Multicanonical sampling of vortex states in magnetic nanoelements

An auto-adaptive multicanonical Monte Carlo (MMC) simulation method is suggested and tested on a single-vortex model of magnetic nanoelement. Simulation process consisting of nonequilibrium and equilibrium stages that circumvents ergodicity sampling problems which stem from a potential barrier standing between the vortex and counter-vortex states is proposed. The method is formulated by the means of an effective Hamiltonian with additional term proportional to the overlap of given configuration and bistable ground-state vortex configuration. The self-organized neural network is used to construct the synopsis of the vortex reversal process.