Quantum Phases Classified From Subsystems of Four Sites or Less

Researchers have developed a new approach to identifying quantum phases that bypasses a longstanding limitation in the field: the need to measure properties across an entire quantum system. The work, led by Mehran Khosrojerdi of the University of Florence and Sougato Bose of University College London, demonstrates a supervised learning framework capable of classifying quantum phases from small subsystems. This data efficiency is achieved through a quantum kernel constructed from the reduced density matrices of these subsystems, allowing for experimental estimation without full system access. The findings demonstrate that local reduced density matrices preserve vital signatures of global topological phases, offering a practical route to characterize…