arXiv:2610.03673
Quantum Simulation on Riemannian Manifolds
Dylan Herman, Jacob Watkins, Guneykan Ozgul, Jiayu Shen, Brandon Augustino, Junhyung Lyle Kim, Shouvanik Chakrabarti
Abstract
We investigate algorithms for the quantum simulation of the Schrödinger equation on a Riemannian manifold, where the kinetic operator is defined by the Laplace--Beltrami operator corresponding to the metric. Our first algorithms are based on a global spectral method based on the identification of an efficient transform to the eigenbasis of the Laplace--Beltrami operator. We use this method to provide explicit, efficient, quantum simulation algorithms for the Riemannian Schrödinger equation on tori and spheres with their standard metrics, simplices with the Wright--Fisher metric, truncated positive orthants and their invertible affine images with the log-barrier Hessian metric, and $\ell_p$ balls with a metric induced by the Duffy map. Our second algorithm is based on a coherent simulation of local spectral methods on multiple charts, and is in principle applicable to any compact manifold. We first analyze this algorithm in the continuum and derive conditions under which a polynomial spectral cutoff suffices. We also provide a discretization analysis of a polynomial spectral cutoff for tensor-products of constant-dimensional manifolds. Finally, we consider applications of these methods to optimization and physical simulation. For optimization, we provide results including a generalization and convergence analysis of Quantum Hamiltonian Descent for geodesically convex functions that leads to explicit algorithms on the sphere and simplex, and a Riemannian generalization of the Real-Space Adiabatic Algorithm. For physical simulation, we show that our algorithms can simulate certain spatially discretized field theories, including a variant of the nonlinear sigma model.
AI Summary
The paper studies digital simulation of $i\partial_t\psi=[-a(t)\Delta_g+b(t)f]\psi$ on Riemannian configuration spaces, including discretization errors between continuum dynamics and quantum circuits. Motivations include constrained quantum-dynamical optimization and spatially discretized field theories. The first approach uses global Laplace–Beltrami eigenfunctions. A uniformly accurate band-limited proxy $f_K$ is introduced only in the analysis; the algorithm still queries $f$. When multiplication adds spectral degrees, a Dyson-series argument bounds leakage beyond cutoff $N\geq K_0+(q-1)K$. Oversampled exact quadrature removes aliasing for the proxy, giving state error at most $2B_1\|f-f_K\|_\infty+2^{2-q}$, where $B_1=\int_0^T|b(t)|dt$ and $q\geq2eB_1\|f_K\|_\infty$. Explicit digital algorithms cover round spheres, the Wright–Fisher simplex, and flat tori; coordinate changes also cover selected truncated orthants and $\ell_p$ domains. On tori, Lipschitz potentials admit polynomial grid resolution in dimension and inverse accuracy, using $O(d\log M)$ qubits for $M$ nodes per coordinate. A second approach coherently combines local Fourier representations using a partition of unity and the IMS kinetic-energy correction. It establishes polynomial continuum cutoffs under geometric and Sobolev assumptions, with a finite-dimensional analysis for products of constant-dimensional manifolds. Applications include a lattice $O(N)$ sigma-model simulation, convergence guarantees for Riemannian Quantum Hamiltonian Descent, and an adiabatic optimization formulation. General-manifold results remain conditional on substantial regularity and interpolation assumptions. Some domain constructions require artificial periodic identifications or specially chosen flat metrics. Most applications do not establish complete quantum speedups; state preparation, measurement, spectral gaps, optimized resource bounds, and classical comparisons remain separate issues.
Large language model tools were used interactively at all stages of the preparation of this manuscript: from the earliest exploratory research stages to the final exposition and writing steps.
Error found: The orthant and affine-image simulation guarantees require periodic identification in logarithmic coordinates and face-compatible potentials, qualifications absent from the abstract's description of these domains. The explicit ℓ_p-domain guarantees concern coordinate-truncated Duffy domains with a specially constructed metric and periodic compatibility assumptions, rather than unrestricted full ℓ_p balls. The supplied simplex QHD result requires Dirichlet evolution on a geodesically convex subdomain, so it cannot directly be combined with the displayed simplex simulation guarantee to support the claimed end-to-end optimization algorithm.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.