Stochastic Computing

Stochastic computing is at the forefront of computational approaches, utilizing randomness to efficiently process information. This groundbreaking method has unlocked new opportunities for tackling intricate computational challenges, particularly those that traditional computing struggles with, such as various NP problems. In this context, our research is centered on the development of innovative circuitry using non-volatile electronic devices for stochastic computing applications. Our goal is to address a spectrum of challenges, including solving problems like integer factorization, Boolean satisfiability, the traveling salesman problem, and protein structure prediction. Furthermore, our investigation delves into the relationship between device reliability and circuit performance, as well as the advancement of effective annealing techniques.

​

In this domain, our primary research revolves around tackling diverse challenges through the implementation of invertible circuits comprising P-bits, which are created using non-volatile electronic devices. These challenges, such as complex combinatorial optimization problems, typically fall under the category of NP problems, which are notoriously challenging for conventional computers to solve. Through the application of stochastic computing, we can effectively resolve these problems swiftly and with minimal energy consumption. As a result, our research emphasis lies in the development of more efficient reversible logic circuits within this sphere.

Furthermore, our research delves into the realm of random operations from a comprehensive and macroscopic viewpoint. Broadly, we evaluate the efficiency of our designed reversible circuit within the context of the entire system. Our analysis takes into account the influence of device instability on the overall circuit stability, and we conduct in-depth investigations into the algorithmic perspective of the entire circuit's annealing method.