Magneto-Thermal Evolution Models

• Magneto-thermal evolution models are time-dependent frameworks that co-evolve magnetic and thermal states via mechanisms like Joule heating, anisotropic conductivity, and Hall drift.
• They apply a range of governing equations and numerical strategies to address systems such as neutron stars, superconducting magnets, and micromagnetic devices.
• These models enhance understanding of energy dissipation, field decay, and mechanical stress by integrating detailed thermal transport and magnetic diffusion in a coupled evolution.

Magneto-thermal evolution models are time-dependent frameworks in which magnetic variables and thermal variables are evolved together in space and time, with feedback mediated by dissipation, anisotropic transport, temperature-dependent constitutive laws, and, in some settings, circuit dynamics, stochastic forcing, or mechanical loading. In the literature, the term encompasses crustal Hall–Ohmic cooling of neutron stars, bidirectionally coupled micromagnetic–heat systems, quench evolution in superconducting accelerator magnets, radiative MHD of the solar atmosphere, convective suppression in magnetic white dwarfs, and thermal histories of rocky planets with possible dynamos (Maciejewski et al., 2017, Yi et al., 11 Jun 2026, 0812.3018).

1. Scope, taxonomy, and defining features

A magneto-thermal evolution model is not a single equation set but a class of coupled initial–boundary-value problems. What remains invariant across implementations is the co-evolution of a magnetic state and a thermal state, together with explicit mechanisms by which one modifies the transport or source terms of the other. In compact-object work this usually means magnetic diffusion, Hall drift, anisotropic heat transport, and Joule heating; in micromagnetics it means stochastic spin dynamics with local heat-bath feedback; in superconducting magnets it means quench heating, current redistribution, and Lorentz-force loading; in solar-atmosphere simulations it means radiative MHD augmented by anisotropic thermal conduction and radiative losses (Viganò et al., 2013, Yi et al., 11 Jun 2026, Maciejewski et al., 2017, Navarro et al., 7 Nov 2025).

System	Evolving quantities	Dominant coupling mechanism
Superconducting accelerator magnets	$T(\mathbf{x},t)$, $\mathbf{A}$, circuit currents, stresses	Ohmic and coupling-loss heating, Lorentz forces, thermal expansion
Micromagnetics	$\mathbf{m}(\mathbf{r},t)$, $T(\mathbf{r},t)$	Damping and stochastic work exchanged with a finite thermal bath
Isolated neutron stars	$\mathbf{B}$, $T$	Hall drift, Ohmic decay, Joule heating, anisotropic conductivity
Solar atmosphere	$\rho,\mathbf{v},\mathbf{B},T$	Radiative losses and anisotropic thermal conduction in MHD
White dwarfs and rocky planets	Envelope convection or interior heat flow, magnetic activity proxies	Magnetic suppression of convection or dynamo viability from thermal history

One major distinction is between one-way and bidirectional coupling. In the superconducting-magnet case study, the magneto-thermal solution is the driver and the mechanical solve is post-processed from interpolated temperatures and Lorentz-force densities, with no feedback from deformation to the quench simulation (Maciejewski et al., 2017). By contrast, the micromagnetic formulation based on the stochastic Landau–Lifshitz–Gilbert equation and a generalized heat equation is explicitly closed-loop: the local temperature sets the noise amplitude, while damping-induced dissipation and stochastic work feed back into the thermal bath (Yi et al., 11 Jun 2026).

A second distinction concerns the role of auxiliary physics. Some models are purely magneto-thermal; others are magneto-thermal–mechanical, magneto-thermal–rotational, or magneto-thermal–radiative. Population-synthesis studies of isolated neutron stars, for example, do not re-solve the crustal PDEs directly; instead they import $B(t)$ from detailed magneto-thermal simulations and couple it to spin evolution and selection effects across radio and X-ray bands (Gullón et al., 2014, Gullón et al., 2015).

2. Governing equations and coupling architectures

The mathematical core is usually an induction-type equation for magnetic evolution plus a heat-balance equation. In crustal neutron-star models, the canonical induction equation includes both Ohmic diffusion and Hall drift,

$$\frac{\partial \vec{B}}{\partial t} = - \vec{\nabla}\times\left[ \frac{c^2}{4\pi\sigma}\,\vec{\nabla}\times\vec{B} + \frac{c}{4\pi e n_e}\,(\vec{\nabla}\times\vec{B})\times\vec{B} \right],$$

while the thermal equation includes anisotropic conductivity and Joule heating (Viganò et al., 2013). In the earlier 2D crustal framework, the mutual feedback of Joule heating and magnetic diffusion is strong enough that stronger initial fields dissipate faster during the first million years, and fields larger than about $5\times 10^{13}\,\mathrm{G}$ evolve toward similar late-time strengths (0812.3018).

In superconducting accelerator magnets, the magnetic state is represented magnetoquasistatically through a vector potential $\mathbf{A}$0, with current supplied by a coupled circuit model rather than prescribed externally. The thermal field satisfies

$\mathbf{A}$1

where $\mathbf{A}$2 encodes cable coupling-current losses, and the Lorentz force density is

$\mathbf{A}$3

These fields then act as loads in a structural model (Maciejewski et al., 2017).

The micromagnetic bidirectional framework makes the energy exchange especially explicit. Starting from the stochastic LLG equation, the model derives a local power balance in which the damping term is a positive-definite heat source and the stochastic term can either heat or cool the local bath. The associated heat equation is

$\mathbf{A}$4

and the authors prove, using Itô calculus, that the coupled system obeys the first law and recovers the correct Boltzmann statistics at equilibrium (Yi et al., 11 Jun 2026).

Radiative MHD formulations enlarge the thermal sector further. In MAGEC, the total-energy equation includes gravity, anisotropic thermal conduction, optically thick LTE radiation, and optically thin losses. The internal-energy form isolates the terms most relevant to magneto-thermal interpretation: $\mathbf{A}$5 The conductive flux can be treated in Braginskii form, retaining parallel, perpendicular, and Hall components relative to $\mathbf{A}$6 (Navarro et al., 7 Nov 2025).

Taken together, these formalisms show that “magneto-thermal” may refer either to direct magnetic-energy dissipation into heat, to thermal control of transport coefficients, or to a fully conservative energy-exchange problem in which magnetic and thermal subsystems are coequal dynamical reservoirs. That distinction is methodologically important and is explicit in the underlying literature (Yi et al., 11 Jun 2026, Maciejewski et al., 2017).

3. Numerical strategies, dimensional reduction, and multiphysics coupling

The numerical realization of magneto-thermal evolution is highly problem-dependent. The superconducting-magnet study uses a 2D half-turn magneto-thermal finite-element model in COMSOL, a detailed mechanical model in ANSYS APDL, and a circuit solved in ORCAD PSpice under the STEAM framework. Field/circuit coupling is performed by waveform relaxation over time windows, while transfer from the magneto-thermal mesh to the structural mesh is handled by MpCCI through finite-element shape-function interpolation rather than nearest-neighbor mapping (Maciejewski et al., 2017).

That workflow is explicitly one-way in time. The magneto-thermal solution is computed over $\mathbf{A}$7, then the structural response is evaluated at 103 communication times chosen to resolve the CLIQ oscillation: every 1 ms between 0 and 30 ms and every 10 ms up to 500 ms (Maciejewski et al., 2017). This architecture is representative of a large class of industrial magneto-thermal models in which the thermal and electromagnetic problem is the expensive transient driver and mechanical integrity is assessed snapshot by snapshot.

Three-dimensional neutron-star crust models have followed two very different numerical traditions. The PARODY-based framework expands angular dependence spectrally in spherical harmonics and discretizes radius on a finite grid, evolving poloidal and toroidal magnetic scalars together with temperature in a spherical shell. The scheme is mixed implicit–explicit: Crank–Nicolson for linear diffusive terms, backward Euler for the isotropic part of thermal diffusion, and Adams–Bashforth for Hall, thermoelectric, and Joule-heating terms (Grandis et al., 2020). By contrast, MATINS adopts a finite-volume scheme on a cubed-sphere grid, thereby avoiding axial singularities and enabling fully 3D Hall-dominated crustal evolution with constrained-transport-like divergence control (Dehman et al., 2022).

The MATINS formalism is explicitly designed for strong-field crustal evolution with complex non-axisymmetric topology. Its first published version focuses on magnetic evolution with a prescribed cooling law rather than a full heat solve, but it already incorporates realistic temperature-dependent conductivity from Potekhin’s microphysics code and general-relativistic factors from a TOV background (Dehman et al., 2022). This suggests a staged methodology in which geometric and Hall-cascade fidelity is established before full thermal back-reaction is activated.

Solar-atmosphere simulations face a different stiffness problem: anisotropic thermal conduction would impose a parabolic timestep restriction if handled directly. MAGEC therefore uses a finite-volume, shock-capturing MHD solver with constrained transport for $\mathbf{A}$8, combined with a hyperbolic treatment of thermal conduction through a relaxation equation for $\mathbf{A}$9. This converts the conductive operator into a telegraph-like form and mitigates restrictive timesteps while retaining anisotropic heat transport (Navarro et al., 7 Nov 2025).

At smaller scales, the bidirectional micromagnetic framework relies on stochastic time integration consistent with Itô calculus. In the cited implementation, 0D macrospin SDE simulations are complemented by 1D COMSOL micromagnetic chains of YIG with $\mathbf{m}(\mathbf{r},t)$0, $\mathbf{m}(\mathbf{r},t)$1, and $\mathbf{m}(\mathbf{r},t)$2, using a generalized-$\mathbf{m}(\mathbf{r},t)$3 scheme tuned to handle white noise (Yi et al., 11 Jun 2026). The numerical requirement here is not mesh interpolation or Hall stability but preservation of the fluctuation–dissipation structure and of the equilibrium measure.

4. Compact objects: neutron-star crusts, cores, and populations

In neutron-star physics, magneto-thermal evolution has become a unifying framework for magnetic-field decay, anisotropic surface heating, timing evolution, and source classification. The first long-term 2D coupled crustal simulations showed that Joule heating and magnetic diffusion are strongly coupled, so that neutron stars born with fields larger than a critical value of about $\mathbf{m}(\mathbf{r},t)$4 reach similar late-time field strengths, approximately $\mathbf{m}(\mathbf{r},t)$5, while after $\mathbf{m}(\mathbf{r},t)$6 years the temperature becomes so low that magnetic diffusion is effectively frozen on radio-pulsar timescales (0812.3018). Later 2D models that included the Hall term and state-of-the-art kinetic coefficients argued that the observed diversity of magnetars, high-$\mathbf{m}(\mathbf{r},t)$7 radio pulsars, and nearby isolated neutron stars can be reproduced by varying initial magnetic field, mass, and envelope composition (Viganò et al., 2013).

Three-dimensional crustal simulations altered the geometric picture substantially. The PARODY-based 3D Hall–Ohmic–thermal calculations found that a resistive tearing instability develops in about a Hall time if the initial toroidal field exceeds $\mathbf{m}(\mathbf{r},t)$8, producing localized high-$\mathbf{m}(\mathbf{r},t)$9 patches, enhanced dissipation, and crustal stresses near the yield threshold (Grandis et al., 2020). The same framework showed that localized crustal heating generates surface hot spots that drift and deform along magnetic field lines, and that crescent-like patterns reminiscent of NICER inferences for PSR J0030+0451 can arise without invoking strong surface multipoles (Grandis et al., 2020).

MATINS complements that pseudo-spectral approach with a finite-volume 3D Hall code. Its long-term crustal calculations show that the non-axisymmetric Hall cascade redistributes energy over different spatial scales and that, during a few tens of kyr, an equipartition of energy between the poloidal and toroidal components happens at small scales, while the large-scale field retains a strong memory of the initial topology (Dehman et al., 2022). This is important because spin-down and observational classification are controlled mainly by the large-scale component, whereas dissipation and surface temperature structure depend strongly on the small-scale cascade.

A separate line of work addresses the hot neutron-star core in the strong-coupling regime. There the particles behave as a single, stably stratified, non-barotropic fluid, and the relevant magneto-thermal question is whether Urca reactions can drive the field toward a Grad–Shafranov equilibrium before the core cools. The answer given by the cited simulations is negative: even for magnetar-strength fields $T(\mathbf{r},t)$0, the feedback from magnetic evolution on thermal evolution is negligible, the star cools essentially passively, and the Grad–Shafranov equilibrium is not reached before the transition to the weak-coupling regime (Moraga et al., 2023).

Three-dimensional simulations of tangled crustal fields in central compact objects extend the framework further by abandoning large-scale dipole-dominated initial conditions. For initial magnetic energies of $T(\mathbf{r},t)$1, the models produce multiple hot regions reaching $T(\mathbf{r},t)$2 keV against a bulk temperature of $T(\mathbf{r},t)$3 keV, with pulsed fractions $T(\mathbf{r},t)$4 (Igoshev et al., 2021). The small pulsed fraction follows from the simultaneous visibility of multiple hot spots under gravitational light bending, not from weak temperature anisotropy.

Population-synthesis studies then project these magneto-thermal tracks into source counts and timing distributions. One such study found that the absence of isolated neutron stars with $T(\mathbf{r},t)$5 s imposes an upper limit to the fraction of magnetars born with $T(\mathbf{r},t)$6, less than $T(\mathbf{r},t)$7 (Gullón et al., 2015). This indicates that magneto-thermal evolution models can be constrained not only by individual objects but by ensemble demographics.

5. Other physical systems and coupled extensions

In superconducting accelerator magnets, magneto-thermal evolution is driven by quench initiation, propagation, and protection circuitry. The cited 11 T Nb$T(\mathbf{r},t)$8Sn dipole example couples the field solution to the circuit, including a CLIQ unit with $T(\mathbf{r},t)$9, $\mathbf{B}$0, and pole inductances $\mathbf{B}$1, then maps time-dependent temperature and Lorentz-force density into a structural model. The maximum equivalent stress is $\mathbf{B}$2 before CLIQ, $\mathbf{B}$3 at the first current peak around $\mathbf{B}$4 ms, and rises again to $\mathbf{B}$5 at $\mathbf{B}$6 ms because late thermal gradients dominate after the electromagnetic transient has subsided (Maciejewski et al., 2017). A central implication is that the mechanically most severe state need not coincide with the maximum current.

In micromagnetics, the 2026 bidirectional model provides an explicitly thermodynamic interpretation of magneto-thermal evolution at cell scale. The temperature is a dynamical field rather than a prescribed reservoir; damping-induced dissipation heats the bath, stochastic work can cool it, and the coupled stochastic process recovers the correct Boltzmann equilibrium (Yi et al., 11 Jun 2026). The same framework predicts finite-bath cooling,

$\mathbf{B}$7

for an equilibrating macrospin, showing that even thermal noise can no longer be treated as an external, cost-free source once the reservoir is finite (Yi et al., 11 Jun 2026).

Solar-atmosphere models bring radiative transfer and magnetic topology to the foreground. MAGEC combines MANCHA and MAGNUS in an MPI-parallelized finite-volume, shock-capturing framework with LTE radiative losses and hyperbolic thermal conduction. In the reported 2D magneto-convection experiments, all simulations reached thermal equilibrium, open-field cases produced higher coronal temperatures than closed, arcade-like fields, and perpendicular thermal conduction, though often neglected, influenced plasma dynamics near reconnection and could cumulatively modify the average coronal temperature (Navarro et al., 7 Nov 2025).

For magnetic white dwarfs, the relevant magneto-thermal mechanism is not Hall drift or Joule heating but magnetic inhibition of convection. Radiation-MHD atmosphere simulations show that convective energy transfer is seriously impeded when the plasma-beta parameter falls below unity, with critical field strengths in the photosphere in the range $\mathbf{B}$8 (Gullón et al., 2015). Yet evolutionary calculations show that complete suppression of convection has no effect on cooling rates until $\mathbf{B}$9 reaches around $T$0, because only then does convective coupling connect the outer envelope to the degenerate thermal reservoir (Gullón et al., 2015).

Rocky-planet models generalize magneto-thermal evolution to dynamos in both a liquid iron core and a magma ocean. The cited framework solves mantle and core energy balances, recomputes hydrostatic structure with a Henyey solver, and evaluates dynamo operation using a critical magnetic Reynolds number. It finds that only layers with melt fraction greater than a critical value of $T$1 may contribute to the dynamo source region in the magma ocean, that mantle mass controls the thermal insulating effect on the iron core, and that for a core thermal conductivity of $T$2 the dynamo lifetime in the iron core is limited by the liquid-core lifetime for $T$3 planets and by the lack of thermal convection for $T$4 planets (Zhang et al., 2022).

6. Limitations, validation, and recurring misconceptions

A recurring limitation is dimensional reduction. The 2D crustal neutron-star models neglect non-axisymmetric structure; the superconducting-magnet quench model is 2D in cross-section and neglects end effects and axial heat conduction; the first MATINS paper prescribes a temperature history instead of solving the full thermal PDE; and the MAGEC validation study is restricted to 2D magneto-convection with a fixed-temperature upper boundary (0812.3018, Maciejewski et al., 2017, Dehman et al., 2022, Navarro et al., 7 Nov 2025). These are not minor implementation choices but governing assumptions that determine which branches of the coupled dynamics are accessible.

A second limitation concerns closure of the coupling. One-way strategies are common and often deliberate. In the superconducting-magnet application there is no feedback from mechanical deformation to the magneto-thermal solution, so strain-dependent $T$5, geometry change, and dynamic mechanical effects are excluded (Maciejewski et al., 2017). In the adolescent neutron-star core, the situation is almost the opposite: the simulations show that magnetic feedback on thermal evolution is negligible in the strong-coupling regime, so a decoupled passive-cooling treatment becomes asymptotically accurate there (Moraga et al., 2023). The phrase “magneto-thermal coupling” therefore does not, by itself, specify the strength or symmetry of the feedback.

Validation practices vary accordingly. The micromagnetic bidirectional model is validated by analytic energy conservation and by recovery of the correct Boltzmann distribution (Yi et al., 11 Jun 2026). MATINS is validated through pure Ohmic-decay eigenmode tests, comparison with a 2D axisymmetric Hall–Ohm code, and explicit monitoring of the discrete divergence measure $T$6 relative to current-related norms (Dehman et al., 2022). The superconducting-magnet coupling paper emphasizes that the underlying field/circuit and mechanical models had been benchmarked separately and that the coupling workflow preserved the spatial structure of temperature and force fields visually after interpolation (Maciejewski et al., 2017). The 3D neutron-star crustal calculations focus more on phenomenological plausibility and on consistency with previously known Hall-attractor behavior and observational hot-spot morphologies (Grandis et al., 2020).

Several recurring misconceptions are explicitly challenged by the literature. First, peak current or peak temperature is not, by itself, a sufficient design metric: in the Nb$T$7Sn dipole example, the largest stresses occur late, when thermal gradients dominate rather than at the initial current overshoot (Maciejewski et al., 2017). Second, complex neutron-star hot-spot geometry does not necessarily require strong surface multipoles; drifting and crescent-like hot regions can arise from anisotropic heat transport in comparatively simple large-scale fields (Grandis et al., 2020). Third, magnetic suppression of surface convection does not imply an immediate change in global cooling rates; in magnetic white dwarfs the effect is delayed until convective coupling reaches the degenerate reservoir (Gullón et al., 2015). Fourth, a model that injects thermal noise into magnetization dynamics without allowing energetic feedback to the bath is not thermodynamically closed; the bidirectional micromagnetic formulation was developed precisely to eliminate that inconsistency (Yi et al., 11 Jun 2026).

Taken together, these works define magneto-thermal evolution models less by any particular equation set than by a methodological commitment: thermal transport, magnetic transport, and energy conversion must be solved on the same dynamical footing appropriate to the system, with the dominant couplings made explicit and the neglected ones stated as modeling assumptions.