Southwest Quantum Information and Technology

Predicting the success rates of quantum circuits with artificial neural networks

Presenting Author: Daniel Hothem, Sandia National Laboratories
Contributing Author(s): Tommie Catanach, Timothy Proctor, Kevin Young

Current quantum computers are noisy and error-prone. As devices grow in size, we need scalable methods for predicting the performance of a given device. In this work, we explore the potential of neural networks to play this role. We demonstrate a convolutional neural network’s ability to predict circuit success rates under both simulated and experimental error models; in each case outperforming non-neural network models that are based on per gate error rates. We also experimentally investigate how a convolutional network’s ability to predict success rates varies as a function of dataset size and measurement accuracy, achieving exponential improvements in performance on simulated data as measurement precision increases. Finally, we present proof-of-concept work detailing a convolutional neural network’s ability to handle non-Markovian noise on wide circuits (greater than 30 qubits), a regime in which traditional techniques become inaccurate and highly expensive. This work was supported by the LDRD program at Sandia National Labs. Sandia National Labs is a multimission laboratory managed and operated by NTESS, LLC, a wholly owned subsidiary of Honeywell International Inc., for DOE’s NNSA under contract DE-NA0003525.

(Session 5 : Thursday from 5:00 pm - 7:00 pm)