a turbulent statistical equilibrium, and self-sustaining magnetic coherence appears above a critical value near

$\gamma_t\sim\gamma$5

The paper interprets this as turbulence maintained by instability in the statistical state rather than by a prescribed linear instability of the laminar base flow (Kim et al., 22 Oct 2025).

A closely related, but spectrally reduced, formulation treats structure formation in two-dimensional incompressible MHD as modulational instability. With a primary harmonic $\gamma_t\sim\gamma$6 and a weak modulation $\gamma_t\sim\gamma$7, the perturbation occupies the full sideband chain

$\gamma_t\sim\gamma$8

which leads to an infinite block-tridiagonal harmonic system. Low-order closures such as 4MT can reproduce modulational growth rates in some regimes, but they can also falsely predict instability because they fail to capture propagating spectral waves (PSWs), which carry energy ballistically along the harmonic chain to dissipative scales and thereby break the feedback loop sustaining modulational instability (Jin et al., 2024). The outgoing-wave spectral closure proposed there treats instability and spectral transport on the same footing. This suggests that in ideal MHD, closure accuracy is unusually sensitive to nonlocal harmonic transport.

Mean-field wave kinetics generalizes the modulational picture by replacing local closure coefficients with a nonlocal phase-space description based on the Wigner–Weyl transform. The fluctuation spectrum is encoded in the Wigner matrix

$\gamma_t\sim\gamma$9

which satisfies the Wigner–Moyal equation

$S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$0

The turbulent electromotive force,

$S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$1

is then obtained as a nonlocal functional of the mean fields without assuming scale separation. In the appropriate limits the standard $S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$2-dynamo is recovered, while a new dynamo effect driven by $S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$3 is predicted (Jin et al., 25 Feb 2025). Within this framework, mean-field generation is a modulational instability of MHD turbulence itself.

5. MInIT as a sub-grid mean-field model

The sub-grid MInIT model was introduced as a physically motivated closure for unresolved instability-driven MHD turbulence. In the MRI version, the model evolves the turbulent energy densities of the MRI channels and parasitic instabilities,

$S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$4

$S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$5

with

$S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$6

The stresses are reconstructed as linear functions of $S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$7 and $S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$8, the system is integrated with Strang splitting, and assessment against filtered DNS using the relative $S_c\sim \left(\frac{L_x}{L_y}\right)^{7/3},$9 norm found “less than one order-of-magnitude difference” between model and DNS for the Maxwell and Reynolds stresses, with essentially no dependence on filter size and better robustness than the gradient model in the tested regime (Miravet-Tenés et al., 2022).

The KHI version simplifies the energetic state to a single sub-grid energy density,

$\lambda_c$0

which evolves according to

$\lambda_c$1

Calibration against well resolved 3D periodic-box simulations with the Aenus code yielded an optimal dissipation constant

$\lambda_c$2

while the reported accuracy was $\lambda_c$3 for Maxwell and Reynolds stresses and $\lambda_c$4 for the Faraday stress (Miravet-Tenés et al., 2023). The model’s stated difference from alpha-viscosity and gradient closures is that it tracks independently the turbulent energy density at sub-grid scales rather than inferring all unresolved effects algebraically from resolved gradients.

The model has also been implemented in global Newtonian simulations of MRI-sensitive, differentially rotating, magnetised neutron stars. In that implementation, the turbulent stresses are added directly to the momentum equation, but no sub-grid terms are introduced in the energy or induction equations, so MRI-driven large-scale magnetic amplification is not modeled. In the $\lambda_c$5 run, $\lambda_c$6 grows from about $\lambda_c$7 to $\lambda_c$8 within the first $\lambda_c$9 ms over much of the interior, and the resulting stresses flatten the differential rotation profile far more strongly than numerical dissipation alone (Miravet-Tenés et al., 8 Sep 2025). The same study stresses the scale-separation problem that motivates MInIT: for $\lambda_c \sim L S_L^{-4/7}$0 with $\lambda_c \sim L S_L^{-4/7}$1 G, the radial resolution is about $\lambda_c \sim L S_L^{-4/7}$2 m while $\lambda_c \sim L S_L^{-4/7}$3 m through most of the star, and fully resolving the MRI in 3D for that setup was estimated to require on the order of $\lambda_c \sim L S_L^{-4/7}$4 years of CPU time (Miravet-Tenés et al., 8 Sep 2025).

6. Astrophysical domains, universality, and limitations

The astrophysical range of MInIT is broad. The sub-grid model was developed with binary neutron star merger remnants, hypermassive neutron stars, proto-neutron stars, and magnetized accretion flows in view (Miravet-Tenés et al., 2022), and its first global implementation targeted differentially rotating neutron stars in post-merger conditions (Miravet-Tenés et al., 8 Sep 2025). MRI–Pm instability was proposed as a mechanism for local thermal-viscous instability in black hole binaries, with implications for hard/soft state transitions and propagating heating and cooling fronts (Potter et al., 2017). KHI-induced turbulence was analyzed as a possible contributor to coronal heating (Hillier et al., 2020). Relativistic tearing-mediated variants were connected to pulsar wind nebulae, relativistic jets, and gamma-ray flares (Demidov et al., 2024).

At the same time, several papers restrict the range of validity of any unified MInIT picture. One important qualification is universality at small scales: MRI-driven turbulence in incompressible shearing boxes was argued to become statistically indistinguishable from standard homogeneous MHD turbulence at sufficiently small scales, with dissipation concentrated in thin sheet-like structures and a preferred tilt of approximately $\lambda_c \sim L S_L^{-4/7}$5 degrees in the toroidal direction (Zhdankin et al., 2017). This suggests that an instability-driven large-scale origin does not imply non-universal small-scale structure.

Another qualification concerns closure and forcing assumptions. In plasmoid-mediated turbulence, the spectral break and steepening were demonstrated in decaying two-dimensional turbulence, and the same paper explicitly noted that theoretical predictions for the break scale had largely been developed for forced turbulence (Dong et al., 2018). In modulational reduced models, low-order truncations can falsely predict instability because PSWs are omitted, especially in ideal MHD (Jin et al., 2024). In guide-field plasmoid reconnection, instability-generated turbulence can either enhance or partially suppress reconnection, depending on the balance between turbulent resistivity and helicity effects (Widmer et al., 2016). In relativistic pair plasmas, tearing can fail to control the cascade if there is insufficient scale separation or if charge starvation intervenes before tearing develops (Demidov et al., 2024).

The sub-grid MInIT program has its own explicit limitations. The neutron-star implementation is a proof of concept in Newtonian, axisymmetric, polytropic models; it is not fully covariant, it omits sub-grid terms in the induction equation, and it therefore does not capture a sub-grid dynamo or the post-saturation magnetic feedback on the MRI itself (Miravet-Tenés et al., 8 Sep 2025). A plausible implication is that “MInIT” is best regarded not as a single universal closure or a single instability, but as a family of instability-mediated feedback mechanisms whose mathematical form and physical consequences depend strongly on geometry, scale separation, symmetry breaking, and the choice of coarse-graining.