Genetic optimization of ansatz expressibility for enhanced variational quantum algorithm performance
Manish Mallapur, Ronit Raj, Ankur Raina · 2025
Read Genetic optimization of ansatz expressibility for enhanced variational quantum algorithm performance by Manish Mallapur, Ronit Raj, Ankur Raina (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-2509-05804.txt.
Introduction
Choosing the right circuit layout for a variational quantum algorithm is the problem addressed by Manish Mallapur and two co-authors in 2025. These algorithms run a tunable quantum circuit and adjust it classically, serving chemistry, optimization, and machine learning tasks. A circuit must reach the needed states yet stay shallow enough to train on today's noisy machines.
Their answer borrows from evolution: candidate circuits mutate and compete, scored by expressibility, a measure of how widely a circuit can reach across possible states. The evolved circuits score high at every target depth while keeping few parameters and shallow depth. Performance matches traditional hand-designed layouts. Because the search ignores the specific problem, one evolved design can be reused across many applications.
Catalog record
| CREATOR | Manish Mallapur, Ronit Raj, Ankur Raina |
|---|---|
| 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/2509-05804
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Genetic optimization of ansatz expressibility for enhanced variational quantum algorithm performance Preprint: APS/123-QED Manish Mallapur, Ronit Raj, Ankur Raina Abstract Variational quantum algorithms have emerged as a leading paradigm that extracts practical computation from near-term intermediate-scale quantum devices, enabling advances in quantum chemistry simulations, combinatorial optimization, and quantum machine learning. However, the performance of Variational Quantum Algorithms is highly sensitive to the design of the ansatze. To be effective, ansatze must be expressive enough to capture target states but shallow enough to be trainable. We propose a genetic algorithm-inspired framework for designing ansatze that achieve high expressibility while maintaining shallow depth and low parameter count. Our approach evolves ansatze through mutation and selection based on an expressibility metric. The circuit generated by our framework consistently demonstrates high expressibility at any target depth and performs comparably to traditional ansatz design approaches while showing minimal to no signs of barren plateau issues. This work presents a general, scalable solution for ansatz design, producing expressive, low-depth circuits that need to be designed only once and can serve a wide range of applications. I Introduction Variational quantum algorithms Cerezo et al. 2021 stand at the forefront of quantum computing’s transition from theoretical promises to practical applications. In the noisy intermediate-scale quantum (NISQ) era Preskill 2018 ; Bharti et al. 2022 , where quantum processors are characterized by their limited qubit counts and susceptibility to noise, variational quantum algorithms (VQAs) emerge as the leading algorithmic framework Bergholm et al. 2018 for practical quantum computing applications. The primary advantage of VQAs is their ability to run shallow-depth circuits, which is crucial in NISQ devices, given their low coherence times and noise accumulation with an increase in circuit depth. The hybrid nature of this algorithm delegates the optimization process to classical computers while using the quantum computer for state preparations and state measurements. VQAs have demonstrated their practical utility across a range of applications that align with the capabilities of NISQ devices. For example, the QAOA Farhi et al. 2014 has been employed to solve combinatorial optimization, including its similar variants like the Max-Cut problem Wang et al. 2018 and other constraint satisfaction problems Lin and Zhu 2016 , error correction Johnson et al. 2017 ; Bhattacharyya and Raina 2023 , and for solving linear systems of equations Harrow et al. 2009 . Additionally, Variational Quantum Eigen solver (VQE) Peruzzo et al. 2014 is a prominent algorithm for finding ground state energies of molecular systems. However, the choice of ansatz determines the performance of VQAs Hubregtsen et al. 2021 . This impact emerges from an ansatz’s ability to explore the appropriate solution space, which decides how close the algorithm can get to approximating the true solution to the given problem. The performance of the ansatz depends on balancing the competing requirements. On one hand, the expressibility of an ansatz is important to represent the complex and wide range of states for accurate solutions. The trainability of the ansatz is also crucial for the algorithm’s convergence. An ansatz with a huge number of quantum gates is supposedly highly expressible, but has poor trainability due to increased depth and parameter count. Several works have explored automated designs of ansatze to improve the performance of VQAs. These quantum architecture search methods aim to optimize variational circuit design, often through evolutionary strategies like genetic algorithms and gate-level optimization Williams and Gray 1998 ; Lamata et al. 2018 ; Huang et al. 2022 . Reinforcement learning based quantum architecture search explores these quantum architectures through policy-driven agents Kuo et al. 2021 . Predictor-based QAS Zhang et al. 2021 ; He et al. 2023 and SuperCircuit frameworks Du et al. 2022 ; Wang et al. 2022 have gained attention due to improved scalability factors, but they still pose a high classical computational cost. The challenge is to find a balance between expressibility and trainability of the ansatz, keeping the scalability in mind. Figure 1: There is a trade-off between circuit complexity and expressibility in variational quantum circuits. The diagram shows three important areas in the expressibility-complexity landscape. Region C (red) shows a small search space because the circuit depth is not deep enough, which makes it hard to get close to the target states. The green area in Region B shows the best expressibility zone that our genetic algorithm (GA)-generated ansatz can find. It does this by choosing the best gate configurations at the right depths to balance expressibility and trainability. At high levels of expressibility and complexity, Region A (gray) shows the entire Hilbert space that can be reached. Circuits suffer from the barren plateau phenomenon, which makes them hard to train. Our depth-aware GA framework moves around these areas in real time to create resource-efficient ansatze that are highly expressive and converge quickly across a wide range of problem sizes. Our work introduces a genetic algorithm-based ansatz design framework Creevey et al. 2023 ; Sünkel et al. 2023 that addresses the limitations of current ansatz design schemes, which are often fixed and greedy in nature. Our method leverages the global search capabilities of evolutionary algorithms to explore the vast circuit design space. We guide the search towards circuits that maximize expressibility while operating on a fixed depth. In other words, we use a classical algorithm as a pre-processing step before using variational principles on an NISQ computer. We introduce depth-aware circuit optimization strategies where, at each depth, the framework identifies optimal configurations that balance gate efficiencies and expressibility. This systematic approach provides us a practical way to navigate the expressibility-trainability Holmes et al. 2022a tradeoff, which is inherent in variational quantum algorithms. To validate the effectiveness of our framework, we apply the resulting ansatze to one of the most prominent VQA applications: finding the ground state energy of complex quantum systems using VQE. We benchmark their performance on a range of molecular Hamiltonians ( IV.2 ) as well as spin Hamiltonians ( IV.1.6 ), demonstrating practical utility for both quantum chemistry and condensed matter problems. Through this work, we demonstrate the potential of genetic algorithms as a general-purpose ansatz design tool capable of generating resource-efficient, highly expressible circuits across various problem sizes and depths. This paper is organized as follows. In Section II , we briefly review the fundamentals of the VQE, including the critical tradeoff between ansatz expressibility and trainability. Section II.3 talks about the related works and established ansatz designs like UCCSD and ADAPT-VQE, which also serve as benchmarks. We then provide an overview of genetic algorithms in section II.3 and the evolutionary optimization techniques that form the basis of our method. Our novel genetic algorithm-based framework for designing highly expressible quantum circuits is detailed in Section III . In Section IV , we present a comprehensive analysis of our framework, where we evaluate the impact of its various parameters, benchmarking the generated ansatze on various Hamiltonians and assessing their trainability. Finally, we discuss the implications of our findings in Section IV.5 and present our conclusion in Section V . II Variational Quantum Eigensolver (VQE) The variational principle in quantum mechanics states that for any normalized trial wavefunction, the expectation value of the Hamiltonian operator will always give a value greater than the true ground state energy of the