Reliably operating noisy quantum computers

Summary

The overall goal of the project is to develop practical methods to be able to reliably run useful applications on near-term quantum computers. This requires identifying and overcoming the ubiquitous errors that currently limit quantum computing capabilities. Traditional methods of quantifying errors in quantum computers fail to predict how errors affect the output of a quantum computation because the exact impact depends upon the exact form of the noise, additional errors arising from interactions between control mechanisms (e.g., crosstalk) and idle qubits, and how the gates are translated and scheduled into temporal pulses. One promising way to account for global errors is to define a parallel quantum instruction (PQI) to be a set of quantum operations executed in a fixed temporal order, including all idle gates for qubits that are not explicitly targeted by any quantum operation. In this project we develop a general method for reconstructing global noise during a cycle of parallel quantum gates and a framework for mitigating and/or extrapolating errors, leading to an experimental demonstration of their effectiveness. This will enable near-term quantum computers to be used to accurately simulate quantum systems and to determine the accuracy of the simulations.

Figure 1. Schematic circuit implementation of the experimental cycle benchmarking protocol for characterizing a noisy parallel quantum instruction G (red). The green gates are used to probe how closely G maps different input states to the desired output state. Random gates (blue) are used to engineer a simple noise model to make the characterization protocol accurate and efficient.