Probing interacting topological matter with neural-network quantum states

Recently, many-body quantum matter with topological orders has been efficiently represented by neural-network quantum states in the form of restricted Boltzmann machines. We extend its application to 2D topological materials with interactions, including generalized Kitaev honeycomb model in a magnetic field and frustrated bond-random Heisenberg antiferromagnets on a square lattice. The stability of the emergent spin liquid phases is tested by dynamical spin structure factor, flux excitations as well as bipartite fluctuations, an indicator for many-body entanglement.