Salvato in:
Dettagli Bibliografici
Autori principali: Gaberle, Cedric, Jattana, Manpreet Singh
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2509.13034
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915497364160512
author Gaberle, Cedric
Jattana, Manpreet Singh
author_facet Gaberle, Cedric
Jattana, Manpreet Singh
contents We propose an optimization method for the Variational Quantum Eigensolver (VQE) that combines adaptive and physics-inspired ansatz design. Instead of optimizing multiple layers simultaneously, the ansatz is built incrementally from its operator subsets, enabling subspace optimization that provides better initialization for subsequent steps. This quasi-dynamical approach preserves expressivity and hardware efficiency while avoiding the overhead of operator selection associated with adaptive methods. Benchmarks on one- and two-dimensional Heisenberg and Hubbard models with up to 20 qubits show improved fidelities, reduced function evaluations, or both, compared to fixed-layer VQE. The method is simple, cost-effective, and particularly well-suited for current noisy intermediate-scale quantum (NISQ) devices.
format Preprint
id arxiv_https___arxiv_org_abs_2509_13034
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Slice-Wise Initial State Optimization to Improve Cost and Accuracy of the VQE on Lattice Models
Gaberle, Cedric
Jattana, Manpreet Singh
Quantum Physics
We propose an optimization method for the Variational Quantum Eigensolver (VQE) that combines adaptive and physics-inspired ansatz design. Instead of optimizing multiple layers simultaneously, the ansatz is built incrementally from its operator subsets, enabling subspace optimization that provides better initialization for subsequent steps. This quasi-dynamical approach preserves expressivity and hardware efficiency while avoiding the overhead of operator selection associated with adaptive methods. Benchmarks on one- and two-dimensional Heisenberg and Hubbard models with up to 20 qubits show improved fidelities, reduced function evaluations, or both, compared to fixed-layer VQE. The method is simple, cost-effective, and particularly well-suited for current noisy intermediate-scale quantum (NISQ) devices.
title Slice-Wise Initial State Optimization to Improve Cost and Accuracy of the VQE on Lattice Models
topic Quantum Physics
url https://arxiv.org/abs/2509.13034