AI has learned to determine phase transitions in complex physical models. This will simplify the study of spin models, according to a study by scientists from the University of Tokyo, published in the journal Nature Physics.
AI is changing not only the everyday aspects of people’s lives but also science. The main advantage of using such algorithms for scientific purposes is that they can be trained using pre-classified data (for example, images of handwritten letters) and used to classify a much wider range of data.
In a new work, physicists used AI to accelerate research on spin models that scientists use to study phase transitions. In a previous study, they created a neural network to track the phase transitions of matter (liquid, gas, or solid) in simple physical models.
Refining the AI, the researchers were able to use it in a similar way in complex physical models. The authors of the work note that the neural network can be used to study various conditions in which a phase transition occurs.
Testing showed that AI trained in one model and applied to another can reveal key similarities between different phases in different systems.