Identifying Quantum Phase Transitions with Minimal Empirical Knowledge by Machine Learning

In this work, we proposed a novel approach for identifying quantum phase transitions in one-dimensional quantum many-body systems using Autoencoders (AE), an unsupervised machine learning technique, with minimal empirical knowledge. The training of the AEs is done with data across entire range of the driving parameter and thus no prior knowledge of the phase diagram is required. With this method, we successfully detect the phase transitions in a wide range of models with multiple phase transitions of different types, including the topological and the Berezinskii-Kosterlitz-Thouless ones by tracking the changes in the reconstruction loss of the AE. The learned representation of the AE has potential utility in elucidating the physical phenomena underlying different quantum phases. Our methodology demonstrates an effective and promising new approach to studying quantum phase transitions.