Benchmarking and fidelity response theory of high-fidelity Rydberg entangling gates

TL;DR

This paper demonstrates a high-fidelity Rydberg CZ gate with 99.71% fidelity, develops a fidelity response theory to predict errors from laser noise, and provides scaling laws to guide future improvements in neutral atom quantum gates.

Contribution

It introduces a time-optimal Rydberg CZ gate, benchmarks its fidelity, and develops a fidelity response theory to predict and optimize gate errors due to laser noise.

Findings

Achieved a fidelity of 0.9971(5) for the Rydberg CZ gate.

Developed a fidelity response theory to predict infidelity from laser noise.

Predicted that a fidelity exceeding 0.999 is achievable with current technology improvements.

Abstract

The fidelity of entangling operations is a key figure of merit in quantum information processing, especially in the context of quantum error correction. High-fidelity entangling gates in neutral atoms have seen remarkable advancement recently. A full understanding of error sources and their respective contributions to gate infidelity will enable the prediction of fundamental limits on quantum gates in neutral atom platforms with realistic experimental constraints. In this work, we implement the time-optimal Rydberg CZ gate, design a circuit to benchmark its fidelity, and achieve a fidelity, averaged over symmetric input states, of 0.9971(5), downward-corrected for leakage error, which together with our recent work forms a new state-of-the-art for neutral atoms. The remaining infidelity is explained by an ab initio error model, consistent with our experimental results over a range of gate speeds, with varying contributions from different error sources. Further, we develop a fidelity response theory to efficiently predict infidelity from laser noise with non-trivial power spectral densities and derive scaling laws of infidelity with gate speed. Besides its capability of predicting gate fidelity, we also utilize the fidelity response theory to compare and optimize gate protocols, to learn laser frequency noise, and to study the noise response for quantum simulation tasks. Finally, we predict that a CZ gate fidelity of ${\gtrsim} 0.999$ is feasible with realistic experimental upgrades.