Probabilistic Computing on Stochastic Magnetic Tunnel Junctions
摘要
Abstract
A solution to the traveling salesman problem was experimentally and theoretically investigated using a stochastic magnetic tunnel junction emulator. The relationship between theoretical and experimental results was analyzed using the Ising model. A series of experiments were conducted to determine the system parameters at which the solution to the problem was most likely. General dependences of the reliability of the obtained solution on the hyperparameters were demonstrated.