A reconfigurable, analog system for efficient stochastic biological computation

Abstract

Motivated by the many stochastic processes found in biology that allow for ultra-efficient computing, this paper explores circuit implementations for stochastic computation in hardware. Several novel contributions are presented in this paper, namely a dynamically controllable system of random number generators that produces Bernoulli random variables, exponentially distributed random variables, and allows for random variables of an arbitrary distribution to be generated. This system is implemented on a reconfigurable analog chipset allowing for the first time ever a hardware stochastic process with a user input to control the probability distribution. The utility of this system is demonstrated by implementing the well-known Gillespie algorithm for simulating an arbitrary biological system trajectory of sufficiently small molecules where over a 127times performance improvement over current software approaches is shown.