Reinforcement Learning-Assisted Shot Assignment for Improved Convergence in Variational Quantum Eigensolver

The Variational Quantum Eigensolver (VQE) holds promise for solving quantum chemistry problems, yet its convergence can be hindered by measurement errors, especially under a limited measurement budget. This work introduces reinforcement learning-assisted shot assignment strategies to mitigate measurement errors and improve VQE convergence. By estimating the standard deviation of measurements, two shot assignment strategies are devised for scenarios of overallocated and underallocated shots [1]. The reinforcement learning approach further optimizes shot allocation, demonstrated through numerical experiments on H2, HeH+ and LiH molecules, showcasing enhanced convergence and reduced shot count. This advancement contributes to making VQE a more practical tool for quantum chemistry applications on quantum computers.

[1] Zhu, Linghua, et al. arXiv:2307.06504 (2023).