Modified Focused Transport Equation

• Modified focused transport equations are Fokker–Planck-type models that adjust traditional formulations to conserve particle number in varying magnetic fields.
• They incorporate conservative reformulations, finite-dimensional subspace reductions, and telegraph closures to better capture pitch-angle dynamics and early-time anisotropy.
• These adaptations ensure numerical consistency with the linear Boltzmann equation and enhance the modeling accuracy of cosmic-ray and energetic-particle transport processes.

Searching arXiv for relevant papers on modified focused transport equations and related formulations. The modified focused transport equation denotes a class of Fokker–Planck-type transport models in which a standard focused or forward-peaked transport operator is altered to enforce a specific structural property of the underlying high-order dynamics. In the energetic-particle literature, the term most commonly refers to the conservative form obtained from a transformed distribution function in a spatially varying magnetic field, while in forward-peaked slab transport it refers to a modified Fokker–Planck low-order equation augmented by a consistency term so that it shares the fixed point of the linear Boltzmann equation. Related work also uses finite-dimensional subspace reductions and telegraph closures as modified focused transport descriptions when a full pitch-angle equation is replaced by a closed system for a small number of moments (Klippenstein et al., 15 Jul 2025, Kuczek et al., 2020, Klippenstein et al., 14 Apr 2025).

1. Standard formulations from which modified equations are built

In one standard energetic-particle formulation, the focused transport equation for a gyrotropic distribution $f(t,z,\mu)$ in a non-uniform mean magnetic field is

$$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$

where $z$ is the coordinate along the field, $\mu=\cos\theta$ is the pitch-angle cosine, $v$ is the particle speed, $D_{\mu\mu}$ is the pitch-angle diffusion coefficient, and $L$ is the focusing length. In the “standard form” emphasized in the subspace-approximation literature, the model is one-dimensional in space, one-dimensional in pitch angle, and contains streaming, pitch-angle diffusion, and focusing only, with no energy changes, no perpendicular spatial transport, and no background plasma flow term (Klippenstein et al., 14 Apr 2025).

A related SEP review writes the one-dimensional focused transport equation along a field line $s$ in conservative form,

$$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$

with $L^{-1}(s)=-(1/B)\,dB/ds$. In this formulation, focusing is the deterministic pitch-angle drift generated by the spatial variation of the large-scale magnetic field magnitude, and the coefficient $$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$0 encodes resonant pitch-angle scattering (Berg et al., 2020).

In slab-geometry particle transport with highly forward-peaked scattering, the starting point is instead the monoenergetic, steady-state linear transport equation

$$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$1

or its Legendre-expanded form

$$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$2

Its standard forward-peaked Fokker–Planck limit is

$$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$3

with $$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$4 and $$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$5 (Kuczek et al., 2020).

2. Conservative modification in energetic-particle transport

The most explicit use of the term “modified focused transport equation” in recent energetic-particle work is the conservative reformulation obtained by removing the geometric dilution associated with an expanding flux tube. For a field $$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$6, with cross-sectional area $$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$7, the transformed distribution

$$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$8

leads to

$$\frac{\partial f}{\partial t} + v \mu \frac{\partial f}{\partial z} = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right] - \frac{v}{2L}(1-\mu^2)\frac{\partial f}{\partial \mu},$$9

Compared with the standard equation, the focusing term is rewritten as a divergence in $z$0 acting on $z$1, not on $z$2. The paper then drops the tilde and denotes the solution of the modified equation again by $z$3 (Klippenstein et al., 15 Jul 2025).

The principal distinction is structural. The standard form does not conserve the norm $z$4 because the physical particle number in a flux tube carries the geometric weight $z$5. In the modified form, the pitch-angle operator becomes a full divergence,

$z$6

so that, under the usual endpoint conditions at $z$7 and suitable spatial boundary conditions,

$z$8

The same reformulation changes the pitch-angle relaxation problem: the modified equation conserves the norm but does not describe pitch-angle isotropization in the same way as the standard form (Klippenstein et al., 15 Jul 2025).

This distinction is visible already in low-order observables. In the two-dimensional subspace approximation with isotropic scattering $z$9, the modified equation yields

$\mu=\cos\theta$0

so that

$\mu=\cos\theta$1

rather than relaxing to zero. By contrast, the corresponding standard-form analysis gives $\mu=\cos\theta$2, which is the signature of isotropization (Klippenstein et al., 15 Jul 2025, Klippenstein et al., 14 Apr 2025).

3. Subspace approximations as reduced modified equations

A second important line of work treats finite-dimensional moment systems as modified focused transport equations. After Fourier transforming in space,

$\mu=\cos\theta$3

the pitch-angle dependence is expanded in Legendre polynomials,

$\mu=\cos\theta$4

For isotropic scattering $\mu=\cos\theta$5, projection onto $\mu=\cos\theta$6 gives an infinite tridiagonal chain. For the modified conservative equation the coefficients satisfy

$\mu=\cos\theta$7

Truncating at $\mu=\cos\theta$8 defines an $\mu=\cos\theta$9-dimensional subspace approximation, so that the original PDE is replaced by a finite matrix problem $v$0 (Klippenstein et al., 15 Jul 2025).

At $v$1, only $v$2 and $v$3 are retained: $v$4 The corresponding $v$5 system has eigenvalues

$v$6

and the small-$v$7 expansion yields

$v$8

with $v$9. In real space this implies drift-diffusion behavior for the pitch-angle averaged distribution (Klippenstein et al., 15 Jul 2025).

The standard-form subspace paper makes the broader conceptual claim that such finite-$D_{\mu\mu}$0 systems are themselves modified focused transport equations, because the full Fokker–Planck operator in $D_{\mu\mu}$1 is replaced by a closed set of coupled evolution equations for a finite set of pitch-angle moments. In that sense, the two-dimensional truncation is telegraph-like, the three-dimensional truncation retains the $D_{\mu\mu}$2 moment, and higher-dimensional reductions such as $D_{\mu\mu}$3 function as hybrid analytical-numerical surrogates for the full equation (Klippenstein et al., 14 Apr 2025).

The practical hierarchy is explicit. The modified-form study finds that $D_{\mu\mu}$4 is useful for analytical insight, $D_{\mu\mu}$5 is noticeably improved, and $D_{\mu\mu}$6 shows excellent agreement with full numerical solutions for the quantities tested for realistic focusing parameters $D_{\mu\mu}$7, while remaining much faster than a full implicit Euler solver in $D_{\mu\mu}$8. The same work also stresses that truncated systems can develop positive-real-part eigenvalues for certain $D_{\mu\mu}$9, so convergence of the inverse Fourier transform must be checked, especially at large focusing parameter $L$0 (Klippenstein et al., 15 Jul 2025).

4. Diffusion–advection and telegraph closures

Moment closure at still lower order produces diffusion–advection and telegraph equations for the isotropic density. In focused cosmic-ray transport, the modified telegraph equation for $L$1 is

$L$2

where $L$3. If one instead uses the linear density

$L$4

the equation becomes

$L$5

For isotropic pitch-angle scattering $L$6, the dimensionless transport coefficients are

$L$7

so the telegraph approximation adds both a focusing-induced convective term and a finite-speed hyperbolic correction (Litvinenko et al., 2015).

The same formulation admits boundary conditions that are not the diffusion limits. At reflecting boundaries, one obtains

$L$8

At absorbing boundaries on $L$9, one obtains

$s$0

or equivalently

$s$1

with $s$2 at the left boundary and $s$3 at the right boundary. These conditions reflect the hyperbolic character of the telegraph equation and differ from the diffusion prescription $s$4 at an absorbing boundary (Litvinenko et al., 2015).

Within SEP transport, the diffusion–advection and telegraph equations are best regarded as reduced modified equations rather than full replacements for focused transport. The SEP review concludes that the diffusion–advection approximation is too diffusive at early times and predicts instantaneous propagation, while the telegraph approximation improves causality and early-time behavior but still does not capture the full complexity of the physical processes involved, particularly strong time-dependent anisotropy, ballistic early propagation, and realistic pitch-angle scattering effects. The review therefore treats these closures as limited approximations to the full focused transport equation rather than equivalent descriptions (Berg et al., 2020).

5. Modified Fokker–Planck acceleration for forward-peaked slab transport

In slab geometry with highly forward-peaked scattering, a distinct modified focused transport equation arises as a low-order model inside a nonlinear high-order/low-order acceleration scheme. The standard Fokker–Planck equation,

$s$5

is altered by an additive consistency term $s$6,

$s$7

The correction is defined by

$s$8

that is, by the discrepancy between the full scattering operator and the Fokker–Planck operator plus isotropic scattering, evaluated on the current high-order solution. This makes the low-order equation algebraically consistent with the high-order transport equation when both see the same flux (Kuczek et al., 2020).

In the NFPA formulation, the high-order transport sweep uses low-order moments,

$s$9

while the low-order modified Fokker–Planck solve uses the consistency term computed from the high-order flux. At convergence, the Legendre moments of the low-order flux match those of the high-order flux for all $$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$0 up to the truncation order $$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$1, and the two systems share the same fixed point. The paper identifies this as the first Fokker–Planck-like equation that is discretely consistent with the linear Boltzmann equation and the first nonlinear HOLO method that accelerates all $$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$2 moments of the angular flux, not just the first moment (Kuczek et al., 2020).

The numerical behavior depends on the scattering kernel. The study considers the Screened Rutherford kernel, an Exponential kernel constructed to possess a clear Fokker–Planck limit, and the Henyey–Greenstein kernel. For the Screened Rutherford and Exponential kernels, NFPA and the standard FP-based synthetic acceleration achieve speed-ups of $$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$3–$$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$4 orders of magnitude in wall-clock time relative to DSA. The low-order modified FP solution remains very close to the full transport solution over a wide range of the forward-peaking parameter, whereas the stand-alone FP solution’s error grows by orders of magnitude as the problem becomes less forward-peaked. For Henyey–Greenstein scattering, which does not admit a valid Fokker–Planck limit, the method still outperforms DSA and GMRES, but the acceleration efficiency is reduced as anisotropy increases (Kuczek et al., 2020).

6. SDE realizations, anisotropy, and relation to Parker transport

Focused transport equations are frequently implemented through stochastic differential equations. In a CRPropa-based solver, the focused pitch-angle transport equation is written as

$$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$5

with corresponding Itô equations

$$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$6

Reflective boundaries are imposed in $$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$7, rather than periodic boundaries, so that pitch-angle changes remain physically continuous near $$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$8. For $$\frac{\partial f}{\partial t} + \frac{\partial}{\partial s}\big[\mu v f\big] + \frac{\partial}{\partial \mu}\left[\frac{(1-\mu^2)v}{2L(s)}f\right] = \frac{\partial}{\partial \mu}\left[D_{\mu\mu}(\mu)\frac{\partial f}{\partial \mu}\right],$$9, focusing shifts the stable fixed point of the deterministic drift and generates a nonzero drift velocity along the field line,

$L^{-1}(s)=-(1/B)\,dB/ds$0

where $L^{-1}(s)=-(1/B)\,dB/ds$1 is the stable fixed point of

$L^{-1}(s)=-(1/B)\,dB/ds$2

The SDE implementation is therefore a direct realization of a modified focused transport operator in which the focusing term appears simultaneously as drift in pitch-angle space, ballistic streaming in $L^{-1}(s)=-(1/B)\,dB/ds$3, and, if needed, a transport weight in the Monte Carlo reconstruction of the distribution function (Merten et al., 2024).

In galactic cosmic-ray modulation, a further modification enters through the focusing length itself. A recent comparison of Parker and focused transport equations defines

$L^{-1}(s)=-(1/B)\,dB/ds$4

which generalizes the usual magnetic focusing length by including flow-acceleration terms. Under otherwise identical diffusion conditions and without drifts, that study finds that the Parker transport equation overestimates the galactic cosmic-ray intensity at Earth’s orbit for low energies by $L^{-1}(s)=-(1/B)\,dB/ds$5, and by $L^{-1}(s)=-(1/B)\,dB/ds$6 over the poles, relative to the focused transport equation. The difference is traced to a small first-order anisotropy caused by particle fluxes over the poles: particles gain easier access to the inner heliosphere by streaming in over the poles, where pitch-angle scattering is generally weaker and the magnetic field is typically less wound. The same work also finds that the focused transport equation yields nearly identical results for different pitch-angle dependencies of the diffusion coefficients when the mean free paths are matched, implying that spectral and anisotropy data alone cannot distinguish between scattering theories with similar mean free paths but different pitch-angle dependencies (Berg et al., 8 Jun 2026).

Taken together, these developments establish the modified focused transport equation not as a single universal PDE but as a family of structurally adapted transport models. In energetic-particle transport, the central distinction is between the standard isotropizing form and the conservative modified form. In reduced descriptions, the modification may take the form of a telegraph correction, a finite-dimensional Legendre subspace, or an SDE realization with explicit focusing drift. In forward-peaked slab transport, the modification is a consistency correction that forces a Fokker–Planck surrogate to preserve the angular flux and retained moments of the linear Boltzmann equation. Across these settings, the common objective is the same: to alter a standard focused or forward-peaked transport equation so that it preserves a physically or numerically essential property that the unmodified approximation does not.