arXiv:2609.20802
Locally optimized variational evolution for quantum many-body systems
Carolin Wille, Maximilian Marvell, Lauren Stewart, Maximillian Murphy, Vinul Wimalaweera, Lesley Gover, Andrew G. Green
Abstract
Quantum many-body dynamics is difficult to simulate because the global wavefunction develops extensive complexity, whereas local observables may retain only an intrinsic, system-size-independent complexity. This distinction is especially relevant in thermalizing systems, where local subsystems lose microscopic information and relax towards equilibrium. We introduce a variational time-evolution principle that exploits this separation by replacing global-state fidelity with a cost function defined on local reduced density matrices. The resulting locally optimized matrix-product-state (LOMPS) algorithm retains coherent short-time dynamics while using the simplification generated by thermalization at later times, and admits closed-form equations of motion analogous to the time-dependent variational principle. The same principle yields a quantum-classical algorithm in which a quantum processor evaluates the local cost and a noise-robust classical procedure performs the optimization. Proof-of-principle implementations on Quantinuum H2 and IBM Heron processors recover the characteristic local dynamics.
AI Summary
Can many-body dynamics be simulated by retaining locally relevant information rather than global wavefunction fidelity? This paper introduces locally optimized matrix-product-state evolution (LOMPS): after each Trotter step, a uniform MPS is fitted to the evolved $L$-site reduced density matrix by minimizing Hilbert–Schmidt distance. The continuous-time limit gives TDVP-like normal equations with metric $S_{ij}=\mathrm{Tr}(\partial_i\rho_L\partial_j\rho_L)$. Exact local matching preserves the target’s energy density for Hamiltonian terms supported within the patch. Benchmarks use integrable and nonintegrable Ising-chain quenches. LOMPS reproduces early-time local Loschmidt signatures and approaches the infinite-temperature reduced state after a strongly thermalizing quench. Tests with $L=2,\ldots,6$ and bond dimensions $D=5,7,12,21,41$ show progressively later deviations from a growing-bond-dimension iTEBD reference, available to $t\simeq7$. At $t=7$, the $L=6$, $D=41$ calculation has smaller reduced-state errors than iTEBD capped at bond dimensions through $164$. Late-time one-site thermal errors decrease exponentially with $L$ over the tested range. Increasing $L$ also improves agreement across weakly thermalizing and nonthermalizing quenches. A hybrid implementation measures reduced-state overlaps using destructive SWAP tests. $D=2$ demonstrations on Quantinuum H2-2 and IBM Heron recover qualitative Loschmidt dynamics using 500 and 5,000 shots per circuit evaluation, respectively. Limitations include uncontrolled influence of larger reduced states on future dynamics: small projection cost does not guarantee trajectory accuracy. Classical tests use $D\sim2^L$; runtime advantages, compressed representations, and scalable quantum advantage remain unestablished. The hardware demonstrations use warm starts and a six-parameter ansatz.
Error found: The quantum protocol's single alternating-bond gate layer followed by translation averaging implements Hamiltonian evolution for Delta t/2 to first order under the stated normalization, although the target is labeled as evolution for Delta t.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 19 pages, 13 figures. Version 2 adds finite-bond-dimension iTEBD comparisons and updated references