Machine learning phases of an Abelian gauge theory

Abstract The phase transition of the two-dimensional U(1) quantum link model on triangular lattice is investigated by employing a supervised neural network (NN) consisting only one input layer, hidden layer two neurons, and output layer. No information studied used when NN training conducted. Instead, artificially made configurations are considered as set. Interestingly, obtained not estimates critical point accurately but also uncovers physics correctly. results presented here imply that NN, which has very simple architecture trained without any from model, can identify targeted structure with high precision.