Error Mitigation and Correction

Zero Noise Extrapolation

Zero Noise Extrapolation, or ZNE, is a numerical integration technique that can be used to reduce the bias in estimating expectation values on a device using post-processing. By deliberately increasing the noise in the system, a function can be fit to the noise/expectation value relationship and traced back to obtain an improved estimate of the expectation value at zero noise. This technique was first used in a quantum computing context by IBM on a superconducting platform, but our focus has been to study the best methods for implementing this technique on a trapped-ion architecture. In https://arxiv.org/abs/2307.07027 we discuss in depth different techniques to scale noise in such a system, and analyze the performance of error mitigation when integrated into a variational optimization algorithm such as VQE. We are now focused on combining ZNE with additional error mitigation techniques such as Randomized Compilation to further improve performance.

Error Correction

The group also works on bridging the gap between error mitigation and error correction. We study how to incorporate methods such as Randomized Compilation and Zero Noise Extrapolation into the hierarchy of error correcting codes, such as with the Solovay-Kitaev theorem.

Page updated

Google Sites

Report abuse