Noise characterization and error mitigation on IBM Heron processors: Part I

Quantum error mitigation has recently been shown to produce accurate expectation values on IBM's fixed-frequency Eagle processor, at scales beyond brute-force classical computation. These methods often rely on the characterization and manipulation of noise in the device to effectively undo its effect on expectation values. However, the success of these methods also crucially relies on access to a representative model of the device noise. In these talks, we will present our implementation of noise learning and mitigation techniques on tunable-coupling Heron processors with two-qubit gate error rates approaching 0.1%. Our results highlight a powerful computational tool for the exploration of near-term quantum applications.

Part 1 will detail characterization and stabilization of noise in our Heron processors.