Magnus (Mathematics & Physics)

• Magnus is a term describing distinct technical concepts in mathematics and physics, ranging from Lie-series methods for solving differential equations to transverse forces observed in fluid and optical dynamics.
• It also denotes a group-theoretic property where normal closures determine conjugacy, providing rigorous insights in combinatorial and geometric group theory.
• Additionally, Magnus refers to a computational MHD code used for solar-atmosphere simulations, showcasing practical applications in numerical analysis and plasma dynamics.

Magnus denotes several distinct technical concepts across mathematics, theoretical physics, soft-matter and condensed-matter physics, numerical analysis, and computational plasma physics. In the literature represented here, the term covers: the Magnus expansion, introduced by Wilhelm Magnus in 1954 as an infinite Lie series for first-order homogeneous linear differential equations; the Magnus property in group theory, where equality of normal closures determines conjugacy up to inversion; and multiple transverse-response phenomena in physics, including the classical Magnus force, memory-induced Magnus forces at microscale, optical Magnus effects, Magnus Hall responses, and gyrotropic dynamics of skyrmions and related quasiparticles (Ebrahimi-Fard et al., 2023, Klopsch et al., 2016, Cao et al., 2023).

1. Classical, microscale, and memory-induced Magnus effects

In classical fluid mechanics, the Magnus force on a body that both translates with velocity $v$ and spins with angular velocity $\omega$ has the vector form

$$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$

where the force is perpendicular to both $v$ and $\omega$ (Cao et al., 2023). At low Reynolds number in a simple Newtonian liquid, Rubinow and Keller derived the transverse force on a spinning sphere,

$$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$

but for micron-sized spheres this force is extremely small (Cao et al., 2023). A direct microparticle experiment on magnetic Janus spheres of diameter $70$–$90~\mu\mathrm{m}$ reported trajectory deflections of the order of $1^\circ$ at Reynolds number around $1$, with measured values in agreement within measurement error with the low-Reynolds-number theory (Solsona et al., 2020).

A major extension is the memory-induced Magnus effect in viscoelastic fluids. There, the lateral force retains the same bilinear vector structure,

$\omega$0

but its origin is not inertial. Instead, viscoelastic memory produces a deformation field around the translating colloid; spinning rotates that deformation and generates a lateral recoil force. The reported coefficient satisfies

$\omega$1

and in the micellar fluid studied experimentally the fitted value was $\omega$2, corresponding to a more than million-fold enhancement over the Newtonian-Stokes estimate (Cao et al., 2023). In that regime the deflection angle can reach $\omega$3, and the response persists after spin is switched off, decaying over the fluid’s memory timescale (Cao et al., 2023).

Related low-Reynolds-number transport phenomena appear in driven ferromagnetic core–shell particles. In a viscous fluid, the translational force balance combines Stokes drag, the Rubinow–Keller Magnus lift, and a periodic external drive; synchronized translational and rotational oscillations then produce a nonzero drift over one forcing period (Denisov et al., 2016). The analysis predicts both unidirectional and bidirectional drift, with the latter enabling separation by core–shell ratio in dilute suspensions (Denisov et al., 2016). This extends the Magnus effect from a single-trajectory deflection problem to a controlled transport mechanism.

2. Optical and Berry-curvature Magnus responses

In wave optics and relativistic field propagation, the optical Magnus effect denotes a transverse shift of a light trajectory caused by polarization and appearing as a linear-in-wavelength correction to geometrical optics (Nishida, 16 Mar 2026). In the semiclassical wave-packet description derived from Maxwell’s equations in curved spacetime, the center obeys

$\omega$4

with helicity $\omega$5 and Berry curvature

$\omega$6

The anomalous velocity term is therefore helicity-dependent and orthogonal to both $\omega$7 and $\omega$8 (Nishida, 16 Mar 2026). In Schwarzschild spacetime this term does not change the photon-sphere radius or the critical impact parameter of the shadow, but it does bend trajectories out of any fixed plane; in weak gravitational lensing it yields a transverse shift $\omega$9 for a point-mass potential (Nishida, 16 Mar 2026).

The same Berry-curvature logic underlies the Magnus Hall effect and Magnus Nernst effect in inversion-broken but time-reversal-symmetric conductors with a built-in electric field (Das et al., 2021). The relevant semiclassical velocity is the Magnus velocity

$$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$0

distinct from the conventional anomalous velocity because it is proportional to the built-in field rather than the external bias (Das et al., 2021). In the ballistic regime, the Magnus Hall conductivity and Magnus Nernst conductivity are Fermi-surface responses weighted by the Berry curvature over forward-moving states, allowing a linear transverse response even when the ordinary anomalous Hall effect vanishes by time-reversal symmetry (Das et al., 2021).

Concrete realizations discussed in the literature include monolayer graphene with trigonal warping, strained bilayer graphene, topological-insulator surface states with hexagonal warping, and tilted multi-Weyl semimetals (Das et al., 2021). A notable result is that Magnus responses in multi-Weyl semimetals survive only for tilted Weyl nodes, so the effect can distinguish untilted from tilted nodes experimentally (Das et al., 2021). This usage of “Magnus” is therefore not hydrodynamic; it is a Berry-curvature-induced transverse transport response.

3. Gyrotropic Magnus terms in skyrmions, vortices, and Magnus-dominated particles

In magnetic skyrmion dynamics, the Magnus term is a nondissipative gyroscopic contribution perpendicular to the net force. In the particle model used for a skyrmion on an asymmetric quasi-one-dimensional substrate,

$$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$1

with $$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$2, the instantaneous velocity is

$$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$3

and the deflection angle obeys $$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$4 (Reichhardt et al., 2015). This Magnus term produces a new ratchet geometry: when the ac drive is perpendicular to the substrate asymmetry, the Magnus deflection converts the transverse drive into an $$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$5-directed velocity component, generating a Magnus-induced transverse ratchet that vanishes in the overdamped limit $$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$6 (Reichhardt et al., 2015). The resulting transport is quantized cycle by cycle, with $$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$7 and $$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$8, and the ratio $$\mathbf{F}_\mathrm{M} = f\,(\boldsymbol{\omega}\times\mathbf{v}),$$9 in the Magnus regime (Reichhardt et al., 2015).

The same gyrotropic mechanism can generate field-controlled nonreciprocal transport. In a skyrmion channel whose top and bottom walls are sawteeth of opposite asymmetry, reversing the out-of-plane magnetic field flips the topological charge $v$0, changes the sign of the gyrovector, and moves the skyrmion from the easy wall to the hard wall or vice versa (Souza et al., 20 Oct 2025). Under spin-transfer torque, the skyrmion velocity reverses and its magnitude changes; under spin–orbit torque, the velocity remains in the same direction but can drop to a much lower value, with negative differential conductivity in the hard-wall configuration (Souza et al., 20 Oct 2025). The paper interprets this as a magnetic-field-induced diode effect generated by the Magnus force (Souza et al., 20 Oct 2025).

Magnus-dominated kinematics also organizes interacting particle systems. In a minimal point-particle model,

$v$1

pairs of particles with the same Magnus coefficient form stable rotating dimers, while pairs with opposite Magnus force form translating dipoles whose speed scales as $v$2 (Reichhardt et al., 2020). Repulsive obstacles can trap these particles in localized orbits, and ac driving near an obstacle line produces commensuration ratchets and ratchet reversals with no overdamped analogue (Reichhardt et al., 2020). In 2D superfluids, an analogous “vortex–particle Magnus effect” arises when a massive particle trapped on a quantized vortex core feels the superflow induced by other vortices; the reduced equations of motion reproduce the cycloid-like trajectories seen in Gross–Pitaevskii simulations (Griffin et al., 2019).

4. The Magnus expansion and its modern extensions

The Magnus expansion is the representation of the solution of a linear non-autonomous system

$v$3

in the form

$v$4

where $v$5 is an infinite Lie series of time-ordered integrals of nested commutators (Ebrahimi-Fard et al., 2023). The first terms are

$v$6

$v$7

$v$8

and the differential form of the expansion is

$v$9

with Bernoulli numbers $\omega$0 (Ebrahimi-Fard et al., 2023). A standard sufficient condition for convergence is $\omega$1 (Ebrahimi-Fard et al., 2023).

Several recent works use this structure as a systematic effective-dynamics tool. In weakly coupled spin-$\omega$2 systems driven by shaped selective radiofrequency pulses, the interaction-picture Magnus solution exists whenever

$\omega$3

which for nonnegative-amplitude pulses reduces to the flip-angle condition $\omega$4 (Miao, 2012). In driven three-level systems, coarse-graining the exact evolution over a time window $\omega$5 and truncating the Magnus series yields ambiguity-free effective Hamiltonians that systematically improve upon adiabatic elimination (Macrì et al., 2022). In stiff time-varying stochastic systems,

$\omega$6

Magnus-based exponential integrators preserve statistical structures and fluctuation–dissipation balance while handling noncommuting $\omega$7 (Jasuja et al., 2022). The same strategy extends to stochastic delay-differential equations through Magnus–Euler–Maruyama and Magnus–Milstein schemes applied on Bellman intervals, with mean-square order $\omega$8 for MEM and $\omega$9 for MM (Griggs et al., 20 Jun 2025).

The expansion also appears in relativistic quantum field theory. There the $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$0-matrix is written as

$$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$1

with $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$2 Hermitian order by order, and the matrix elements of $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$3 are termed Magnus amplitudes (Brandhuber et al., 4 Dec 2025). At tree level they are built from retarded and advanced propagators weighted by Murua coefficients; at one loop they are determined by forward limits of tree amplitudes together with a Hadamard cut function (Brandhuber et al., 4 Dec 2025). This formulation is particularly useful because Magnus amplitudes are free of hyper-classical terms and are directly related to the radial action in classical two-body scattering (Brandhuber et al., 4 Dec 2025).

5. The Magnus property in combinatorial and geometric group theory

A group $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$4 has the Magnus property if equality of normal closures determines conjugacy up to inversion: $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$5 Here $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$6 is the normal closure of $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$7 in $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$8 (Klopsch et al., 2016). This notion is distinct from the Freiheitssatz, although Freiheitssatz arguments are frequently used to prove Magnus-type results (Feldkamp, 2017).

One line of work studies permanence under products. If $$\mathbf{F}_\mathrm{M}^{\text{(Newtonian, Stokes)}} = \pi\rho a^3 \,(\boldsymbol{\omega}\times\mathbf{v}),$$9 is an odd prime and $70$0 are residually finite-$70$1 groups with the Magnus property, then $70$2 has the Magnus property (Klopsch et al., 2016). The paper also exhibits finitely generated, torsion-free, residually finite groups $70$3, each with the Magnus property, such that $70$4 does not have the Magnus property, showing that the odd-prime residually finite-$70$5 hypothesis is substantive rather than cosmetic (Klopsch et al., 2016).

A second line studies amalgams and locally indicable groups. For groups of the form

$70$6

where $70$7 is a non-trivial reduced word in the letters $70$8, the Magnus property holds (Feldkamp, 2017). This encompasses the fundamental group of the closed non-orientable surface of genus $70$9, thereby completing the surface-group case left open by earlier work (Feldkamp, 2017). A broader Magnus extension theorem shows that for an indicable as well as locally indicable group $90~\mu\mathrm{m}$0, a nontrivial $90~\mu\mathrm{m}$1, and any group $90~\mu\mathrm{m}$2,

$90~\mu\mathrm{m}$3

has the Magnus property if and only if $90~\mu\mathrm{m}$4 has the Magnus property; when $90~\mu\mathrm{m}$5, the indicability assumption on $90~\mu\mathrm{m}$6 can be omitted (Feldkamp, 2020). This places the Magnus property within a systematic framework of free products, amalgamated products, and direct factors (Feldkamp, 2020).

In this mathematical sense, “Magnus” no longer refers to transverse physical response or exponential integrators; it refers to a rigidity property of normal closures in nonabelian groups.

6. MAGNUS as a computational MHD code

MAGNUS is also the name of a resistive MHD code for solar-atmosphere simulations. It is a finite-volume code on a uniform 3D Cartesian grid designed for wave propagation in the solar atmosphere under electrical resistivity and heat transference, with those non-ideal effects present but not dominant (Navarro et al., 2017). The code solves the resistive MHD equations with gravity, includes isotropic heat flux

$90~\mu\mathrm{m}$7

and field-aligned conduction

$90~\mu\mathrm{m}$8

and advances the induction equation with flux constrained transport so that $90~\mu\mathrm{m}$9 remains at machine round-off error (Navarro et al., 2017).

Its hyperbolic part is discretized by finite-volume HRSC methods with HLLE and HLLC approximate Riemann solvers and MINMOD, MC, or WENO5 reconstruction (Navarro et al., 2017). Time integration includes TVD Runge–Kutta schemes, and the paper demonstrates 1D and 2D ideal-MHD benchmarks, resistive reconnection, thermal conduction along magnetic field lines, a 3D vertical-velocity pulse in a photosphere–transition-region–corona configuration, and a 2D transverse pulse in a coronal loop (Navarro et al., 2017). In this context, “MAGNUS” is a code name rather than a Magnus effect, a Magnus expansion, or a Magnus property.

Across these usages, Magnus functions less as a single concept than as a family of technically unrelated but highly developed constructs: a Lie-series exponential method for noncommuting dynamics, a conjugacy criterion in group theory, a class of transverse-response phenomena in media and wave dynamics, a gyrotropic force term in topological quasiparticles, and a named computational framework for resistive MHD.