This article discusses the use of machine learning force fields to perform large-scale DFT calculations for tMoTe2 and tWSe2 at various twist angles. The results show that the twist-angle dependence of the moiré potential can lead to changes in the Chern numbers and the emergence of multiple flat bands with Chern numbers of 1. This opens up new possibilities for designing topological and correlated states in TMD homobilayers.
