Probabilistic computing with spins

Conventional computers operate deterministically using bits, which are strings of 0’s and 1’s that represent information in binary code. There are many classes of problems that are difficult to address using such computation, however, such as optimization and sample, which is leading to a growing interest in alternative computing schemes. Probabilistic computing is one such example, which aims to exploit probabilistic bits (p-bits) that fluctuate in time between 0 and 1, and interact with other p-bits using principles inspired by neural networks. Shunsuke Fukami and colleagues now demonstrate probabilistic computing using stochastic magnetic tunnel junctions, showing integer factorization, which is an illustrative example of an optimization problem. This scalable spintronics platform could be a promising new hardware approach to the difficult problems of optimization and sampling.