Improving quantum computer performance with machine learning

Excitement about the promise of quantum computers is tempered by the reality that the hardware remains exceptionally fragile and error-prone, forming a bottleneck in the development of novel applications. In this talk we will introduce the concept and experimental implementation of quantum control, providing a pathway to maximizing hardware performance in near term systems, and forming a complement to quantum error correction in future systems. We will describe the challenge of designing low error and robust quantum logic and present a series of experiments on cloud-accessible superconducting quantum computers, demonstrating how redefining the analog waveforms used to implement quantum logic gates can lead to improvements in gate error, resilience against fabrication variance, and resilience against temporal drifts. We will then present the first demonstration of Deep Reinforcement Learning (DRL) to autonomously design a Universal gateset showing superior performance to calibrated default gates. DRL-designed two-qubit cross-resonance gates exhibit ~2.5X improvements relative to standard gates, and obviate the need for additional compensating signals designed to mitigate crosstalk. We demonstrate drift-robust two-qubit gate performance at the level of ~99.5% fidelity (near T1 limits) up to 25 days from gate design with no recalibration, while default gates require recalibration every 12-24 hours. These experiments reveal a pathway to autonomously designing error-robust quantum logic at scale across complex systems with unknown couplings and Hamiltonian terms. We will complement my talk by showing that such machine-learning based automated routines can lead to dramatic algorithmic improvements on current mid-scale devices. We show up to 25X improvements in success probability of both deterministic algorithms, such as QFT and BV, and hybrid algorithms such as QAOA and VQE.