Reinforcement learning-guided optimization of critical current in high-temperature superconductors

High-temperature superconductors (HTS) are essential for next-generation energy and quantum technologies, but their performance is often limited by the critical current density (Jc), which is strongly influenced by microstructural defects. Optimizing Jc through defect engineering is challenging due to the complex interplay of defect type, density, and spatial correlation. Here we present an integrated workflow that combines reinforcement learning (RL) with time-dependent Ginzburg–Landau (TDGL) simulations to autonomously optimize defect landscapes. In our framework, TDGL simulations generate current–voltage characteristics to evaluate Jc, which serves as the reward signal guiding the RL agent to iteratively refine defect distributions. We find that the agent discovers optimal defect densities and correlations in two-dimensional thin-film geometries, enhancing vortex pinning and improving Jc beyond baseline designs. This RL-driven approach provides a scalable strategy for defect manipulation, with implications for advancing HTS applications in fusion magnets, particle accelerators, and other high-field technologies.