Evolutionary-based searching method for quantum circuit architecture

Quantum architecture search (QAS) is desired to construct a powerful and general QAS platform that can significantly accelerate quantum advantages in error-prone and depth-limited quantum circuits in today’s Noisy Intermediate-Scale Quantum era. In this paper, we propose an evolutionary-based quantum architecture search (EQAS) scheme for the optimal layout to balance the higher expressive power and the trainable ability. In our EQAS, each layout of quantum circuits, i.e., quantum circuit architecture (QCA), is first encoded into a binary string, also called genes. Next, an algorithm is designed to remove the redundant parameters in QCA according to the eigenvalues of the corresponding quantum Fisher information matrix (QFIM). Later, the fitness values of the QCAs are calculated by evaluating the performance of QCAs and are processed with softmax function so that the sum of all fitness values is to 1, and it is used as the probabilities to prepare the parent generation by the Roulette Wheel selection strategy. After that, the mutation and crossover are applied to obtain the next generation. EQAS is verified by the classification task in quantum machine learning over three datasets. The results show that the proposed EQAS can search for the optimal QCA with fewer parameterized gate. And higher accuracies are also obtained by using the proposed EQAS for the classification tasks over the three datasets. Overall, EQAS presents a promising way in quantum architecture search, by taking advantage of QFIM and the evolutionary algorithm.