A Nanomagnetic Voltage-Tunable Correlation Generator between Two Random Bit Streams for Stochastic Computing (1907.07532v1)

Abstract: Graphical probabilistic circuit models of stochastic computing are more powerful than the predominant deep learning models, but also have more demanding requirements. For example, they require "programmable stochasticity", e.g. generating two random binary bit streams with tunable amount of correlation between the corresponding bits in the two streams. Electronic implementation of such a system would call for several components leaving a large footprint on a chip and dissipating excessive amount of energy. Here, we show an elegant implementation with just two dipole-coupled magneto-tunneling junctions (MTJ), with magnetostrictive soft layers, fabricated on a piezoelectric film. The resistance states of the two MTJs (high or low) encode the bits in the two streams. The first MTJ is driven to a random resistance state via a current or voltage generating spin transfer torque and/or voltage controlled magnetic anisotropy, while the second MTJ's resistance state is determined solely by dipole coupling with the first. The effect of dipole coupling can be varied with local strain applied to the second MTJ with a local voltage (~0.2 V) and that varies the correlation between the resistance states of the two MTJs and hence between the bits in the two streams (from 0% to 100%). This paradigm can be extended to arbitrary number of bit streams.