Optical Magnus Effect

• Optical Magnus effect is defined as a polarization-dependent transverse phenomenon arising from spin–orbit coupling in free space, focused-beam scattering, and curved spacetime.
• It emerges through varied mechanisms, including intrinsic transverse momentum currents under Maxwell theory, interference effects in atom–light interactions, and Berry-curvature corrections in gravitational lensing.
• The effect has practical implications in optical tweezers, trapped-ion experiments, and potentially cosmological observations, offering insights into spin-dependent forces and beam deflections.

Searching arXiv for recent and foundational papers on the Optical Magnus Effect to ground the article in the literature. The optical Magnus effect denotes a class of helicity- or polarization-dependent transverse optical phenomena in which spin-like degrees of freedom couple to propagation, spatial structure, or scattering. In current arXiv usage, the term covers several physically distinct regimes: a free-space polarization-dependent rotation of localized wavepackets driven by transverse momentum currents (Luo et al., 2010); an interaction-induced transverse deflection of a tightly focused beam scattered by a circular atomic dipole, with an equal and opposite force on the atom and an off-axis tweezer equilibrium up to $\lambda/(2\pi)$ (Spreeuw, 2021); and a Berry-curvature correction to photon trajectories in curved spacetime that modifies gravitational lensing at linear order in wavelength (Nishida, 16 Mar 2026). A trapped-ion experiment later directly mapped the focused-beam version at the level of the atom–light interaction profile (Leindecker et al., 30 Jan 2026).

1. Terminological scope and conceptual setting

The expression “optical Magnus effect” has not remained tied to a single mechanism. In older beam-optics usage it was associated with polarization-dependent beam-profile rotation and related Berry-phase phenomena. In the free-space formulation, it denotes a polarization-dependent rotation of a localized optical wavepacket caused by its own transverse-momentum currents, explicitly without light–matter interaction (Luo et al., 2010). In the focused-beam atomic formulation, it denotes a transverse deflection generated when a nonparaxial focused field excites a circular dipole whose spiral radiation interferes with the incident beam (Spreeuw, 2021). In curved spacetime, it denotes a helicity-dependent anomalous velocity term in semiclassical ray dynamics, transverse to both the local wavevector and the gradient of the effective optical medium defined by the metric (Nishida, 16 Mar 2026).

These usages are linked by spin–orbit coupling, but they are not interchangeable. The free-space version is formulated within Maxwell theory as a consequence of transverse spin and orbital momentum currents. The focused-beam atomic version is interaction-induced and mechanically consequential: the beam is deflected and the scatterer recoils. The gravitational version is a Berry-curvature correction to geometrical optics, linear in wavelength, and is usually discussed as a gravitational spin Hall effect of light. A recurring source of confusion is that all three are helicity-dependent and transverse, yet only one of them necessarily involves a localized scatterer, and only one of them is formulated directly as a correction to null geodesics.

Historically, the modern focused-beam discussion emerged from the broader development of optical orbital angular momentum after the recognition that vortex beams carry OAM, an insight that the review paper places in Leiden about 30 years before its publication (Spreeuw, 2021). That historical thread is important because the focused-beam Magnus analogy depends not on input OAM of the beam axis, but on nonparaxial spin-to-orbit conversion in the focal region.

2. Free-space optical Magnus effect from transverse-momentum currents

In the free-space treatment, the starting point is the vector Helmholtz equation

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$

together with an angular-spectrum representation and a Maxwell-consistent vector-field construction based on Whittaker scalar potentials (Luo et al., 2010). The electric field is decomposed as

$$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$

with the fixed unit vector

$$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$

This framework permits longitudinal components and separates different polarization geometries through the Jones parameters $\alpha,\beta$ and the helicity parameter

$\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$

for which $\sigma=\pm1$ for circular polarization.

The central dynamical quantity is the time-averaged linear momentum density

$$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$

with the decomposition

$$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$

where the spin and orbital parts are written as

$$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$

The optical Magnus effect in this setting is the polarization-dependent rotation of the centroid or, when the centroid itself does not move, the rotation of the transverse momentum currents.

For the full Whittaker model, the centroid is defined from $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$0 as

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$1

and the explicit centroid shifts are

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$2

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$3

The azimuthal rotation angle is

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$4

and the instantaneous angular velocity is

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$5

The rotation vanishes for $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$6, so linear polarization does not generate the same effect.

A key result is that the rotation is “unavoidable” when the wavepacket possesses transverse angular momentum (Luo et al., 2010). The integrated transverse angular momenta are related to centroid motion through

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$7

For the Whittaker model,

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$8

This formulation differs sharply from conventional medium-based optical Magnus descriptions because it requires no refractive-index gradient and no scatterer. The mechanism is internal to the vector field: transverse spin and orbital currents, including polarization-dependent screw wavefront structure, generate the rotation (Luo et al., 2010).

3. Focused-beam scattering by a circular dipole

In the tightly focused light–matter formulation, the optical Magnus effect is an interference-driven deflection of a focused beam by an atom that has been excited into a circular dipole. The physical picture is explicit: a tightly focused, linearly polarized beam excites an atom to a circular dipole, the spiral wavefront radiated by that dipole interferes with the incident focused field, and because the spiral wavefront is slightly tilted relative to the forward-propagating local wavefronts, optical power is redistributed across the beam cross-section. The beam therefore acquires a transverse deflection, and momentum conservation imposes an equal and opposite transverse force on the atom (Spreeuw, 2021).

The nonparaxial mechanism rests on two matching spiral phase structures. First, the co-rotated polarization field of an $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$9-polarized focused beam has, in the meridional plane,

$$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$0

so tight focusing generates helicity-dependent phase factors $$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$1, i.e. transverse OAM accompanying SAM. Second, radiation from a circular dipole $$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$2 rotating in the $$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$3-plane has

$$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$4

and in the dipole plane $$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$5 this becomes

$$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$6

The same spiral phase factor therefore appears in both the focused incident field and the circular-dipole radiation.

The scattered field for a coherent $$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$7 dipole is written as

$$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$8

with

$$\mathbf{E}(\mathbf{r})=\nabla\times\nabla\times(\mathbf{u}V_1)-ik\nabla\times(\mathbf{u}V_2),$$9

The total angular radiant intensity is

$$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$0

where

$$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$1

The pure dipole term obeys

$$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$2

so the transverse deflection is entirely carried by the interference term.

The average propagation-direction shift is

$$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$3

and the transverse reaction force is

$$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$4

For Gaussian and angular-tophat beams, the leading-order deflection is

$$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$5

and it is maximal at $$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$6. The effect vanishes in the plane-wave limit $$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$7, which expresses the dependence on nonparaxiality (Spreeuw, 2021).

A geometrically striking feature is the apparent source shift associated with the tilted spiral wavefront. Darwin’s observation, as discussed in the review, is that the dipole’s light appears to originate from a point displaced by

$$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$8

on a viewing-direction-dependent circle. In the tweezer problem this becomes a mechanical equilibrium condition. If the atom is laterally displaced by $$\mathbf{u}=\sin\theta\,\mathbf{e}_x+\cos\theta\,\mathbf{e}_z.$$9, the deflection acquires a factor $\alpha,\beta$0 to lowest order, and the transverse force vanishes at

$\alpha,\beta$1

Hence an atom in an optical tweezer is trapped at an off-axis equilibrium up to $\alpha,\beta$2, independent of detuning, beam divergence, trap frequency, and beam shape (Spreeuw, 2021).

This is the sense in which the effect is a true optical analogue of the classical Magnus effect. In the co-moving frame of a spinning ball, air is deflected by the rotating surface and the ball receives a transverse force. Here the “air stream” is the focused beam, the “spinning body” is the circular atomic dipole, and the recoil is an optical force on the atom.

4. Optical tweezers, trapped ions, and state-dependent forces

The tweezer context converts the interference mechanism into spin-dependent trapping geometry and control. For far-off-resonant operation, a level scheme such as $\alpha,\beta$3 allows spin selectivity to arise from selection rules rather than spectral isolation. Then $\alpha,\beta$4 couple to $\alpha,\beta$5 even at large detuning, and the equilibrium positions become

$\alpha,\beta$6

so the two spin states sit at $\alpha,\beta$7 for the $\alpha,\beta$8 quantization axis (Spreeuw, 2021). The review explicitly describes this as a Stern–Gerlach-like spin-dependent splitting in a single tweezer. Rotating the magnetic field in the $\alpha,\beta$9-plane moves the spin-dependent trap centers in anti-phase across the optical axis, enabling spin–motion control, symmetric splitting and merging, and microwave transitions between motional states. Driving the magnetic field at the trap frequency can induce large coherent motion with opposite phases for the two spins.

A trapped-ion proposal recast this focused-beam polarization-gradient physics as a quantum-gate resource. In that formulation, a tightly focused optical tweezer with $\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$0 produces circular components $\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$1 whose intensity maxima are laterally displaced by

$\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$2

so that a Zeeman qubit experiences qubit-state-dependent AC Stark minima (Mazzanti et al., 2023). The optical potential is modeled as

$\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$3

which expands near the center as

$\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$4

The spin-dependent force constant is therefore

$\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$5

and the force is transverse, perpendicular to the propagation direction. With amplitude modulation near the center-of-mass mode, the proposal obtains an effective Ising interaction

$\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$6

For two $\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$7 ions with $\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$8, $\sigma=i(\alpha\beta^\ast-\alpha^\ast\beta),$9, $\sigma=\pm1$0, $\sigma=\pm1$1, and $\sigma=\pm1$2, the proposal gives $\sigma=\pm1$3 and

$\sigma=\pm1$4

$\sigma=\pm1$5

It also states that pointing errors $\sigma=\pm1$6 reduce the gate fidelity from $\sigma=\pm1$7 to $\sigma=\pm1$8 (Mazzanti et al., 2023).

The first direct observation of the tightly focused optical Magnus effect at the atom–light interaction level used a single $\sigma=\pm1$9 ion on the $$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$0 quadrupole transition at $$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$1 (Leindecker et al., 30 Jan 2026). The experiment employed a custom $$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$2 objective, a focal size $$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$3, and a magnetic field $$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$4 along the $$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$5 axis. In the nonparaxial field model,

$$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$6

so the longitudinal component

$$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$7

produces intrinsic longitudinal-vortex structure and transverse polarization gradients beyond the paraxial approximation.

The quadrupole interaction is driven by the gradient tensor

$$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$8

with spherical-tensor components

$$\mathbf{p}^{[\mathrm{M}]}(\mathbf{r})=\frac{1}{2c^2}\mathrm{Re}\!\left[\mathbf{E}^{[\mathrm{M}]}(\mathbf{r})\times\mathbf{H}^{[\mathrm{M}]\ast}(\mathbf{r})\right],$$9

$$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$0

The experiment directly mapped $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$1 and found, for linear polarization, two-lobe separations $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$2 for $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$3 and $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$4 for $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$5, in agreement with the predicted $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$6 and $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$7, respectively. For right-hand circular polarization, the measured separations were $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$8 for $$\mathbf{p}^{[\mathrm{M}]}=\mathbf{p}^{[\mathrm{M}]}_O+\mathbf{p}^{[\mathrm{M}]}_S,$$9 and $$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$0 for $$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$1 (Leindecker et al., 30 Jan 2026). A phase-sensitive $$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$2–$$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$3 protocol further showed opposite phases of the two lobes for $$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$4, establishing that the observed spatial structure was a genuine polarization-gradient effect rather than a scalar intensity artifact.

The trapped-ion work is careful to distinguish this effect from interface beam shifts such as Imbert–Fedorov and Goos–Hänchen. No reflection or refractive interface is involved; the shift arises in free space through vectorial focusing and the resulting intrinsic $$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$5 and polarization gradients (Leindecker et al., 30 Jan 2026).

5. Berry-curvature formulation in curved spacetime and gravitational lensing

In curved spacetime, the optical Magnus effect is formulated as a semiclassical Berry-curvature correction to geometrical optics. Starting from Maxwell’s equations in a conformally flat metric class

$$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$6

with

$$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$7

one arrives at a wave-packet Lagrangian

$$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$8

where the Berry connection is

$$\mathbf{p}^{[\mathrm{M}]}_{S}=\mathrm{Im}[(\mathbf{E}^{[\mathrm{M}]}\cdot\nabla)\mathbf{E}^{[\mathrm{M}]\ast}],\qquad \mathbf{p}^{[\mathrm{M}]}_{O}=\mathrm{Im}[\mathbf{E}^{[\mathrm{M}]\ast}\cdot(\nabla)\mathbf{E}^{[\mathrm{M}]}].$$9

and the Berry curvature is

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$00

The equations of motion are

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$01

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$02

The first term is the geodesic limit; the second is the anomalous velocity, linear in wavelength and opposite for the two helicities (Nishida, 16 Mar 2026).

In the weak-potential FRW case,

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$03

the linearized equations become

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$04

with $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$05. Under the thin-lens approximation, the source–image mapping is modified to

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$06

where

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$07

The new term is transverse to the standard lensing deflection and produces a helicity-dependent twist of the lens map (Nishida, 16 Mar 2026).

For an axially symmetric thin lens,

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$08

A notable consequence is that a point source at $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$09 cannot produce an Einstein ring when $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$10, because the conditions $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$11 and $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$12 cannot be satisfied simultaneously for $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$13 (Nishida, 16 Mar 2026). For point-mass and singular isothermal sphere lenses, the paper gives analytic image radii, critical curves, caustics, and magnifications, with a finite caustic radius proportional to $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$14 replacing the usual point caustic.

In exact Schwarzschild spacetime, the same formalism yields

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$15

The radial dynamics are unchanged: the photon sphere remains at the standard Schwarzschild value and the shadow’s critical impact parameter is unchanged. The effect instead appears as nonplanarity and a helicity-dependent transverse drift whose sign flips with $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$16 (Nishida, 16 Mar 2026).

Despite these qualitative consequences, the magnitude is extremely small for astrophysical light. For optical or near-infrared photons with $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$17–$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$18 and cosmological $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$19, the paper estimates

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$20

so the induced transverse rotation scale $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$21 is negligible for current lensing observations (Nishida, 16 Mar 2026).

A cosmological extension shows that weak gravitational lensing plus the optical Magnus effect can induce circular polarization of the CMB from temperature anisotropies alone (Nishida, 16 May 2026). The modified lens equation is written as

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$22

with

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$23

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$24

If the unlensed CMB has no intrinsic circular polarization, the observed Stokes parameter obeys

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$25

Thus right- and left-handed components observed at the same sky position are sourced from different points on the last-scattering surface. The predicted signal is minuscule: at $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$26 and $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$27,

$(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$28

far below current detection capability (Nishida, 16 May 2026).

6. Distinctions, misconceptions, and open directions

A common misconception is that the optical Magnus effect is a single phenomenon. The literature summarized here instead shows a family resemblance: all variants involve polarization or helicity and a transverse response, but the underlying mechanisms differ. The free-space version is generated by transverse momentum currents of the electromagnetic field itself (Luo et al., 2010). The focused-beam atomic version is an interaction-induced interference effect and yields a real transverse force on a localized scatterer (Spreeuw, 2021). The curved-spacetime version is a Berry-curvature correction to ray transport and is best regarded as a helicity-dependent modification of geometrical optics (Nishida, 16 Mar 2026).

A second misconception is that the effect is synonymous with the spin Hall effect of light. The relation is close but not exact. The curved-spacetime and inhomogeneous-medium formulations are explicitly described as optical or gravitational spin Hall effects. The tightly focused atomic effect, however, is distinguished in the review from earlier Berry-phase-based optical Magnus or SHEL phenomena because it is interaction-induced, vanishes in the plane-wave limit, is maximized by strong focusing, and produces a direct mechanical force setting an off-axis equilibrium at $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$29 (Spreeuw, 2021). The trapped-ion measurements likewise emphasize that the observed displacement does not require any interface or reflection, unlike Imbert–Fedorov or Goos–Hänchen shifts (Leindecker et al., 30 Jan 2026).

A third misconception is that helicity reversal always corresponds simply to changing the input beam from $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$30 to $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$31. In the trapped-ion quadrupole measurements, simulations indicate that reversing $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$32 introduces only a global phase between $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$33 and $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$34 and does not change the gradients, whereas reversing the magnetic field flips the displacement direction along $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$35 (Leindecker et al., 30 Jan 2026). In the focused-beam dipole-scattering problem, by contrast, the sign flips with detuning and also flips if the dipole helicity is reversed simultaneously (Spreeuw, 2021). The relevant “spin” variable is therefore context-dependent: photon helicity, induced dipole helicity, or atomic quantization axis can play distinct roles.

Several open directions are identified explicitly in the recent literature. The focused-beam review proposes exploration of different level schemes with larger $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$36, as well as higher-multipole transitions such as magnetic dipole and electric quadrupole radiation; quadrupole radiation is noted to have tighter spiral phases $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$37, potentially doubling deflection and displacement (Spreeuw, 2021). Collective enhancement via subwavelength atomic arrays or dense micro-clouds is also suggested for the beam-deflection signature. The gravitational lensing work remarks that the same semiclassical formalism extends heuristically to gravitational waves with helicity $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$38, implying a doubled Berry curvature and anomalous velocity at leading order (Nishida, 16 Mar 2026). The CMB work emphasizes that Magnus-induced circular polarization is a fundamental geometric source of Stokes $(\nabla^2+k^2)\mathbf{E}(\mathbf{r},\omega)=0,$39, even though it is observationally negligible compared with foregrounds and nonlinear radiative mechanisms (Nishida, 16 May 2026).

Taken together, these results establish the optical Magnus effect not as a single canonical shift, but as a technically precise label for several helicity-dependent transverse transport phenomena whose common structure is spin–orbit or Berry-curvature coupling. In focused optical matter systems it can be a quantitatively useful force mechanism. In free-space beam theory it is a diagnostic of transverse spin and orbital momentum flow. In gravitation and cosmology it is a theoretically robust but extremely small correction to photon propagation.