Application of machine learning to investigate the stabilization of skyrmions on a triangular Kondo lattice

Skyrmions are key contenders for future spintronic devices because of the topological protection caused by the whirling spin texture. The ideal candidates to realize skyrmions are materials with a band of conduction electrons coupled to the net moment of the localized f-electrons. Such materials are modelled via Kondo lattice model (KLM) which can be approximated to RKKY model in the weak-coupling regime. Away from this regime, significant four spin interactions that are non-analytic functions of the coupling constant emerge in the low-energy model rendering a perturbative treatment inapplicable. We use supervised machine learning to obtain these effective interactions in both real space and momentum space. The resultant low-energy effective Hamiltonian is used to study multiple-Q magnetic orderings on a triangular Kondo Lattice Model. We demonstrate the presence of the skyrmion phase even in absence of spin anisotropy, which is not possible within the pure RKKY model.