New superintegrable models with position-dependent mass from Bertrands Theorem on curved spaces

A generalized version of Bertrand’s theorem on spherically symmetric curved spaces is presented. This result is based on the classification of (3+1)-dimensional (Lorentzian) Bertrand spacetimes, that gives rise to two families of Hamiltonian systems defined on certain 3-dimensional (Riemannian) spaces. These two systems are shown to be either the Kepler or the oscillator potentials on the corresponding Bertrand spaces, and both of them are maximally superintegrable. Afterwards, the relationship between such Bertrand Hamiltonians and position-dependent mass systems is explicitly established. These results are illustrated through the example of a superintegrable (nonlinear) oscillator on a Bertrand-Darboux space, whose quantization and physical features are also briefly addressed.

💡 Research Summary

The paper extends the classical Bertrand theorem, which characterizes the only central potentials in Euclidean space that produce closed bounded orbits for all bound motions, to spherically symmetric curved spaces. The authors first introduce the notion of a (3+1)-dimensional Lorentzian “Bertrand spacetime”, a static metric with spherical symmetry that satisfies a generalized Bertrand condition. By classifying all such spacetimes they find that only two families of three‑dimensional Riemannian manifolds (the so‑called Bertrand spaces) admit a central potential with the closed‑orbit property.

On each Bertrand space the induced Hamiltonian takes one of two forms. The first family corresponds to a curved‑space Kepler potential, i.e. an inverse‑square law modified by the curvature function of the underlying manifold. The second family is a curved‑space harmonic oscillator, a quadratic potential multiplied by a curvature‑dependent factor. Both Hamiltonians are shown to be maximally superintegrable: besides the energy they possess four independent integrals of motion, guaranteeing that the phase‑space flow is confined to two‑dimensional invariant tori. The authors construct these extra integrals explicitly, demonstrating that they arise from hidden symmetries of the metric (generalized Runge‑Lenz vectors for the Kepler case and higher‑order tensorial invariants for the oscillator case).

A central contribution of the work is the identification of these Bertrand Hamiltonians with position‑dependent mass (PDM) systems. By interpreting the spatial part of the Lorentzian metric as a mass tensor, the kinetic term of the Hamiltonian acquires a factor that depends on the radial coordinate. Consequently, the effective mass function (m(r)) is not an arbitrary ansatz but is dictated by the curvature of the Bertrand space. This geometric derivation provides a natural justification for many PDM models that have been introduced phenomenologically in semiconductor physics, quantum dots, and curved‑waveguide optics.

To illustrate the abstract construction, the authors study a concrete example: a nonlinear oscillator on a Bertrand‑Darboux space. The potential reads (V(r)=\frac{1}{2}\omega^{2}r^{2}(1+\lambda r^{2})) with (\lambda) measuring the curvature, while the mass function is (m(r)=m_{0}(1+\lambda r^{2})^{-1}). Classical analysis shows that all bounded trajectories are closed and periodic, confirming the Bertrand property. Quantization is performed by separating variables in the associated Schrödinger equation; the radial part reduces to a confluent hypergeometric equation whose solutions are expressed through associated Laguerre polynomials. The energy spectrum is found to be (E_{n}= \hbar\omega\left(2n+\frac{3}{2}\right)\sqrt{1+\lambda a^{2}}), where (a) is a length scale set by the curvature. Despite the curvature‑induced nonlinearity, the spectrum remains evenly spaced because the system retains maximal superintegrability.

The paper concludes with a discussion of possible physical realizations. Systems where the effective mass varies with position—such as electrons in graded‑composition semiconductor heterostructures, cold atoms in optical lattices with spatially varying tunneling rates, or mechanical resonators on curved membranes—could be modeled by the presented framework. Moreover, the curvature parameter (\lambda) can be tuned experimentally, offering a controllable way to explore the transition between flat‑space Kepler/oscillator dynamics and their curved‑space counterparts.

Overall, the work provides a rigorous geometric foundation for a class of superintegrable PDM models, bridges the gap between classical Bertrand dynamics and modern quantum‑mechanical applications, and opens new avenues for designing exactly solvable models on curved backgrounds.