Sample-Based Krylov Quantum Diagonalization for the Schwinger Model on Trapped-Ion and Superconducting Quantum Processors
Emil Otis Rosanowski, Jurek Eisinger, Lena Funcke, Ulrich Poschinger, Ferdinand Schmidt-Kaler · 2025
Read Sample-Based Krylov Quantum Diagonalization for the Schwinger Model on Trapped-Ion and Superconducting Quantum Processors by Emil Otis Rosanowski, Jurek Eisinger, Lena Funcke, Ulrich Poschinger, Ferdinand Schmidt-Kaler (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-26951.txt.
Introduction
Emil Otis Rosanowski works with Jurek Eisinger, Lena Funcke, Ulrich Poschinger, and Ferdinand Schmidt-Kaler to test a quantum method on real hardware. The method, called sample-based Krylov quantum diagonalization, estimates the lowest energy state of a system. It builds a small working space from bit strings sampled on a quantum device, then solves the problem on a normal computer.
The test case is the Schwinger model, a simple model of electric charge and fields, with an added angle term. The team tracks energy and particle number as that angle changes and recovers the phase pattern. Runs on trapped-ion and superconducting processors agree, and the working space stays much smaller than the full space.
Catalog record
| CREATOR | Emil Otis Rosanowski, Jurek Eisinger, Lena Funcke, Ulrich Poschinger, Ferdinand Schmidt-Kaler |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-26951
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Sample-Based Krylov Quantum Diagonalization for the Schwinger Model on Trapped-Ion and Superconducting Quantum Processors Emil Otis Rosanowski, Jurek Eisinger, Lena Funcke, Ulrich Poschinger, Ferdinand Schmidt-Kaler Abstract We apply the recently proposed Sample-based Krylov Quantum Diagonalization (SKQD) method to lattice gauge theories, using the Schwinger model with a θ \theta -term as a benchmark. SKQD approximates the ground state of a Hamiltonian, employing a hybrid quantum–classical approach: (i) constructing a Krylov space from bitstrings sampled from time-evolved quantum states, and (ii) classically diagonalizing the Hamiltonian within this subspace. We study the dependence of the ground-state energy and particle number on the value of the θ \theta -term, accurately capturing the model’s phase structure. The algorithm is implemented on trapped-ion and superconducting quantum processors, demonstrating consistent performance across platforms. We show that SKQD substantially reduces the effective Hilbert space, and although the Krylov space dimension still scales exponentially, the slower growth underscores its promise for simulating lattice gauge theories in larger volumes. I Introduction Efficient and accurate determination of the low-energy spectrum of a Hamiltonian is a central challenge in particle and condensed-matter physics, quantum chemistry, and materials science, and serves as a key benchmark for quantum simulation and algorithm development. Exact diagonalization of the Hamiltonian becomes exponentially costly with system size on classical computers. The best-known quantum algorithm, quantum phase estimation Kitaev 1995 , remains impractical on current noisy intermediate-scale quantum (NISQ) devices Preskill 2018 due to the large circuit depths required (see, e.g., Refs. Lin and Tong 2022 ; Ding and Lin 2023 ). Hybrid quantum-classical methods, such as the Variational Quantum Eigensolver (VQE) Peruzzo et al. 2014 , offer greater hardware efficiency but rely on noisy parametric optimization and extensive sampling of observables, resulting in unfavorable scaling, high runtime, and Barren plateaus Wang et al. 2021 ; Anschuetz and Kiani 2022 . As an alternative approach, Krylov Quantum Diagonalization (KQD) has been proposed for probing spectral properties on pre-fault-tolerant quantum devices Parrish and McMahon 2019 ; Motta et al. 2019 ; Evangelista 2020 ; de Jong 2020 ; Cohn et al. 2021 ; Seki and Yunoki 2021 ; Baker 2021 ; Baker 2024 ; Klymko et al. 2022 ; Jamet et al. 2022 ; Lee et al. 2024 ; Kirby et al. 2023 ; Shen et al. 2023 ; Tkachenko et al. 2024 ; Epperly et al. 2022 ; Kirby 2024 . In KQD, a Krylov space is built by evolving a reference state over a series of time intervals, after which the Hamiltonian is diagonalized classically within this subspace. The method uses time-evolution circuits to reach system sizes inaccessible to classical simulation, while ensuring convergence when the initial state has a polynomial overlap with the true ground state, meaning that the initial state’s overlap with the ground state does not shrink exponentially with system size but instead decreases only as an inverse polynomial Kim et al. 2023 ; Shinjo et al. 2024 . KQD has been used to compute eigenenergies of quantum many-body systems on lattices of up to 56 sites Yoshioka et al. 2025 and to extract ground-state properties of the lattice Schwinger model Anderson et al. 2025 . Building on these ideas, the Sample-Based Krylov Quantum Diagonalization (SKQD) method has been recently introduced Yu et al. 2025 . SKQD combines KQD with key concepts from subspace algorithms using independent samples from a quantum computer Kanno et al. 2023 ; Robledo-Moreno et al. 2025 ; Kaliakin et al. 2024 : estimating ground-state energies by sampling quantum states and performing classical post-processing and diagonalization on noisy data. Specifically, SKQD constructs a subspace from bitstrings sampled from time-evolved reference states and diagonalizes the Hamiltonian classically within this subspace to approximate the ground state. This hybrid approach inherits the noise resilience of sample-based methods while preserving the convergence guarantees of Krylov methods. SKQD has been proven to approximate the ground-state energy in polynomial time, assuming sparsity of the ground state and that the reference state has polynomial overlap with the true ground state Yu et al. 2025 . The method has been demonstrated experimentally for quantum many-body systems with up to 41 sites Yu et al. 2025 . In this paper, we present the first application of SKQD to lattice gauge theories, in particular, to the Schwinger model Schwinger 1962 . The Schwinger model describes quantum electrodynamics in 1+1 dimensions and exhibits QCD-like features such as confinement, chiral symmetry breaking, and a topologically non-trivial vacuum with a \theta -term. Owing to these properties, its lattice formulation serves as a common benchmark for developing and testing numerical methods aimed at Lattice QCD applications, including new approaches to address the sign problem, tensor network techniques, and quantum algorithms (see, e.g., Refs. Bañuls et al. 2018 ; Funcke et al. 2023 for reviews). Specifically, we investigate the dependence of the ground-state energy and particle number on the \theta -term, aiming to explore the first-order phase transition that occurs in the Schwinger model at \theta=\pi Coleman 1976 ; Hamer et al. 1982 ; Buyens et al. 2017 ; Byrnes et al. 2002 . For the quantum hardware implementations, we execute SKQD on both a trapped-ion quantum processor Kaushal et al. 2020 and the IBM superconducting processors ibm_marrakesh and ibm_kingston IBM Quantum 2025 . Trapped-ion devices offer enhanced coherence times and full qubit connectivity, whereas superconducting processors currently provide faster gate times and better scalability. Accordingly, in this study, the trapped-ion processor serves as a proof-of-principle demonstration, using N=4 qubits, while the IBM devices illustrate the scaling of the method to larger systems with up to N=20 qubits. We observe excellent agreement with exact diagonalization results. Notably, although the Krylov space has a much smaller dimension than the full Hilbert space, it still allows for accurate reconstruction of energies and phase transitions. SKQD reduces the effective Hilbert space by up to 80%, while keeping the relative energy deviations to the true ground state in the order of {10}^{-3} . The Krylov space dimension still scales exponentially, but its slower growth highlights the potential of this approach for simulating larger quantum systems. The paper is organized as follows: In Section II , we introduce the Schwinger model. The SKQD method is then discussed in Section III.1 , followed by details on its application to the Schwinger model in Section III.2 . Results from the trapped-ion quantum computer are presented in Section IV.1 , while results from the IBM devices are discussed in Section IV.2 . We conclude in Section V . In Appendix A , we provide a detailed description of the trapped-ion quantum processor Kaushal et al. 2020 used in this work. Additional results for the ground-state energy at different qubit numbers, completing the scaling analysis in Section IV.2 , are presented in Appendix B . In Appendix C , we show further plots on the scaling of the Krylov space dimension with the number of Trotter steps. Finally, in Appendix D , we investigate the phase transition using SKQD with a specific ground-state initialization of the mass Hamiltonian for systems up to N=30 qubits, where exact solutions are not available. II The Schwinger Model II.1 Continuum Formulation In the continuum, the Hamiltonian density of the Schwinger model Schwinger 1962 with a single fermion flavor