Enhanced Quantum behavior on frustrated Ising model: Quantum Approximate Optimization Algorithm study

Seunghan Lee, Hunpyo Lee · 2025

Read Enhanced Quantum behavior on frustrated Ising model: Quantum Approximate Optimization Algorithm study by Seunghan Lee, Hunpyo Lee (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-2507-07457.txt.

Introduction

Seunghan Lee and Hunpyo Lee probe quantum effects in a frustrated magnet model with a quantum optimizer. Frustration pits spins against each other so no layout pleases all. Classic studies force each spin up or down, which can miss quantum wiggle. Their 2025 study runs the Quantum Approximate Optimization Algorithm on a square grid of such spins. A new score gauges how much of the behavior is truly quantum.

In weakly frustrated zones, the algorithm rarely lands on the first excited state, which sits far above the ground. Near the phase change point, excited states show up often in the reads. That crowding marks where quantum fluctuations take over. The metric gives a clean dial for the quantum-classical border.

Catalog record

CREATORSeunghan Lee, Hunpyo Lee
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE16 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-07457

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Enhanced Quantum behavior on frustrated Ising model: Quantum Approximate Optimization Algorithm study Seunghan Lee, Hunpyo Lee Abstract We investigated the quantum effects of a frustrated Ising model on a two-dimensional square lattice using the Quantum Approximate Optimization Algorithm (QAOA). While strong spin frustration is known to induce quantum fluctuations at low temperatures, previous classical approaches—restricted to binary (up/down) spin configurations—have been insufficient to fully capture the quantum contributions of frustration. In this study, we introduced a quantitative metric to evaluate the quantum effects arising from frustration and employed QAOA to differentiate between classical and quantum regimes. Notably, we found that in the weakly frustrated region, QAOA measurements rarely capture first excited states, as they are energetically well separated from the ground state. In contrast, near the quantum phase transition point, excited states appear more frequently in QAOA measurements, highlighting the increased role of quantum fluctuations. pacs 71.10.Fd,71.27.+a,71.30.+h I Introduction The frustrated Ising model, which undergoes a quantum phase transition and represents a class of combinatorial optimization problems, serves as an effective testbed for evaluating the performance of newly developed computational methods. A key advantage of this model is that the energy gap between the ground state and the first excited state can be systematically tuned, making it ideal for benchmarking algorithmic efficiency and accuracy. The energy gap is a critical parameter that directly influences the complexity and precision of combinatorial optimization tasks; when the gap becomes small—particularly near the quantum phase transition—the problem becomes significantly more difficult to solve accurately. To address this challenge, a variety of numerical techniques have been developed, including Monte Carlo simulations, machine learning algorithms, simulated annealing, and quantum annealing Jin2012 ; Jin2013 ; Park2022 ; Lee2024 ; Romer2025 . Recent advances in quantum hardware—including gate-based quantum computers and quantum annealers have significantly enhanced their ability to address combinatorial optimization problems encountered in fields such as economics, computational science, and machine learning Preskill2018 ; Bharti2022 ; arute2019 ; postler2022 ; nation2025 ; Johnson2011 ; Higgott2019 ; King2021 ; Park2024(1) ; Park2024(2) . These quantum machines are believed to offer new insights into the underlying quantum nature of such problems by exploiting quantum entanglement, which emerges during the evolution of quantum states associated with optimization tasks. Consequently, it is of great interest to investigate how quantum behavior depends on the energy gap, using quantum algorithms. The Quantum Approximate Optimization Algorithm (QAOA) stands out as a prominent method for finding approximate solutions to combinatorial optimization problems using quantum entanglement farhi2014 ; farhi2015 ; farhi2019 ; choi2019 ; farhi2020 ; zhou2020 ; moussa2022 ; blekos2024 . Inspired by the Quantum Adiabatic Algorithm (QAA), it employs Trotterization to simulate time-dependent adiabatic evolution and approximate the ground state of a system, thereby addressing the original optimization task farhi2000 ; farhi2001 . Unlike QAA, which performs optimization through continuous adiabatic evolution between an initial and a final Hamiltonian, QAOA adopts a discrete, layered structure that alternates between cost and mixer Hamiltonians. When a combinatorial optimization problem is formulated in terms of binary variables, its objective function can be encoded into a cost Hamiltonian within the QAOA framework. The alternating application of cost and mixer Hamiltonians forms a parameterized quantum ansatz, which serves as a sequence of unitary time-evolution operators. A classical optimizer iteratively updates the variational parameters based on measurement outcomes from the quantum ansatz, aiming to approximate the optimal solution. The number of layers p determines the ansatz depth; as p increases, the ansatz explores a larger portion of the solution space and can yield better approximations. In the limit p=\infty , QAOA converges to quantum adiabatic computing, which theoretically guarantees the optimal solution but at the cost of significantly longer run times. In this paper, we focused on how the energy gap between the ground state and the first excited state influences both the accuracy and quantum behavior on the QAOA approach. To explore this, we considered the frustrated Ising model on a two-dimensional 4\times 4 square lattice with nearest-neighbor feromagnetic (FM) interaction J_{1} ( J_{1}>0 ) and diagonal antiferomagnetic (AF) interaction J_{2} ( J_{2}<0 ). This model, under periodic boundary conditions, exhibits a quantum phase transition between FM and stripe AF phases at zero temperature, where the energy gap \Delta/J_{1} between the ground state and the first excited state vanishes at |J_{2}|/J_{1}=0.5 . Notably, the ratio |J_{2}|/J_{1} provides a systematic means of controlling \Delta/J_{1} . Using the QAOA approach with an appropriate number of layers, via computing the order parameters of spin configurations, we first confirmed the FM-to-stripe AF phase transition at |J_{2}|/J_{1}=0.5 , which is well-known from previous Monte Carlo, simulated annealing and D-Wave quantum annealing studies ( ) . Next, assuming that the exact ground state and first excited state energies are known, we computed the probability of obtaining the correct solution as a function of the number of layers. We confirmed that a larger \Delta/J_{1} allows for a high success probability even with fewer layers thank to absence of quantum effects, whereas a smaller \Delta/J_{1} requires deeper ansatz to achieve similar accuracy. Finally, we observed that the expectation value of the energy converges to the ground state energy in the large \Delta/J_{1} , while it deviates in \Delta/J_{1} due to enhanced quantum fluctuations. The remainder of this paper is organized as follows. Section II provides a detailed description of the frustrated Ising model and the QAOA simulation. In Section III , we present and analyze the results. Finally, Section IV summarizes our conclusions and outlines the directions for future work. Figure 1: (Color online) Schematic diagram of the workflow for the Quantum Approximate Optimization Algorithm (QAOA) with p layers. H_{C} and H_{M} represent the cost Hamiltonian and mixing Hamiltonian, respectively. The variational parameters \gamma_{k} and \beta_{k} at layer index k are optimized using a classical optimizer. The initial spin configuration is prepared in an equal superposition of all possible spin states by applying the Hadamard operator to each state. II Model and Simulation The Hamiltonian of the frustrated Ising model on a two-dimensional L\times L square lattice, incorporating competing nearest-neighbor FM interactions and diagonal AF interactions, is given by H=-J_{1}\sum_{\langle i,j\rangle}\sigma_{i}^{z}\sigma_{j}^{z}-J_{2}\sum_{\langle\langle i,j\rangle\rangle}\sigma_{i}^{z}\sigma_{j}^{z}, (1) where the nearest- and diagonal-neighbors are denoted by <i,j> and <<i,j>> , respectively. We assume J_{1}>0 and J_{2}<0 , corresponding to preferences of FM and AF, respectively. Unlike classic algorithms, which assume each spin variable takes discrete values of +1 or -1 , the Ising Hamiltonian must be reformulated in terms of quantum gates that are implemented in a quantum ansatz, where spins are represented by the simulated qubits Jin2012 ; Jin2013 ; Park2022 ; Lee2024 ; Romer2025 . The spin interactions are then translated into simulated qubit interactions. We reformulated the frustrated Ising model as a combinatorial optimization problem, aiming to find the minimum (or maximum) of a cost function C(z) , where z is a classic bitstring. In this formulation, the cost Hamiltonian H_{C} , expressed using Pauli-Z operators, is given by H_{C}=-J_{1}\sum_{\langle i,j\rangle}Z_{i}Z_{j}-J_{2}\sum_{\langle\langle i,j^{\prime}\rangle\rangle}Z_{i}Z_{j}^{\prime}, (2) where Z_{i} is the Pauli-Z operator acting on the qubit i and Z_{i}Z_{j} represents a two-qubit interaction. The corresponding mixing Hamiltonian H_{M} , which introduces quantum fluctuations, is defined as H_{M}=\sum_{j}X_{j}, (3) where X_{j} is the Pauli-X operator acting on the simulated qubit j .

Read the complete text