This study evaluates the performance of a Quantum Genetic Algorithm (QGA) executed on quantum devices. The algorithm was implemented in MATLAB and Python and deployed on the Amazon Web Services (AWS) platform. The QGA was utilized to optimize a set of continuous single-variable functions. The implementation employed the Hadamard quantum gate to initialize the population and the Ry rotation gate for mutation and crossover. The findings revealed significant differences in execution time and costs were observed between the two implementations, undescoring the performance of quantum devices available on AWS. The results demonstrate that the QGA can achieve optimal solutions in a few generations, suggesting its potential for efficiently solving complex problems. However, the costs and availability of quantum devices remain restrictive. This work exemplifies the potential of leveraging AWS cloud-based quantum computing platforms for the research and development of quantum algorithms.
We develop a new class of genetic algorithm that computationally determines efficient pulse sequences to implement a quantum gate U in a three-qubit system. The method is shown to be quite general, and the same algorithm can be used to derive efficient sequences for a variety of target matrices. We demonstrate this by implementing the inversion-on-equality gate efficiently when the spin-spin coupling constants
J12=J23=J and
J13=0. We also propose new pulse sequences to implement the Parity gate and Fanout gate, which are about 50% more efficient than the previous best efforts. Moreover, these sequences are shown to require significantly less RF power for their implementation. The proposed algorithm introduces several new features in the conventional genetic algorithm framework. We use matrices instead of linear chains, and the columns of these matrices have a well defined hierarchy. The algorithm is a genetic algorithm coupled to a fast local optimizer, and is hence a hybrid GA. It shows fast convergence, and running on a MATLAB platform takes about 20 minutes on a standard personal computer to derive efficient pulse sequences for any target 8X8 matrix
U.