betaFeedback

WorksPhysics and Astronomy

Partially isophase electro-magnetic configurations

Abstract

Separation of variables in Schrödinger equation allows an infinite-dimensional family of physically distinct field configurations that are dynamically indistinguishable in one selected sector, labeled by a conserved quantum number, while differing in all others. We call such a symmetry a partial isophase symmetry. Two main examples are considered: axially symmetric magnetic backgrounds (a rigid spectral flow of the Fock–Darwin oscillator, purely magnetic isophase families, and a two-channel extension) and a Cartesian slab geometry (magnetic barriers in two-dimensional electron gases, treated in the Appendices). We further discuss the consequences of this symmetry for the inverse spectral problem and consider its quasiclassical counterpart.

Introduction

The quantum dynamics of a charged particle in an external electromagnetic field is governed by gauge-invariant combinations of the scalar and vector potentials [1]. Standard gauge transformations leave the electromagnetic fields unchanged while modifying the wave-function phase. In this Letter, we identify a different type of invariance which we call a partial isophase symmetry. This transformation changes the electromagnetic field itself, yet preserves exactly the effective Hamiltonian, and therefore the full quantum evolution, in one selected sector singled out by the symmetries of the problem.

A partial isophase symmetry is a generic consequence of separation of variables which in turn is related with the symmetry algebra of the differential operators [2]. Whenever continuous symmetries commuting with the Hamiltonian decompose the wave function into sectors labeled by conserved quantum numbers, electromagnetic potentials can be varied leaving the corresponding 1D effective potential fixed in exactly one sector, at the price of shifting every other sector. This mechanism is worked out below first for the physically richest example, an axially symmetric background, where the two symmetry directions are a rotation and a translation; § 5 then abstracts it to an arbitrary separable geometry. Our findings followed the opposite direction. Having at hand a variety of non-trivial exactly solvable examples of both stationary and non-stationary one-dimensional Schrödinger equations, we were interested in more realistic 3D applications for a charged particle in an electromagnetic field. However, such an embedding lacks uniqueness. While analyzing the ambiguities that appear, the partial isophase symmetry was discovered. This symmetry is not restricted to exactly solvable models.

Consider the three-dimensional Schrödinger equation

(iℏ∂t−eϕ)Ψ(r,t)=12m(p−eA)2Ψ(r,t),(1)(i\hbar\partial_t-e\phi)\Psi(\boldsymbol{r},t)=\frac{1}{2m}(\boldsymbol{p}-e\boldsymbol{A})^2\Psi(\boldsymbol{r},t), \tag*{(1)}

in axially symmetric electromagnetic backgrounds that are translation invariant along the zz axis [3]. Such configurations are relevant, in particular, to the dynamics of electron states carrying orbital angular momentum, including vortex electrons, whose controlled generation and propagation has become an active topic in both fundamental and applied research [4].

For zz-independent fields, the potentials can be parametrized in cylindrical coordinates as

ϕ=ϕ(ρ),A=f(ρ)ρeφ+Az(ρ)ez,(2)\phi=\phi(\rho),\qquad\boldsymbol{A}=\frac{f(\rho)}{\rho}\boldsymbol{e}_{\varphi}+A_z(\rho)\boldsymbol{e}_z, \tag*{(2)}

so that

E=−∂ρϕ(ρ)eρ,B=−∂ρAz(ρ)eφ+f′(ρ)ρez,(3)\boldsymbol{E}=-\partial_\rho\phi(\rho)\boldsymbol{e}_\rho,\qquad\boldsymbol{B}=-\partial_\rho A_z(\rho)\boldsymbol{e}_\varphi+\frac{f'(\rho)}{\rho}\boldsymbol{e}_z, \tag*{(3)}

with 2πf(ρ)=Φ(ρ)2\pi f(\rho)=\Phi(\rho) magnetic flux.

Because the system is invariant under rotations around and translations along the zz axis, the operators L^z\hat{L}_z and p^z\hat{p}_z are conserved, and the wave function can be decomposed into sectors labeled by (l,kz)(l,k_z). The radial equation for u=ψl(ρ)/ρu=\psi_l(\rho)/\sqrt{\rho} takes the form

−ℏ22mu′′(ρ)+Vl,kz(ρ)u(ρ)=Eu(ρ),(4)-\frac{\hbar^2}{2m}u''(\rho)+V_{l,k_z}(\rho)u(\rho)=Eu(\rho), \tag*{(4)}

with

Vl,kz(ρ)=(ef(ρ)−ℏl)22mρ2−ℏ24ρ2+eϕ(ρ)+(ℏkz−eAz(ρ))22m.(5)V_{l,k_z}(\rho)=\frac{(ef(\rho)-\hbar l)^2}{2m\rho^2}-\frac{\hbar^2}{4\rho^2}+e\phi(\rho)+\frac{(\hbar k_z-eA_z(\rho))^2}{2m}. \tag*{(5)}

Fix now a partial wave ll and consider an arbitrary variation of f(ρ)f(\rho). The radial Hamiltonian (4) is invariant, for all kzk_z, under the transformation

f(ρ)⟶f~(ρ)=f(ρ)+δf(ρ),(6)f(\rho)\longrightarrow\tilde{f}(\rho)=f(\rho)+\delta f(\rho), \tag*{(6)}
ϕ(ρ)⟶ϕ~(ρ)=ϕ(ρ)−2δf(ρ)(ef(ρ)−lℏ)+eδf2(ρ)2mρ2,\phi(\rho)\longrightarrow\tilde{\phi}(\rho)=\phi(\rho)-\frac{2\delta f(\rho)(ef(\rho)-l\hbar)+e\delta f^2(\rho)}{2m\rho^2},

with arbitrary physically admissible δf(ρ)\delta f(\rho), while Az(ρ)A_z(\rho) remains unchanged. In contrast, the effective potentials in all other angular-momentum channels transform as

V~l′(ρ)=Vl′(ρ)+ℏ(l−l′)eδf(ρ)mρ2.(7)\tilde{V}_{l'}(\rho)=V_{l'}(\rho)+\frac{\hbar(l-l')e\delta f(\rho)}{m\rho^2}. \tag*{(7)}

We therefore obtain an infinite-dimensional family of physically distinct electromagnetic backgrounds

B~z=Bz+1ρδf′(8)\widetilde{B}_z = B_z + \frac{1}{\rho}\delta f' \tag*{(8)}
E~ρ=Eρ+(ef−ℏl)(ρδf′−2δf)mρ3+eδf(ρf′+ρδf′−δf)mρ3(9)\widetilde{E}_\rho= E_\rho+ \frac{(ef-\hbar l)\left(\rho\delta f'-2\delta f\right)}{m\rho^3} +\frac{e\delta f\left(\rho f'+\rho\delta f'-\delta f\right)}{m\rho^3} \tag*{(9)}

that are dynamically equivalent in one selected Lz=lL_z=l sector and inequivalent in all others.

We refer to this property as a partial isophase symmetry because not only the energy levels from the selected spectrum are invariant, but also any wave packet with a definite Lz=lL_z=l undergoes identical evolution in all possible electromagnetic configurations {(xref)}(8).

The remainder of this Letter develops this mechanism along the axially symmetric example above, then generalizes it. Since transformations (7) have a functional ambiguity, additional constraints lead to rich possibilities presented in § 1.1–§ 2. Variation of the homogeneous part of the magnetic field naturally produces a rigid spectral flow of the Fock–Darwin oscillator (and brings some information beyond the harmonic background). Fixing the electric field, we get a one-parameter family of magnetic fields. Choosing (7) to be a Darboux isospectral deformation, we extend our spectral control to two channels. Section V shows that the symmetry is a genuinely new obstruction for the inverse spectral problem of the magnetic Schrödinger operator [5], complementary to the classical Borg–Gel’fand–Levitan theory for a scalar potential; Sec. VI is devoted to the quasiclassical counterpart of the partial isophase symmetry. Section VII then abstracts the construction to an arbitrary separable geometry, and Appendix B works it out in full for a Cartesian slab geometry that maps directly onto magnetic-barrier and velocity-filter devices already realized in two-dimensional electron gases [6, 7], which suggests possible applications.

Uniform-field deformation and rigid spectral flow

For an initially homogeneous magnetic field,

f(ρ)=bρ22,Bz(0)=b,(10)f(\rho)=\frac{b\rho^2}{2}, \qquad B_z^{(0)}=b, \tag*{(10)}

choose

δf(ρ)=ρ2δb2.(11)\delta f(\rho)=\frac{\rho^2\delta b}{2}. \tag*{(11)}

The transformation changes the magnetic field by

δBz=δb,(12)\delta B_z=\delta b, \tag*{(12)}

and generates the compensating radial electric field

δEρ=eρ2m(bδb+δb22).(13)\delta E_\rho=\frac{e\rho}{2m}\left(b\delta b+\frac{\delta b^2}{2}\right). \tag*{(13)}

The protected Lz=lL_z=l radial Hamiltonian is unchanged, whereas the other partial waves are shifted by constants,

δVl′=eℏ(l−l′)2mδb,δEn,l′=δVl′,(14)\delta V_{l'}=\frac{e\hbar(l-l')}{2m}\delta b,\qquad\delta E_{n,l'}=\delta V_{l'}, \tag*{(14)}

and the radial eigenfunctions are unchanged. Applying global gauge transformation ϕ(ρ)→ϕ(ρ)−ℏl2mδb\phi(\rho)\to\phi(\rho)-\frac{\hbar l}{2m}\delta b we can always choose l=0l=0 as a protected sector without losing generality. Note that the initial radial component of the electric field can be arbitrary. That is, the initial spectrum En,l′E_{n,l'} and eigen-functions may not be analytically known. Nevertheless, the spectral flow (see Figure 2) is given by (14) and wave functions in all channels are invariant.

Figure 2. Spectral flow (14) for l=0l=0 (protected), 11, 22 versus ξ=−δb\xi=-\delta b, on top of a non-equidistant bare spectrum En,0(b0)E_{n,0}(b_0) (levels n=0,…,3n=0,\ldots,3). Every nn shifts linearly by δEn,l=eℏlξ/2m\delta E_{n,l}=e\hbar l\xi/2m, whatever the bare spectrum; the l=0l=0 levels stay flat.

When Eρ(0)(ρ)=λρE_\rho^{(0)}(\rho)=\lambda\rho, family (12),(13) generates a rigid spectral flow for the Fock–Darwin system [8–10]. Transformation (12),(13) gives a family of parabolas in the parameter space of the Fock–Darwin system (see Figure 1). In this case, we have an exactly solvable 2D2D harmonic oscillator with the spectrum En,l=(2n+∣l∣+1)ℏω−12ℏωcE_{n,l}=(2n+|l|+1)\hbar\omega-\frac{1}{2}\hbar\omega_c. The Larmor frequency ωc=eBz/m\omega_c=eB_z/m shifts the ground state energies of radial Hamiltonians E0,l(Bz)E_{0,l}(B_z) accompanying variations of BzB_z. While the energy quantum ℏω=ℏωc2/4−eλ/m\hbar\omega=\hbar\sqrt{\omega_c^2/4-e\lambda/m} stays fixed under the flow (12),(13)

Figure 1. Trajectories of the transformation (12),(13) in the (Bz,λ)(B_z,\lambda) plane, λ≡Eρ(0)/ρ\lambda\equiv E_\rho^{(0)}/\rho. Along each parabola λ=Bz2/4−Ω2\lambda= B_z^2/4-\Omega^2 the spectral quantum ℏΩ\hbar\Omega (color) is constant; the shaded region is Ω2<0\Omega^2<0. The arrow is one step b→b+δbb\to b+\delta b. For a non-linear background Eρ(0)(ρ)E_\rho^{(0)}(\rho), λ\lambda is the coefficient of its linear part, the only one changed by the flow.

e2(b+δb)24m2−em(λ+eδb4m(2b+δb))(15)\sqrt{\frac{e^2(b+\delta b)^2}{4m^2}-\frac{e}{m}\left(\lambda+\frac{e\delta b}{4m}(2b+\delta b)\right)} \tag*{(15)}
=e2b24m2−eλm(16)=\sqrt{\frac{e^2b^2}{4m^2}-\frac{e\lambda}{m}} \tag*{(16)}

Unmodified electric field

A particularly important case of symmetry appears in the case when we require δEρ(ρ)=0\delta E_\rho(\rho)=0. In this case

δϕ=2δf(ρ)(ef(ρ)−ℏl)+eδf2(ρ)2mρ2=C\delta\phi=\frac{2\delta f(\rho)\left(ef(\rho)-\hbar l\right)+e\delta f^2(\rho)}{2m\rho^2}=C

Thus we get [11]

eδf(ρ)=(ℏl−ef(ρ))±(ef(ρ)−ℏl)2+2meρ2C,e\delta f(\rho)=(\hbar l-ef(\rho))\pm\sqrt{(ef(\rho)-\hbar l)^2+2me\rho^2C},
B~z(ρ)=∣ef(ρ)−ℏl∣Bz(ρ)+2mC(ef(ρ)−ℏl)2+2meρ2C\widetilde{B}_z(\rho)=\frac{|ef(\rho)-\hbar l|B_z(\rho)+2mC}{\sqrt{(ef(\rho)-\hbar l)^2+2me\rho^2C}}

where the sign is chosen to provide a continuity of δf(ρ)\delta f(\rho) and B~z(ρ)\widetilde{B}_z(\rho).

Parameter CC, eC≥0eC\geq0, is responsible for a gauge symmetry (a constant shift of the scalar potential accompanied by a global phase transformation of the wave function exp⁡(−iCt)\exp(-iCt)). If the initial electric field E=0E=0, we get a family of magnetic fields with isophase quantum dynamics in the llth OAM channel.

Two-channel isospectrality

For Az(ρ)=0A_z(\rho)=0 (and hence Bϕ=0B_\phi=0), the partial-isophase construction can be extended to preserve one additional angular-momentum channel spectrally. Let ll denote the protected sector of the exact partial-isophase symmetry, and let ls≠ll_s\ne l be another fixed partial wave.

We may require deformation ΔV~ls(ρ)\Delta\widetilde V_{l_s}(\rho) to coincide with a standard isospectral Darboux transformation,

V~ls(ρ)−Vls(ρ)=−2ℏ22md2dρ2ln⁡W(ρ),(17)\widetilde V_{l_s}(\rho)-V_{l_s}(\rho)=-2\frac{\hbar^2}{2m}\frac{d^2}{d\rho^2}\ln W(\rho), \tag*{(17)}

where W(ρ)W(\rho) is chosen so that VlsV_{l_s} and V~ls\widetilde V_{l_s} are isospectral. This yields

δf(ρ)=ℏρ2e(ls−l)d2dρ2ln⁡W(ρ).(18)\delta f(\rho)=\frac{\hbar\rho^2}{e(l_s-l)}\frac{d^2}{d\rho^2}\ln W(\rho). \tag*{(18)}

Hence, the electromagnetic deformation generated by (18) preserves the ll-channel exactly and makes the lsl_s channel isospectral to its Darboux partner.

Explicit realizations are obtained by starting from an exactly solvable radial potential Vls(ρ)V_{l_s}(\rho). In the general Crum-Darboux construction,

W(ρ)=Wr⁡[u1(ρ,α1),…,uN(ρ,αN)],(19)W(\rho)=\operatorname{Wr}[u_1(\rho,\alpha_1),\ldots,u_N(\rho,\alpha_N)], \tag*{(19)}

where the auxiliary functions satisfy

−ℏ22muj′′(ρ,αj)+Vls(ρ)uj(ρ,αj)=αjuj(ρ,αj),(20)-\frac{\hbar^2}{2m}u_j''(\rho,\alpha_j)+V_{l_s}(\rho)u_j(\rho,\alpha_j)=\alpha_j u_j(\rho,\alpha_j), \tag*{(20)}

with factorization constants chosen below the ground-state energy, αj<E0(ls)\alpha_j<E_0^{(l_s)}, so that the resulting Wronskian is nonsingular. The corresponding deformation (18) then produces a family of electromagnetic backgrounds that is exactly invariant in the ll sector and spectrally equivalent in the lsl_s sector.

Although Eq. (20) is written for kz=0k_z=0, the conclusion extends to all longitudinal momenta. Indeed, for Az=0A_z=0 the kzk_z dependence enters the radial Hamiltonian only through an additive constant, so Vls,kzV_{l_s,k_z} and V~ls,kz\widetilde V_{l_s,k_z} remain isospectral for arbitrary kzk_z.

An explicit realization with two Crum seeds below the ground state of the harmonic oscillator is worked out in Appendix § 7.

Inverse spectral problem perspective

In standard literature on the inverse spectral problem for 3D Schrödinger equation, electric and magnetic fields are considered separately. In the standard setting H=−Δ+UH=-\Delta+U with separable variables, the single function UU enters all channels, so complete spectral data of one channel determine UU, and with it every other channel. “Complete” means eigenvalues together with norming constants (or two spectra) [12, 13, 14, 15]; for radial problems with a centrifugal term see [16, 17]. Eigenvalues alone are not sufficient: the potentials with a prescribed spectrum form the isospectral (Crum-Darboux) manifold parametrized by the norming constants [15, 18, 19].

Inverse results for the magnetic Schrödinger operator are usually formulated for boundary data [20] [5], which in a separable geometry contain all angular-momentum channels at once. If one considers “spectroscopic” information for magnetic backgrounds, ϕ=const\phi=\mathrm{const}, Eq. (5) gives (ef−ℏl)2(ef-\hbar l)^2 in terms of VlV_l, so one channel fixes ff up to the choice of the branch of the square root and an additive constant (the parameter CC of Sec. III).

For the electromagnetic field the counting is different. A sector (l,kz)(l,k_z) with complete data yields one function of ρ\rho,

Vl,kz=(ef−ℏl)2−ℏ2/42mρ2+eϕ+(ℏkz−eAz)22m,(21)V_{l,k_z}=\frac{(ef-\hbar l)^2-\hbar^2/4}{2m\rho^2}+e\phi+\frac{(\hbar k_z-eA_z)^2}{2m}, \tag*{(21)}

whereas the fields are described by three, ff, ϕ\phi and AzA_z. The partial isophase transformation (7) is precisely the description of the resulting ambiguity: for fixed ll (and all kzk_z) the data determine AzA_z and the combination VlV_l, but the field is fixed only up to an arbitrary δf(ρ)\delta f(\rho). Additional sectors remove it algebraically. Two channels l≠l′l\ne l' give

ef=ℏ(l+l′)2+mρ2(Vl−Vl′)ℏ(l′−l),ef=\frac{\hbar(l+l')}{2}+\frac{m\rho^2(V_l-V_{l'})}{\hbar(l'-l)},
eAz=ℏkz2−m(Vl,kz−Vl,0)ℏkz,(22)eA_z=\frac{\hbar k_z}{2}-\frac{m(V_{l,k_z}-V_{l,0})}{\hbar k_z}, \tag*{(22)}

and eϕe\phi then follows from VlV_l. We therefore conclude that three sectors, e.g. (l,0)(l,0), (l,kz≠0)(l,k_z\ne0), (l′,0)(l',0), are sufficient in principle, provided the potential of each channel can be reconstructed from its data; the exponent ν=∣l−f(0)e/ℏ∣\nu=|l-f(0)e/\hbar| of the centrifugal term is encoded in the asymptotics of the levels and norming constants. Equation (22) also explains Eq. (7): the deformation of channel l′l' is proportional to l−l′l-l'. A single channel is not sufficient even with complete data, and the corresponding non-uniqueness is infinite-dimensional. We do not address the stability of the reconstruction, the sufficiency of finitely many levels, or whether isospectrality can be imposed simultaneously in more than one unprotected channel.

Quasi-classical limit

In the classical limit, ℏ→0\hbar\to0, l→∞l\to\infty with L=ℏlL=\hbar l fixed, the canonical angular momentum LL is conserved and the radial motion is governed by

HL=pρ22m+VL(ρ),VL(ρ)=(L−ef)22mρ2+eϕ,(23)H_L=\frac{p_\rho^2}{2m}+V_L(\rho),\qquad V_L(\rho)=\frac{(L-ef)^2}{2m\rho^2}+e\phi, \tag*{(23)}

the classical counterpart of Eq. (5) (the zz motion is unaffected, since AzA_z is not changed). The fields transform as in the quantum case, Eq. (7) with ℏl→L0\hbar l\to L_0. Then VL0V_{L_0} is invariant, so the whole radial dynamics (ρ(t),pρ(t))(\rho(t),p_\rho(t)) at L=L0L=L_0 is unchanged, and for L0L_0 the orbits fill the same annulus ρmin⁡≤ρ≤ρmax⁡\rho_{\min}\le\rho\le\rho_{\max} determined by E=VL0(ρ)E=V_{L_0}(\rho). Comparing the Hamiltonians at the same (ρ,pρ,L)(\rho,p_\rho,L), the δf2\delta f^2 terms cancel and

H~−H=−Ωδ(ρ)(L−L0),Ωδ(ρ)=e δf(ρ)mρ2.(24)\widetilde{H}-H=-\Omega_\delta(\rho)(L-L_0),\qquad\Omega_\delta(\rho)=\frac{e\,\delta f(\rho)}{m\rho^2}. \tag*{(24)}

The same identity holds for the quantum Hamiltonian, H~−H=−Ωδ(ρ)(L^z−ℏl)\widetilde{H}-H=-\Omega_\delta(\rho)(\hat{L}_z-\hbar l), because Ωδ(ρ)\Omega_\delta(\rho) commutes with L^z\hat{L}_z; it reproduces Eq. (7). The transformation thus acts as a differential rotation with angular velocity Ωδ(ρ)\Omega_\delta(\rho), which vanishes identically in the protected sector. The real-space orbit, however, is not invariant. Since mρ2ϕ˙=L−efm\rho^2\dot{\phi}=L-ef, Eq. (24) gives ϕ~˙=ϕ˙−Ωδ(ρ)\dot{\widetilde{\phi}}=\dot{\phi}-\Omega_\delta(\rho), and because ρ(t)\rho(t) is the same,

ϕ~(t)=ϕ(t)−∫0tΩδ(ρ(t′)) dt′.(25)\widetilde{\phi}(t)=\phi(t)-\int_0^t\Omega_\delta(\rho(t'))\,dt'. \tag*{(25)}

The velocity, the kinetic energy and the Lorentz force along the orbit change; only the combination entering the radial equation, m(ρ¨−ρϕ˙2)=e(Eρ+ρϕ˙Bz)m(\ddot{\rho}-\rho\dot{\phi}^2)=e(E_\rho+\rho\dot{\phi}B_z), is invariant, although each of its terms is modified. The partial isophase property is therefore a property of the reduced radial dynamics rather than of the trajectory in the plane. It is not restricted to circular orbits: a circular orbit keeps its radius (VL′=0V_L'=0) but is traversed with a different angular velocity.

For the Fock–Darwin system, f=bρ2/2f=b\rho^2/2, Eρ(0)=λρE_\rho^{(0)}=\lambda\rho, the orbit is a superposition of two counter-rotating circular modes with frequencies ω±=ω±ωc/2\omega_\pm=\omega\pm\omega_c/2, and ρ(t)\rho(t) oscillates with frequency 2ω2\omega. The orbit is closed if and only if ω+/ω−\omega_+/\omega_- is rational. For the uniform-field deformation, Ωδ=e δb/2m\Omega_\delta=e\,\delta b/2m is constant: Eq. (25) is a rigid rotation, ω\omega (and ρ(t)\rho(t)) is fixed, which is the classical counterpart of the invariance of ℏω\hbar\omega in Eqs. (12),(13), while ω+→ω+±Ωδ\omega_+\to\omega_+\pm\Omega_\delta. The shape of a single loop is preserved, but the closure condition changes (Figure 3, top). For a generic δf\delta f (Figure 3, bottom) the rotation is differential: the azimuth is shifted in steps, mainly where δf\delta f is non-negligible, the orbit in the plane is distorted, and the radial oscillations are not isochronous; nevertheless ρ(t)\rho(t) coincides exactly.

Figure 3. Classical orbits with L=L0L=L_0 before and after the transformation (e=m=1e=m=1), obtained by integrating Newton’s equations with the Lorentz force; the dot marks the initial point, dotted circles are the turning radii. Top: Fock–Darwin system, b=1b=1, λ=−0.75\lambda=-0.75 (ω+/ω−=3\omega_+/\omega_-=3, closed four-lobe orbit), δf=ρ2δb/2\delta f=\rho^2\delta b/2, δb=−1/3\delta b=-1/3 (ω+/ω−=2\omega_+/\omega_-=2, closed three-lobe orbit). Bottom: Bz=0.6+1.4e−ρ2/1.32B_z=0.6+1.4e^{-\rho^2/1.3^2}, E(0)=0E^{(0)}=0, δf=0.5ρ2e−ρ2/2\delta f=0.5\rho^2e^{-\rho^2/2}. Right: ρ(t)\rho(t) is identical before and after (to 10−910^{-9}), while φafter−φbefore=−∫Ωδ(ρ(t′)) dt′\varphi_{\mathrm{after}}-\varphi_{\mathrm{before}}=-\int\Omega_\delta(\rho(t'))\,dt' (dotted) grows linearly (rigid rotation) in the quadratic case and by steps (differential rotation) otherwise.

Thus the classical dynamics within a family of partially isophase backgrounds is consistent with the quantum picture. The density of a wave packet with a sharp ll, built from radial states, forms a ring of pulsating radius, and its evolution is the same before and after the transformation. All observables that are functions of (ρ,pρ,L^z)(\rho,p_\rho,\hat{L}_z) coincide. Note that gauge-invariant kinetic quantities transform as follows: the current jϕ=(ℏl−ef)∣ψ∣2/mρj_\phi=(\hbar l-ef)|\psi|^2/m\rho, the kinetic energy and the orbital magnetic moment change, being the quantum counterparts of the shift of ϕ˙\dot{\phi}. In the protected channel the configurations are therefore indistinguishable by the spectrum and the density, but distinguishable by currents and magnetization. For a packet with a spread Δl\Delta l the deviation from identical evolution is controlled by Eq. (24) and grows with Δl\Delta l. Finally, the Bohr–Sommerfeld condition ∮pρ dρ=2πℏ(n+γ)\oint p_\rho\,d\rho=2\pi\hbar(n+\gamma) depends only on VlV_l and gives the same levels in the protected channel.

General scheme: separable coordinates and sector protection

The construction of § 1–§ 2 rests on a structural feature of the cylindrical problem that is worth isolating, since it can be extended to other separable geometries.

Suppose the electromagnetic background is invariant under a set of commuting continuous symmetries (translations and/or rotations) labeled by a=1,…,ka=1,\ldots,k, so that after separation of variables in the remaining coordinate ξ\xi the wave function carries good quantum numbers nan_a (linear and/or angular momenta), and the vector potential admits, in a suitable gauge, one component along each symmetry direction. It is convenient to write the minimal-coupling combination along direction aa as ℏna−efa(ξ)\hbar n_a-ef_a(\xi), where fa(ξ)=Aa(ξ)f_a(\xi)=A_a(\xi) for a translational direction and fa(ξ)=ρAφ(ξ)f_a(\xi)=\rho A_\varphi(\xi) (the flux function) for rotation about the ξ\xi-axis. The effective one-dimensional

potential in the sector {na}\{n_a\} then takes the general form

V{na}(ξ)=∑aca(ξ)2m(ℏna−efa(ξ))2+eϕ(ξ)+Q(ξ),(26)V_{\{n_a\}}(\xi)=\sum_a\frac{c_a(\xi)}{2m}\left(\hbar n_a-ef_a(\xi)\right)^2+e\phi(\xi)+Q(\xi), \tag*{(26)}

where ca(ξ)c_a(\xi) is a direction-dependent weight (ca=1/ρ2c_a=1/\rho^2 for rotation about ξ=ρ\xi=\rho, ca=1c_a=1 for translation), and Q(ξ)Q(\xi) is a measure-induced quantum correction that is present only for curved (rotational) directions – cf. the −ℏ2/(8mρ2)-\hbar^2/(8m\rho^2) term hidden in Eq. (5) – and vanishes identically for flat (translational) ones. Equation (5) is recovered for k=2k=2, with a=1a=1 the rotational direction (c1=1/ρ2c_1=1/\rho^2, f1=ff_1=f) and a=2a=2 the translational direction along zz (c2=1c_2=1, f2=Azf_2=A_z, absorbed with a minus sign into the definition used in § 1).

Fix one direction a0a_0 and one value na0=na00n_{a_0}=n_{a_0}^0 to be protected. Deform only the corresponding flux function, fa0→f~a0=fa0+δfa0f_{a_0}\to\tilde f_{a_0}=f_{a_0}+\delta f_{a_0}, leaving every faf_a with a≠a0a\ne a_0 untouched. Demanding that V{na}(ξ)V_{\{n_a\}}(\xi) with na0=na00n_{a_0}=n_{a_0}^0 be left invariant as a function of ξ\xi, for arbitrary values of the remaining quantum numbers, fixes the compensating scalar potential

δϕ(ξ)=−ca0(ξ)2δfa0(ξ)(efa0(ξ)−ℏna00)+eδfa02(ξ)2m(27)\delta\phi(\xi)=-c_{a_0}(\xi)\frac{2\delta f_{a_0}(\xi)(e f_{a_0}(\xi)-\hbar n^0_{a_0})+e\delta f_{a_0}^{2}(\xi)}{2m} \tag*{(27)}

which reduces to Eq. (7) for ca0=1/ρ2c_{a_0}=1/\rho^2, na00=ln^0_{a_0}=l. Every other value na0′≠na00n'_{a_0}\ne n^0_{a_0} along the same direction (with all other nan_a unchanged) then shifts by

ΔVna0′(ξ)=ca0(ξ)eℏ(na00−na0′)mδfa0(ξ)(28)\Delta V_{n'_{a_0}}(\xi)=c_{a_0}(\xi)\frac{e\hbar(n^0_{a_0}-n'_{a_0})}{m}\delta f_{a_0}(\xi) \tag*{(28)}

which reduces to Eq. (7). Both reductions follow by direct substitution of Eq. (27) into Eq. (26) and are exact, not merely first-order in δfa0\delta f_{a_0}.

Two structural remarks follow immediately from Eqs. (26)–(28). First, the construction of § 1 can equally well be applied to Az(ρ)A_z(\rho) rather than to f(ρ)f(\rho), protecting a chosen kz0k_z^0 for all ll instead of a chosen ll for all kzk_z; since δϕ(ξ)\delta\phi(\xi) is a single function, only one such direction can be protected in full, and protecting a single point (l0,kz0)(l_0,k_z^0) in the space of quantum numbers by splitting the freedom between δf\delta f and δAz\delta A_z is possible only at the cost of the “for all” quantifier. Second, whenever k≥2k\ge2 independent symmetry directions are available, the same trade-off recurs: the single scalar potential can protect one direction completely, or several directions only at isolated points of their joint spectrum. Appendix B illustrates the k=2k=2 case explicitly, with both directions translational. The Cartesian slab geometry is presented in Appendix B.

Conclusion

We have identified a family of electromagnetic deformations, a partial isophase symmetry, that leave the quantum dynamics in one selected separable sector exactly unchanged while modifying the physical field itself, complementary to ordinary gauge invariance. Its general mechanism (§ 5), its axially symmetric realizations (§ 1–§ 2) and its Cartesian-slab realization (Appendix § 8) were worked out above. The two-channel isospectrality presented in § 2 seems very interesting for further mathematical studies, since usually Darboux transformations of an effective radial Hamiltonian destroy analytical structure in other sectors (except for highly symmetric shape-invariant potentials). Ambiguities in spectrum-to-field inversion clarify a set of sufficient spectral data for the inverse 3D spectral problem in a mixed electromagnetic background (§ 3). In the classical limit (§ 4) we found an interesting interplay between the angular and radial dynamics of individual trajectories. Our preliminary results also show that a similar symmetry manifests itself for non-stationary electromagnetic fields directly in the time-dependent Schrödinger equation, and in the Schrödinger–Poisson equation, where the wave packet describes a

Acknowledgments

The author is grateful to Professor P. O. Kazinsky for valuable comments and suggestions. This work was presented at the seminar of the Department of Mathematics (PPGMPA), Universidade Federal de Santa Catarina. The author also thanks Ivan Pontual Costa e Silva and Vladimir Pestov for organizing this seminar.

Explicit example of the two-channel isospectrality

We take ls=0l_s=0, l=1l=1 and a seed background consisting of a uniform field Bz=ωB_z=\omega (in units of h/eh/e) and an Aharonov–Bohm flux μ=1/2\mu=1/2, i.e. F(ρ)≡ef/ℏ=μ+ωρ2/2F(\rho)\equiv ef/\hbar=\mu+\omega\rho^2/2, ϕ=0\phi=0; we use units ℏ2/2m=1\hbar^2/2m=1. Since ∣l−μ∣=1/2|l-\mu|=1/2 in both channels, the centrifugal term in Eq. (5) vanishes, V0=ω/2+ω2ρ2/4V_0=\omega/2+\omega^2\rho^2/4, V1=−ω/2+ω2ρ2/4V_1=-\omega/2+\omega^2\rho^2/4, and both channels are reduced to the oscillator on the half-line with u(0)=0u(0)=0, En(0)=2ω(n+1)E_n^{(0)}=2\omega(n+1), En(1)=ω(2n+1)E_n^{(1)}=\omega(2n+1); the AB flux is what makes the origin regular for l=0l=0 and l=1l=1 simultaneously and allows oscillator seed functions to be used. The two Crum seeds of the ls=0l_s=0 channel, at energies E2<E1<E0(ls)=2ωE_2<E_1<E_0^{(l_s)}=2\omega, are u1=ρe−ωρ2/41F1(1−E12ω;32;ωρ22)u_1=\rho e^{-\omega\rho^2/4}{}_1F_1\left(1-\frac{E_1}{2\omega};\frac{3}{2};\frac{\omega\rho^2}{2}\right), regular at the origin, and u2=e−ωρ2/4U(12−E22ω;12;ωρ22)u_2=e^{-\omega\rho^2/4}U\left(\frac{1}{2}-\frac{E_2}{2\omega};\frac{1}{2};\frac{\omega\rho^2}{2}\right), decaying at infinity; we use ω=1\omega=1, E1=ω/2E_1=\omega/2, E2=−3ω/2E_2=-3\omega/2. Both seeds are nodeless, W(0)<0W(0)<0, W′=(E1−E2)u1u2>0W'= (E_1-E_2)u_1u_2>0 and W→0−W\to0^-, so W=Wr⁡[u1,u2]<0W=\operatorname{Wr}[u_1,u_2]<0 for all ρ>0\rho>0. Equation (18) then gives δF=−ρ2(ln⁡W)′′\delta F=-\rho^2(\ln W)'' (with δF=eδf/ℏ\delta F=e\delta f/\hbar), and the scalar potential follows from Eq. (7), eϕ=−[2δF(F−l)+δF2]/ρ2e\phi=-[2\delta F(F-l)+\delta F^2]/\rho^2.

The AB flux is unchanged, δF(0)=0\delta F(0)=0, so only the regular parts of the fields are modified (Figure 4). The deformation is finite and localized: δF→−2\delta F\to-2 at large ρ\rho (a net flux of −2-2 flux quanta), B~z→ω\widetilde B_z\to\omega, and eϕ~→2ωe\widetilde\phi\to2\omega, a constant that keeps the ls=0l_s=0 channel exactly isospectral; at the origin the fields are finite but contain a term linear in ρ\rho (B~z(0)≈5\widetilde B_z(0)\approx5, eE~ρ(0+)≈5.7e\widetilde E_\rho(0+)\approx5.7). By Eq. (7) all potential deformations are given by one universal function, ΔVl/(l−l′)=2δF/ρ2=−2(ln⁡W)′′\Delta V_l/(l-l')=2\delta F/\rho^2=-2(\ln W)'' [Figure 4(c)]. Numerical diagonalization of −u′′+Vu=Eu-u''+V u=E u (Figure 5) confirms that the l=1l=1 potential, spectrum and eigenfunctions are unchanged; the ls=0l_s=0 channel keeps all levels 2ω(n+1)2\omega(n+1) while its eigenfunctions u~n=Wr⁡[u1,u2,un]/Wr⁡[u1,u2]\widetilde u_n=\operatorname{Wr}[u_1,u_2,u_n]/\operatorname{Wr}[u_1,u_2] are visibly deformed; each unprotected level of l′=2l'=2, degenerate with l=1l=1 before the deformation, is shifted, δEn≃0.18, 0.11, 0.08, 0.07, 0.05,…\delta E_n\simeq0.18,\,0.11,\,0.08,\,0.07,\,0.05,\ldots (Figure 6).

Figure 4. Regular parts of the fields before and after the two-channel deformation (ls=0l_s = 0, l=1l = 1, μ=1/2\mu= 1/2, ω=1\omega= 1, E1=ω/2E_1 = \omega/2, E2=−3ω/2E_2 = -3\omega/2; ℏ2/2m=1\hbar^2/2m = 1): (a) BzB_z, (b) EρE_\rho, (c) universal potential deformation ΔVl′/(l−l′)\Delta V_{l'}/(l-l'). The AB flux μ=1/2\mu= 1/2 is unchanged.

Figure 5. Potentials, levels and eigenfunctions En+un(ρ)E_n + u_n(\rho) before (dashed) and after (solid) the deformation for the isospectral channel ls=0l_s = 0, the protected channel l=1l = 1 (unchanged) and the unprotected channel l=2l = 2.

Figure 6. (a) Lowest levels of the three channels before (dashed) and after (solid); (b) shifts δEn\delta E_n: zero for ls=0l_s = 0 and l=1l = 1, positive and decreasing with nn for l=2l = 2.

Cartesian slab geometry

Setup

As a first application of § 5 beyond cylindrical symmetry, consider a background invariant under translations along yy and zz, with all nontrivial dependence carried by a single Cartesian coordinate xx (a “slab”). In the gauge Ax=0A_x=0,

ϕ=ϕ(x),A=Ay(x)ey+Az(x)ez,(29)\phi=\phi(x),\qquad\mathbf{A}=A_y(x)\mathbf{e}_y+A_z(x)\mathbf{e}_z, \tag*{(29)}

so that

B=−Az′(x)ey+Ay′(x)ez,E=−ϕ′(x)ex.(30)\mathbf{B}=-A_z'(x)\mathbf{e}_y+A_y'(x)\mathbf{e}_z,\qquad\mathbf{E}=-\phi'(x)\mathbf{e}_x. \tag*{(30)}

Here both symmetry directions are translational, cy=cz=1c_y=c_z=1 and Q=0Q=0 identically: unlike Eq. (5), there is no measure-induced correction and no coordinate singularity at x=0x=0.

Reduction and sector protection

With py,pzp_y,p_z conserved and Ψ=ei(kyy+kzz)χ(x)e−iEt/ℏ\Psi=e^{i(k_y y+k_z z)}\chi(x)e^{-iEt/\hbar}, Eq. (1) reduces to

−ℏ22mχ′′(x)+Vky,kz(x)χ(x)=Eχ(x),(31)-\frac{\hbar^2}{2m}\chi''(x)+V_{k_y,k_z}(x)\chi(x)=E\chi(x), \tag*{(31)}
Vky,kz(x)=(ℏky−eAy(x))22m+(ℏkz−eAz(x))22m+eϕ(x),(32)V_{k_y,k_z}(x)=\frac{(\hbar k_y-eA_y(x))^2}{2m}+\frac{(\hbar k_z-eA_z(x))^2}{2m}+e\phi(x), \tag*{(32)}

the k=2k=2, translational instance of Eq. (26). Applying Eqs. (27)–(28) with a0=ya_0=y, ca0=1c_{a_0}=1, na0=ky0n^{a_0}=k_y^0 gives the protecting deformation

δϕ(x)=2δAy(x)(ℏky0−eAy(x))−eδAy2(x)2m,(33)\delta\phi(x)=\frac{2\delta A_y(x)(\hbar k_y^0-eA_y(x))-e\delta A_y^2(x)}{2m}, \tag*{(33)}
ΔVky′(x)=eℏ(ky0−ky′)mδAy(x),(34)\Delta V_{k_y'}(x)=\frac{e\hbar(k_y^0-k_y')}{m}\delta A_y(x), \tag*{(34)}

while Az(x)A_z(x) is left untouched (it can be protected in exactly the same way instead, protecting a chosen kz0k_z^0 for all kyk_y; see the remark at the end of § 5). The magnetic field changes by

B~z(x)=Bz(x)+δAy′(x).(35)\widetilde{B}_z(x)=B_z(x)+\delta A_y'(x). \tag*{(35)}

Rigid spectral flow of a magnetically confined channel

Take the Landau gauge Ay(x)=bxA_y(x)=bx (Bz(0)=bB_z^{(0)}=b) together with a background field Ex(0)(x)=λxE_x^{(0)}(x)=\lambda x, i.e. ϕ(x)=−λx2/2\phi(x)=-\lambda x^2/2; this is the standard model of a quantum wire in a perpendicular magnetic field with parabolic electrostatic confinement. Protecting ky0=0k_y^0=0 (any other value can always be brought to 0 by the same global gauge shift used in § 1.1), Eq. (33) with δAy(x)=x δb\delta A_y(x)=x\,\delta b gives δϕ(x)=−ex22m[(b+δb)2−b2]\delta\phi(x)=-\frac{e x^2}{2m}\left[(b+\delta b)^2-b^2\right]. Adding this to the deformed kinetic term of Eq. (32),

V~0,kz(x)=e2(b+δb)2x22m−eλx22−e2x22m[(b+δb)2−b2]+(ℏkz−eAz)22m=V0,kz(x),(36)\widetilde{V}_{0,k_z}(x)=\frac{e^2(b+\delta b)^2x^2}{2m}-\frac{e\lambda x^2}{2}-\frac{e^2x^2}{2m}\left[(b+\delta b)^2-b^2\right]+\frac{(\hbar k_z-eA_z)^2}{2m}=V_{0,k_z}(x), \tag*{(36)}

the quadratic-in-xx deformation cancels identically, confirming Eq. (27) by direct computation. The protected channel is a harmonic oscillator of frequency Ω2=ωc2+eλ/m\Omega^2=\omega_c^2+e\lambda/m, ωc=eb/m\omega_c=eb/m, exactly as in Eq. (16). For ky0≠0k_y^0\ne0, Eq. (34) gives ΔVky′(x)=−eℏky0mδb x\Delta V_{k_y'}(x)=-\frac{e\hbar k_y^0}{m}\delta b\,x: a term linear in xx, which only translates the center of the parabola without changing its curvature. Hence Ω\Omega is common to every channel, before and after the deformation – a rigid spectral flow of the same kind as in § 1.1, but for a single, non-degenerate transverse mode rather than a full Landau level.

Unmodified electric field

Requiring δEx(x)=0\delta E_x(x)=0 in Eq. (33) forces δϕ=C=const.\delta\phi=C=\text{const.} Writing h(x)≡ℏky0−eAy(x)h(x)\equiv\hbar k_y^0-eA_y(x) (so h′=−eBzh'=-eB_z), the continuous-at-C=0C=0 solution is

e δAy(x)=h(x)−h(x)2−2meC,(37)e\,\delta A_y(x)=h(x)-\sqrt{h(x)^2-2meC}, \tag*{(37)}

and the deformed field follows in closed form,

B~z(x)=∣h(x)∣Bz(x)h(x)2−2meC,(38)\widetilde{B}_z(x)=\frac{\lvert h(x)\rvert B_z(x)}{\sqrt{h(x)^2-2meC}}, \tag*{(38)}

structurally identical to the cylindrical result of Sec. III, but without the metric factors of ρ\rho.

Physical realization: magnetic barriers in a two-dimensional electron gas

The geometry of Eqs. (29)–(32) is typical for magnetic barriers in a two-dimensional electron gas (2DEG), realized by depositing a ferromagnetic or current-carrying stripe on a semiconductor heterostructure so as to produce a perpendicular field Bz(x)B_z(x) that is localized across the transport direction xx and translation invariant along yy [6, 7]. There, kyk_y is the conserved transverse momentum (equivalently, the angle of incidence of an electron on the barrier), and Eq. (32) is exactly the effective 1D scattering potential seen by each incident channel. It is well established in this literature that an inhomogeneous Bz(x)B_z(x) acts as a wavevector-dependent, rather than electrostatic, barrier: transmission depends sensitively on kyk_y, which is the basis of proposed velocity- and angle-filtering devices and of magnetic focusing schemes [6, 7].

Equations (B5)–(B7) show that two magnetic barriers with visibly different profiles Bz(x)≠B~z(x)B_z(x) \ne\widetilde{B}_z(x)—produced, say, by differently shaped ferromagnetic stripes—can be tuned, through an accompanying electrostatic gate voltage that realizes ϕ→ϕ+δϕ\phi\to\phi+\delta\phi, to be exactly indistinguishable in transmission and phase for one selected incidence channel ky0k_y^0, while differing for every other channel. Since split gates alongside the ferromagnetic stripe are already a standard part of these devices [7], no new fabrication step is required, only a specific choice of gate voltage as a function of Bz(x)B_z(x) and B~z(x)\widetilde{B}_z(x) via Eq. (33). This suggests two concrete uses: (i) an experimental test of the present construction, by comparing transmission through two differently fabricated barriers, tuned this way, at normal and oblique incidence; and (ii) a barrier that can be reshaped (e.g. to optimize some other figure of merit) while leaving the response of a fixed injection channel unaffected, so that off-channel changes serve as a built-in diagnostic of the deformation itself.

Numerical study: a resonant magnetic barrier

To illustrate the construction quantitatively we consider transmission through a purely magnetic barrier in an infinite (in yy) 2DEG strip, where kyk_y is a good quantum number, and a deformation that protects one incidence channel ky1≡ky0k_{y1} \equiv k_y^0. Units are ℏ=m=e=1\hbar=m=e=1, with a length scale ℓ\ell (see below).

Barrier profile and deformation family

The field of one edge of a magnetized stripe of thickness HH at depth dd below the 2DEG is B1(x)=bln⁡x2+d2x2+(d+H)2B_1(x)=b\ln\frac{x^2+d^2}{x^2+(d+H)^2}, with the closed-form vector potential

A1(x)=b{xln⁡x2+d2x2+(d+H)2+2darctan⁡xd−2(d+H)arctan⁡xd+H}.(39)A_1(x)=b\left\{x\ln\frac{x^2+d^2}{x^2+(d+H)^2}+2d\arctan\frac{x}{d}-2(d+H)\arctan\frac{x}{d+H}\right\}. \tag*{(39)}

A stripe of length LxL_x is B1(x+Lx/2)−B1(x−Lx/2)B_1(x+L_x/2)-B_1(x-L_x/2) (zero net flux, so that Ay→0A_y\to0 at ∣x∣→∞|x|\to\infty). A single stripe gives a smooth non-resonant barrier. To obtain resonances we use two identical stripes with centers at x=±g/2x=\pm g/2; the region between them is a well in Vky(x)V_{k_y}(x), and the structure acts as a Fabry–Perot resonator. We use b=0.18b=0.18, d=1d=1, H=1.5H=1.5, Lx=3L_x=3, g=6g=6. For a GaAs 2DEG (m∗=0.067mem^*=0.067m_e) and ℓ=50\ell=50 nm this corresponds to an energy unit ℏ2/m∗ℓ2≈0.455\hbar^2/m^*\ell^2\approx0.455 meV and a field unit ℏ/eℓ2≈0.263\hbar/e\ell^2\approx0.263 T. For this profile Ay(x)A_y(x) is even and Bz(x)B_z(x) is odd.

The deformation family is the one-parameter family

A~y=(1+a)Ay,(40)\widetilde{A}_y=(1+a)A_y, \tag*{(40)}
eϕ~=aky1Ay−(a+a22)Ay2,(41)e\widetilde{\phi}=ak_{y1}A_y-\left(a+\frac{a^2}{2}\right)A_y^2, \tag*{(41)}

i.e. δAy=aAy\delta A_y=aA_y in Eq. (33), with a≡ampa\equiv\mathrm{amp} the deformation amplitude (we write amp in the figures). It is localized (the leads are untouched), the magnetic field is simply rescaled, B~z=(1+a)Bz\widetilde{B}_z=(1+a)B_z, and

V~ky(x)=Vky(x)+a(ky1−ky)Ay(x),(42)\widetilde{V}_{k_y}(x)=V_{k_y}(x)+a(k_{y1}-k_y)A_y(x), \tag*{(42)}

which is Eq. (34) for this δAy\delta A_y. The value a=−1a=-1 switches the magnetic field off: B~z=0\widetilde{B}_z=0 and the barrier is purely electrostatic (an “electrostatic twin”), which retains the protected channel only. The fields before and after the deformation are shown in Figure 7. For ky1=0.15k_{y1}=0.15 and a=0.5a=0.5 the maximal field grows from ∣Bz∣max⁡=0.268|B_z|_{\max}=0.268 to 0.4020.402, and the compensating potential ϕ~\widetilde{\phi} consists of two wells over the stripes (minimum −0.528-0.528 at x=+2.88x=+2.88), with ∣eEx∣max⁡=0.231|eE_x|_{\max}=0.231; ExE_x is odd in xx.

Fields for the double-stripe barrier and the deformation

Figure 7. Fields for the double-stripe barrier and the deformation (B12)–(B13) with ky1=0.15k_{y1}=0.15, a=0.5a=0.5. (a) Magnetic field BzB_z before and after (the profile is rescaled by 1+a1+a; shaded: the two ferromagnetic stripes). (b) Compensating potential eφ~e\tilde{\varphi} with its linear-in-AyA_y and diamagnetic (Ay)2(A_y)^2 parts. (c) Electric field eExeE_x, odd in xx and localized near the stripes. (d) Effective 1D potentials VkyV_{k_y} for several kyk_y (thick: before; dashed: after): identical for ky=ky1k_y=k_{y1}, shifted by a(ky1−ky)Aya(k_{y1}-k_y)A_y otherwise. resonance flow versus aa for five values of kyk_y, which is linear for small ∣a∣|a| and has a fixed point at ky=ky1k_y=k_{y1} (horizontal line); (e) Eres−ky2/2E_{\mathrm{res}}-k_y^2/2 versus aa: all lines meet at a=−1a=-1, where the magnetic field is off and the problem reduces to an electrostatic barrier, for which the potential is independent of kyk_y apart from the trivial shift ky2/2k_y^2/2; (f) the width flows too and is frozen at ky1k_{y1}. The peak value of TT at resonance is exactly 1 for any kyk_y and aa, because Vky(x)V_{k_y}(x) is even in xx (the double-stripe potential and the compensation preserve the parity), so that the resonance is a perfect-transmission peak, and the effect of the deformation is a pure displacement and a change of the width of the peak, not of its height.

Exact invariance of the protected channel

The scattering problem for fixed (E,ky)(E,k_y) was solved by Numerov integration of Eq. (31) from x=+Lx=+L with an outgoing wave, backward to x=−Lx=-L, giving the complex amplitudes tt, rr and T=(kR/kL)∣t∣2T=(k_R/k_L)|t|^2. For ky=ky1k_y=k_{y1}, Vky1V_{k_{y1}} coincides before and after to ∼4×10−16\sim4\times10^{-16}, and the complex transmission amplitude t(E)t(E) (not just TT) is identical for all energies to ∼10−13\sim10^{-13}, for any aa. Equivalently, any wave packet incident at fixed ky1k_{y1} evolves identically. For ky≠ky1k_y\ne k_{y1} the transmission changes (Figure 8–Figure 9).

Figure 8. Resonance flow in the double-stripe barrier for ky1=0.15k_{y1}=0.15. (a)–(c) T(E)T(E) at ky=ky1k_y=k_{y1}, 0 and 0.30 for several amplitudes a=ampa=\mathrm{amp}: all curves coincide in (a). (d) Positions of the first (solid) and second (dashed) resonance versus aa for five kyk_y; the line for ky=ky1k_y=k_{y1} is flat. (e) Eres−ky2/2E_{\mathrm{res}}-k_y^2/2: all lines meet at a=−1a=-1 (magnetic field off, electrostatic twin). (f) Resonance width (FWHM) versus aa; frozen at ky1k_{y1}.

Figure 9. Transmission T(E,ky)T(E,k_y) of the resonant barrier (ky1=0.15k_{y1}=0.15, a=0.5a=0.5). (a) TT before deformation. (b) ΔT=T~−T\Delta T=\widetilde{T}-T, linear scale. (c) ∣ΔT∣|\Delta T|, log scale. (d) Relative change ∣Δln⁡T∣|\Delta\ln T| of the suppressed quantity, which isolates the protected line ky=ky1k_y=k_{y1} from the saturated regions. (e) Positions of the resonance ridges before and after. (f) Slices ΔT(ky)\Delta T(k_y) at fixed energy; all pass through zero at ky1k_{y1}.

A remark on saturation is needed when reading difference plots. A map of ΔT=T~−T\Delta T=\widetilde{T}-T is also close to zero where T~≈0\widetilde{T}\approx0 or T≈1T\approx1 for any kyk_y, so a small ΔT\Delta T alone is not a sign of protection. The protected line ky=ky1k_y=k_{y1} has to be recognized by the vanishing of the relative change of the suppressed quantity, ∣Δln⁡T∣|\Delta\ln T| (or ∣Δln⁡R∣|\Delta\ln R| where T≈1T\approx1), which vanishes only there (Figure 9).

Resonance flow

Because the transmission at fixed kyk_y is a one-dimensional scattering problem in the potential (B14), the resonance positions Eres(ky)E_{\mathrm{res}}(k_y) respond to the deformation in a controlled way. To first order in aa, perturbation theory for a quasi-bound state in the well gives

δEres≃a(ky1−ky)⟨Ay⟩res,(43)\delta E_{\mathrm{res}}\simeq a(k_{y1}-k_y)\langle A_y\rangle_{\mathrm{res}}, \tag*{(43)}

where ⟨Ay⟩res\langle A_y\rangle_{\mathrm{res}} is the average of AyA_y over the resonant state. We find ⟨Ay⟩≈−0.63\langle A_y\rangle\approx-0.63 and −0.72-0.72 for the first and second resonances, close to the average over the cavity (−0.62-0.62 to −0.64-0.64). Figure 8 shows: (a) T(E)T(E) at ky=ky1k_y=k_{y1}, where all curves for different aa coincide, with the first resonance at Eres=0.372295E_{\mathrm{res}}=0.372295 and ΓFWHM=0.0183\Gamma_{\mathrm{FWHM}}=0.0183 for ky1=0.15k_{y1}=0.15, and a second resonance near 0.700.70; (b,c) the same at ky=0k_y=0 and ky=0.30k_y=0.30, where the resonances move in opposite directions for ky<ky1k_y<k_{y1} and ky>ky1k_y>k_{y1} (the sign of the shift in Eq. (43) is that of a(ky1−ky)⟨Ay⟩a(k_{y1}-k_y)\langle A_y\rangle); (d) the

The point a=−1a=-1 provides a nontrivial numerical check. There V~ky=12ky2−ky1Ay+12Ay2\tilde{V}_{k_y}=\frac{1}{2}k_y^2-k_{y1}A_y+\frac{1}{2}A_y^2 depends on kyk_y only through the additive constant ky2/2k_y^2/2, so T~\tilde{T} is a function of E−ky2/2E-k_y^2/2 and ∂kyln⁡T~=−ky∂Eln⁡T~\partial_{k_y}\ln\tilde{T}=-k_y\partial_E\ln\tilde{T}. At ky=ky1k_y=k_{y1} we have T~=T\tilde{T}=T for all EE, hence ∂kyln⁡T~∣ky1=−ky1∂Eln⁡T∣ky1\left.\partial_{k_y}\ln\tilde{T}\right|_{k_{y1}}=-k_{y1}\left.\partial_E\ln T\right|_{k_{y1}}; at a test energy the two sides are +5.5361 and +5.5360.

Maps T(E,ky)T(E,k_y)

Figure 9 shows the angle–energy dependence of the transmission: (a) T(E,ky)T(E,k_y) for a=0a=0, where the resonances are narrow ridges whose position depends on kyk_y; (b) ΔT\Delta T on a linear scale; (c) ∣ΔT∣|\Delta T| on a logarithmic scale; (d) the relative change ∣Δln⁡T∣|\Delta\ln T| of the suppressed quantity; (e) the ridge positions before and after; (f) slices ΔT(ky)\Delta T(k_y) at fixed EE, all passing through zero at ky1k_{y1}. The deformation moves the ridges sideways in the (E,ky)(E,k_y) plane around the vertical line ky=ky1k_y=k_{y1}, which stays fixed, while the ridge of the unprotected channels moves through ΔT∼1\Delta T\sim1 (the peak is shifted by more than its width). It is therefore a very sensitive probe: a change of the resonance position by less than the width Γ\Gamma gives a full change of TT at fixed energy.

Resonant beam

The kyk_y-dependence of the resonance has two observable consequences for a beam Ψ(x,y)=∫g(ky)ψky(x)eikyydky\Psi(x,y)=\int g(k_y)\psi_{k_y}(x)e^{ik_y y}dk_y with a narrow Gaussian distribution gg around ky1k_{y1} at the resonance energy (we take g=W/kxg=\sqrt{W/k_x} so that the incident flux density is W(ky)W(k_y)). First, the resonance acts as a filter in angle: at E=EresE=E_{\mathrm{res}} and σk=0.02\sigma_k=0.02 the transmitted fraction is 0.487 before and 0.393 after, because the width of the resonance in kyk_y is comparable to σk\sigma_k and the neighbouring kyk_y components are detuned by Eq. (43). The total transmitted fraction changes at first order in aa whenever σk\sigma_k is not small compared to the width of the resonance in kyk_y: for σk=0.002\sigma_k=0.002, 0.004, 0.008, 0.012, 0.02, 0.04 (at a=0.5a=0.5) we find ΔTtot=−2.4,−6.7,−12.9,−16.2,−19.3,−19.9%\Delta T_{\mathrm{tot}=-2.4,-6.7,-12.9,-16.2,-19.3,-19.9\%}. Protection is therefore a property of the single channel ky1k_{y1}, and not of a beam of finite angular width. Second, the transmitted beam is displaced laterally: the centroid at x=26x=26 (free propagation +4.6) moves from +70.2 before to +89.2 after, with an rms width increasing from 52 to 63 (Figure 10). The corresponding shift follows from ∂kyarg⁡t\partial_{k_y}\arg t (a Goos–Hänchen-like shift) and for non-resonant propagation agrees with the classical drift −a∫Ay/kx dx-a\int A_y/k_x\,dx. At resonance the transverse drift ∂kyEres\partial_{k_y}E_{\mathrm{res}} is 0.783 for a=0a=0, 1.100 for a=0.5a=0.5 and 0.4665 for a=−0.5a=-0.5, corresponding to ∂kyarg⁡t=−95.2,−134.4,−55.9\partial_{k_y}\arg t=-95.2,-134.4,-55.9, in agreement within ∼10%\sim10\% with the estimate −2Eres/Γ=−85.7,−120.3,−51.0-2E_{\mathrm{res}}/\Gamma=-85.7,-120.3,-51.0. The “resonance dragged along yy” is clearly seen in ∣Ψ(x,y)∣2|\Psi(x,y)|^2.

Figure 10. Beam at the resonance energy E=EresE = E_{\mathrm{res}}, ky1=0.15k_{y1} = 0.15, σk=0.02\sigma_k = 0.02, a=0.5a = 0.5. (a) kyk_y-content of the beam at the resonance energy. (b) Transmitted profile at x=26x = 26 (absolute intensity). (c,d) ∣Ψ(x,y)∣2|\Psi(x,y)|^2 before and after: the transmitted beam is dragged along yy and its transmitted fraction is reduced.

For comparison, for a single (non-resonant) stripe, a beam incident at E=0.4E=0.4 with ky1=0.15k_{y1}=0.15, σk=0.1\sigma_k=0.1 and a=0.5a=0.5 has its mean outgoing angle changed from 6.4∘6.4^\circ to 5.75∘5.75^\circ, centroid shift from +10.8 to +13.9 and transmitted fraction from 0.4707 to 0.4776: these effects are small compared with the resonant case, where the sharp kyk_y-dependence of the resonance position amplifies the sensitivity.

Gate realization and tolerances

The compensating potential (B13) is not produced by the ferromagnetic stripes alone: it needs electrostatic gates, which also must cover the region between the stripes. We tested what accuracy is required by fitting ϕ(x)\phi(x) for ky1=0.15k_{y1}=0.15, a=0.5a=0.5 with a few independent, left–right symmetric Gaussian-screened electrodes of width s=0.5ℓs=0.5\ell (Figure 11). With 2, 3, 6 electrodes the resonance of the protected channel is shifted by −0.19-0.19, −0.07-0.07, −0.02-0.02 Γ respectively. If the amplitude of the ideal potential is off by 1% (5%) the resonance shifts by ∓0.17\mp0.17 Γ (∓0.84\mp0.84 Γ), and smoothing of the popotential by the screening length s=0.25s=0.25, 0.50.5, 1.0 ℓ1.0\,\ell gives −0.10-0.10, −0.38-0.38, −1.4 Γ-1.4\,\Gamma. If only the regions over the stripes are gated the shift is +7.8 Γ+7.8\,\Gamma, i.e. the protection is lost entirely. For a resonance of width Γ≃0.018\Gamma\simeq0.018 this means that the gate amplitude must be controlled at the level of a few percent, or better if the resonance is to be held within a fraction of its width. We suggest, but have not tested, a calibration procedure: tune the gate voltages until the transmission at (E,ky1)(E,k_{y1}) coincides with that of the original device, and then use the other channels as a diagnostic.

Figure 11. Gate realization. (a) Ideal compensating potential eϕ~(x)e\widetilde{\phi}(x) and least-squares fits by 2, 3 and 6 independent left–right symmetric electrodes (Gaussian screening s=0.5ℓs = 0.5\ell). (b) The protected resonance at ky1=0.15k_{y1} = 0.15 for each realization, showing how well it is held in place.

Limitations

These results hold for the infinite strip in yy, with a kyk_y-conserving, ballistic and non-interacting electron; scattering, finite temperature, and the lateral confinement of the strip (hard walls break translation invariance in yy and couple the transverse modes) were not included. The exact protection applies to a single ky1k_{y1}, and for beams or finite-temperature distributions the effect on the total transmission is generically of first order in aa.

References

  1. [11]A. M. Pupasov-Maksimov and M. S. Oliveira, Propagators of singular anharmonic oscillators with quasi-equidistant spectra, arXiv preprint arXiv:2512.16695 (2025).arxiv.org/abs/2512.16695
  2. [12]G. Borg, Eine Umkehrung der Sturm-Liouilleschen Eigenwertaufgabe: Bestimmung der Differentialgleichung durch die Eigenwerte, Acta Mathematica 78, 1 (1946).DOI
  3. [13]I. M. Gel’fand and B. M. Levitan, On the determination of a differential equation from its spectral function, Izvestiya Akademii Nauk SSSR, Seriya Matematicheskaya 15, 309 (1951), english transl.: Amer. Math. Soc. Transl. (2) 1, 253–304 (1955).
  4. [14]B. M. Levitan, Inverse Sturm–Liouville Problems (VNU Science Press, Utrecht, 1987).DOI
  5. [15]J. Pöschel and E. Trubowitz, Inverse Spectral Theory, Pure and Applied Mathematics, Vol. 130 (Academic Press, Boston, 1987).
  6. [16]K. Chadan and P. C. Sabatier, Inverse Problems in Quantum Scattering Theory, 2nd ed. (Springer, New York, 1989).
  7. [17]S. Albeverio, R. Hryniv, and Y. Mykytyuk, Inverse spectral problems for Bessel operators, Journal of Differential Equations 241, 130 (2007).DOI
  8. [18]M. M. Crum, Associated Sturm–Liouville systems, Quarterly Journal of Mathematics, Oxford Second Series 6, 121 (1955).
  9. [19]F. Cooper, A. Khare, and U. Sukhatme, Supersymmetry and quantum mechanics, Physics Reports 251, 267 (1995).DOI
  10. [20]G. Eskin, Inverse spectral problem for the schrödinger equation with periodic vector potential, Communications in mathematical physics 125, 263 (1989).DOI

Paper details

vlnc:W_cay7r3ek
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents