ZX-Diagram T-Count Reduction Based on Proximal Policy Optimization

In fault-tolerant quantum computing systems, the T gates consume more fault-tolerant resources. In this paper, we propose an automated optimization method based on the Proximal Policy Optimization (PPO) algorithm. We first translate quantum circuits into ZX-diagrams using ZX-calculus, transforming the circuit optimization problem into a graph rewriting task. Subsequently, the PPO agent, powered by graph neural networks (GNNs) to encode the complex graph structures of the ZX-diagrams, learns to predict efficient transformation paths by focusing on local node and edge features. The PPO algorithm ensures stable policy updates through its clipped objective function, which prevents excessively large and potentially detrimental changes during optimization. Experimental results confirm the method’s effectiveness, showing an average reduction in T-count of 3.12% compared to the baseline approach.