Evaluating Mutation Techniques in Genetic Algorithm-Based Quantum Circuit Synthesis

Evaluating Mutation Techniques in Genetic Algorithm-Based Quantum Circuit Synthesis

Abstract:

Quantum computing has the potential to solve complex problems that are intractable for classical computers, while serving as a cornerstone of next-generation systems offering extreme computational power. This capability arises from the unique properties of qubits and quantum parallelism, allowing quantum computers to perform certain calculations much faster than classical counterparts.
The optimization of quantum circuits is essential for advancing quantum computing, particularly for noisy intermediate-scale quantum (NISQ) devices. These devices face significant challenges due to their limited number of qubits and high error rates, making efficient circuit synthesis critical. Genetic algorithms (GAs) have emerged as a promising solution for optimizing quantum circuits by automating a task that is otherwise manually solved in an inefficient manner.
This thesis investigates the impact of various mutation strategies within a GA frame- work for quantum circuit synthesis. Mutations interact at the most fundamental level of a circuit and can significantly influence overall performance. Collecting data on how these mutations transform circuits and determining which strategies are most efficient is a key step in developing a robust GA optimizer for quantum synthesis.
The experiments conducted in this research employed a fitness function primarily based on fidelity, while also considering circuit depth and the number of T operations. The experiments focused on optimizing four to six qubit circuits with extensive hyperparameter testing to identify optimal solutions for practical quantum computing. The results indicate that the combination of delete and swap strategies, without employing change or add strategies, provided the best performance under the given constraints.

Author:

Tom Bintener

Advisors:

Michael Kölle, Maximilian Zorn, Thomas Gabor, Claudia Linnhoff-Popien

Student Thesis | Published December 2024 | Copyright © QAR-Lab
Direct Inquiries to this work to the Advisors