Problem specific ion native ansatz for combinatorial optimization

Georgii Paradezhenko, Daniil Rabinovich, Ernesto Campos, Kirill Lakhmanskiy · 2025

Read Problem specific ion native ansatz for combinatorial optimization by Georgii Paradezhenko, Daniil Rabinovich, Ernesto Campos, Kirill Lakhmanskiy (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-2505-15898.txt.

Introduction

Georgii Paradezhenko, Daniil Rabinovich, Ernesto Campos, and Kirill Lakhmanskiy build a quantum algorithm for hard optimization problems that is shaped to the native physics of trapped-ion hardware. Variational quantum algorithms need a trial circuit, called an ansatz, that fits the machine. Generic circuits fight the hardware; native ones flow with it.

Their design alternates digital gates with the ions' natural long-range Ising interaction, in the spirit of the well-known approximate optimization algorithm. Performance depends strongly on how the individual interaction strengths are set, and the paper studies that dependence. Problem-specific, hardware-native ansatz design may squeeze more from today's small quantum machines.

Catalog record

CREATORGeorgii Paradezhenko, Daniil Rabinovich, Ernesto Campos, Kirill Lakhmanskiy
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE61 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-15898

Browse the full library · About the Tower of Babel

Full text

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

Problem specific ion native ansatz for combinatorial optimization Georgii Paradezhenko, Daniil Rabinovich, Ernesto Campos, Kirill Lakhmanskiy Abstract Variational quantum algorithms have become a standard approach for solving a wide range of problems on near-term quantum computers. Identifying an appropriate ansatz configuration for variational algorithms, however, remains a challenging task, especially when taking into account restrictions imposed by real quantum platforms. This motivated the development of digital-analog quantum circuits, where sequences of quantum gates are alternated with natural Hamiltonian evolutions. A prominent example is the use of the controllable long-range Ising interaction induced in ion-based quantum computers. This interaction has recently been applied to develop an algorithm similar to the quantum approximate optimization algorithm (QAOA), but native to the ion hardware. The performance of this algorithm has demonstrated a strong dependence on the strengths of the individual ion-ion interactions, which serve as ansatz hyperparameters. In this work, we propose a heuristic for identifying a problem-specific ansatz configuration, which enhances the trainability of the ion native digital-analog circuit. The proposed approach is systematically applied to random instances of the Sherrington-Kirkpatrick Hamiltonian for up to 15 qubits, providing favorable cost landscapes. As a result, the developed approach identifies a well-trainable ion native ansatz, which requires a lower circuit depth to solve specific problems as compared to standard QAOA. This brings the algorithm one step closer to its large scale practical implementation. Preprint: AIP/123-QED I Introduction Current day quantum computers are inherently limited by both environmental and systematic noise. In the absence of error correction, these devices can only support the execution of short depth quantum circuits which fit within their coherence times [ 82 , 66 , 32 , 48 ] . Within these restrictions, the variational model of quantum computing has been developed to explore the full potential of current hardware. In a variational quantum algorithm (VQA), a short depth parametrized quantum circuit is iteratively optimized by a classical co-processor in an attempt to minimize a given cost function [ 14 ] . This model has been proven to be universal [ 9 ] adapted for a wide variety of tasks, notably: the variational quantum eigensolvers (VQEs) [ 57 , 33 , 37 , 58 , 10 , 88 , 38 , 76 , 77 ] designed to approximate the ground energy of Hamiltonians, Hamiltonian simulations [ 86 , 17 , 26 , 20 ] , the quantum approximate optimization algorithm (QAOA) [ 22 , 51 , 23 , 80 , 44 , 90 , 18 , 81 , 64 , 3 , 2 , 12 , 11 ] designed to give approximate solutions to combinatorial optimization problems, quantum circuit compilers [ 39 , 36 , 30 ] , machine learning [ 43 , 47 , 7 , 78 ] , among others [ 49 , 68 , 84 , 15 ] . Despite the benefits offered by the variational model [ 25 , 70 , 16 , 29 , 28 , 52 ] , inaccurate results arising from gate imperfections and restricted circuit depth remain a prime concern [ 63 , 1 ] . One pathway towards alleviating these limitations is the use of digital-analog quantum circuits which alternate between quantum gates and entangling natural Hamiltonian evolutions [ 85 , 24 , 46 , 27 ] , reducing the circuit dependence on entangling gates of limited fidelity. A typical example is the use of the long-range Ising interaction in trapped ion-based quantum computers [ 89 , 45 , 50 , 67 , 73 , 69 ] , which has been applied to quantum Hamiltonian simulations [ 56 ] , QAOA compilation [ 52 , 31 ] and used as an entangling layer in VQE circuits [ 93 , 41 ] . In the same fashion, [ 62 ] has proposed to use this interaction as a substitute to the problem Hamiltonian exponent to generate a QAOA-like sequence capable of minimizing general \mathbb{Z}_{2} symmetric combinatorial problems. The performance of this approach strongly depends on the choice of ion interactions referred to as ansatz hyperparameters. Indeed, problem agnostic hyperparameters can lead to trainability limitations induced by a large circuit expressivity [ 34 ] or by cost landscapes plagued with local minima [ 5 ] . On the other hand, as demonstrated in [ 62 ] , certain problem-specific ion native ansatz hyperparameters can significantly improve the algorithmic performance, yet a procedure to identify them has been lacking so far. Following the framework in [ 62 ] , we develop an approach for identifying hyperparameters suitable for minimizing specific problem instances. For a QAOA-like ansatz we propose a heuristic based on optimizing a single layer circuit expectation value alternately over variational and circuit hyper- parameters using the block coordinate descent method (see, e.g., [ 75 ] ). The proposed heuristic is benchmarked by performing Hamiltonian minimization for the Sherrington-Kirkpatrick (SK) model from 5 to 15 qubits with the coupling coefficients sampled from the standard normal distribution. This is a widely studied model in condensed matter physics, arising as a mean field approach to spin glasses [ 53 , 71 ] . We demonstrate that the proposed heuristic identifies an ansatz structure, which provides a favorable optimization landscape. When compared to standard QAOA [ 22 ] and the previous approach proposed in [ 62 ] , we establish that our heuristic ameliorates the optimization and notably reduces the circuit depth required to solve a problem. We argue that this behavior is associated to ansatz states getting effectively locked to a low-dimensional subspace that has support on the low energy states of the problem Hamiltonian. The paper is structured as follows. Sec. II provides an outline of the ion native digital-analog circuit implementation based on the long-range Ising interactions. In Sec. III , we introduce a heuristic for searching problem-specific hyperparameters of the ion native ansatz that ameliorate the optimization in VQA. Sec. IV shows the numerical results, where we benchmark the proposed heuristic on random instances of the SK model and investigate how the choice of hyperparameters affects the ion-based QAOA performance. Sec. V summarizes and discusses the results. II Ion native quantum ansatz II.1 Global ion interaction Trapped ion-based quantum computers are mostly known for their fully connected architecture, which allows executing entangling gates between any pair of ions in a trap [ 59 , 83 ] . The interaction between ions is turned by the excitation of collective ion oscillation modes (i.e. phonons) at specific frequencies \omega_{m} under laser radiation. If, however, the driving frequency is tuned far from the phonon frequencies (dispersive regime), the phonons are only virtually excited, inducing the evolution under the effective Hamiltonian [ 92 , 50 ] (for details, see Appendix A ), H_{\mathrm{I}}=\sum_{i<j}J_{ij}X_{i}X_{j}, (1) with X_{j} being Pauli X operator applied to the j -th qubit. The coupling coefficients in Eq. ( 1 ) are given by J_{ij}=\Omega_{i}\Omega_{j}\sum_{m}\frac{\eta_{i}^{m}\eta_{j}^{m}\omega_{m}}{\mu^{2}-\omega_{m}^{2}}\equiv\Omega_{i}\Omega_{j}C_{ij}. (2) Here, the matrix C_{ij} depends on the spectrum of phonon frequences \omega_{m} , Lamb-Dicke parameters \eta_{j}^{m} which quantify the displacement of the j -th ion in the m -th phonon mode, and laser detuning \mu from the carrier frequency. The matrix C_{ij} thus can be precalculated and kept fixed (for details, see Appendix B ). A characteristic feature

Read the complete text