Batch Optimization of Frequency-Modulated Pulses for Robust Two-Qubit Gates in Ion Chains
Uncertainty in control parameters limits the quality of operations and algorithms on a quantum computer. Typical methods for building operations that are robust to these errors still optimize for the ideal case. In our recent publication (Kang, M. et al., Phys. Rev. Applied 16, 024039, collaboration with the MIST Lab), we achieve robustness of two-qubit gates on trapped ion systems by optimizing average performance over a range of systematic errors. The key idea is using batch optimization, a technique originated from neural networks. We demonstrate significantly improved robustness to motional mode frequency offsets both in theory and experiment, compared to the typical robust method. Simulation results show that the advantage in robustness is more significant with larger number of ions and uncertainty in motional frequency, which makes our method a promising tool for large-scale experiments. This study inspires further development of machine-learning-inspired pulse optimization tools for robust control in any quantum computing platform.