We propose a framework to construct a real-space spin model based on the inverse Hamiltonian design. The method provides an efficient way of realizing unconventional topological spin textures by optimizing the interaction parameters. In order to demonstrate its usefulness, we show that the tuning of the long-range exchange interactions can give rise to a square skyrmion crystal even without factors that have been previously identified as prerequisites for its stabilization, such as a Dzyaloshinskii-Moriya interaction, a multispin interaction, and bond-dependent magnetic anisotropy. Moreover, we elucidate the essence for the emergence of the square skyrmion crystal by classifying the parameter sets we obtain by the method. Since the present framework by adopting machine learning techniques can be universally applied to any magnetic system irrespective of lattice structures, it serves as the efficient construction of an effective spin model and the understanding of the stabilization mechanisms for unconventional topological spin orders.

Close

PrevNext

View figure in article

• Exchange interaction
• Magnetic vortices
• Skyrmions
• Heisenberg model
• Numerical techniques