Quantum computer-based simulation of Stark many-body localization in a 1D Fermi-Hubbard model

This study investigates Stark many-body localization (MBL) in a one-dimensional Fermi-Hubbard model. Unlike traditional MBL, which requires random disorder, Stark MBL arises from a deterministic linear potential (a Stark field) that suppresses particle transport. The authors aim to demonstrate that current noisy intermediate-scale quantum (NISQ) devices can effectively capture this transition from ergodic (thermalizing) to localized dynamics.

The team mapped a 12-qubit correlated fermionic system onto an IBM superconducting quantum processor. To overcome hardware noise and connectivity limitations, they employed a spin-resolved Jordan-Wigner transformation, which separates spin-up and spin-down sectors to simplify the Hamiltonian. They further optimized the quantum circuits using a combination of SWAP networks and a tensor-network-based routine (AQC-Tensor), achieving an approximately 88% reduction in two-qubit gate counts and 87% reduction in circuit depth. The real-time dynamics were simulated using Trotterized quantum circuits, with error mitigation techniques including dynamical decoupling and Pauli twirling.

The simulations successfully identified the crossover between two distinct dynamical regimes. At weak field tilts, the system exhibits thermalization, where particles redistribute across the lattice and local memory of the initial state is lost. At large field tilts, the system enters the Stark-MBL phase, characterized by the retention of the initial charge-density-wave pattern and suppressed particle transport. These experimental results show strong agreement with exact diagonalization benchmarks, validating the efficacy of the proposed circuit optimization pipeline for studying many-body quantum phenomena on NISQ hardware.

This work provides a practical blueprint for simulating complex many-body systems on existing quantum hardware. By demonstrating that Stark MBL can be studied without the need for complex disorder averaging, the authors highlight a more efficient path for exploring quantum thermodynamics and non-ergodic behavior in isolated systems, which are otherwise computationally expensive to model on classical machines.