Optimizing Quantum Annealing Schedules with Monte Carlo Tree Search enhanced with neural networks

TL;DR

This paper introduces QuantumZero, a neural network-enhanced Monte Carlo Tree Search algorithm that automates the optimization of quantum annealing schedules, significantly improving efficiency in solving 3-SAT problems.

Contribution

It presents a novel hybrid quantum-classical method combining MCTS and neural networks to optimize annealing schedules, outperforming existing reinforcement learning approaches.

Findings

QZero effectively discovers annealing schedules for 3-SAT problems.

Neural network transfer learning boosts QZero's performance.

MCTS and QZero outperform other RL algorithms in schedule design.

Abstract

Quantum annealing is a practical approach to approximately implement the adiabatic quantum computational model under a real-world setting. The goal of an adiabatic algorithm is to prepare the ground state of a problem-encoded Hamiltonian at the end of an annealing path. This is typically achieved by driving the dynamical evolution of a quantum system slowly to enforce adiabaticity. Properly optimized annealing schedules often significantly accelerate the computational process. Inspired by the recent success of deep reinforcement learning such as DeepMind's AlphaZero, we propose a Monte Carlo Tree Search (MCTS) algorithm and its enhanced version boosted with neural networks, which we name QuantumZero (QZero), to automate the design of annealing schedules in a hybrid quantum-classical framework. Both the MCTS and QZero algorithms perform remarkably well in discovering effective annealing schedules even when the annealing time is short for the 3-SAT examples we consider in this study. Furthermore, the flexibility of neural networks allows us to apply transfer-learning techniques to boost QZero's performance. We demonstrate in benchmark studies, that MCTS and QZero perform more efficiently than other reinforcement learning algorithms in designing annealing schedules.