Revealing the Phase Diagram of Kitaev Materials by Machine Learning

Kitaev materials are promising materials for hosting quantum spin liquids and investigating the interplay of topological and symmetry-breaking phases. We use an unsupervised and interpretable machine-learning method, the tensorial-kernel support vector machine (TK-SVM), to study the phase diagram of candidate models for Kitaev materials. Our machine learns the global classical phase diagram for the honeycomb Kitaev-Γ model in a magnetic field and the associated order parameters, identifying several distinct spin liquids and unconventional orders. We find that the emergence of orders in the Kitaev-Γ model can be consistently explained by competition and cooperation between two spin liquids. We then apply our TK-SVM method to the Heisenberg-Kitaev-Γ model and discuss the effects of the Γ' and the J3 interaction.