Purdue University engineers are building a probabilistic computer that would not need new hardware or extreme cold to operate.
The team believes it could solve some of the problems a quantum computer would solve, such as calculating the best solution from a very large number of solutions. In one logistics example, that could be the best route for goods to travel to market.
Instead of qubits used in quantum computers, a probabilistic computer is made from p-bits to solve problems involving optimization. In 2019, researchers from Purdue and Tohoku University in Japan first demonstrated a probabilistic computer.
Supriyo Datta, who led the Purdue team, said p-bits are a “poor man’s qubit” and could be useful for a subset of problems solvable with qubits. Datta is a professor of electrical and computer engineering at Purdue.
The researchers see probabilistic computing as a step up from classical computing to reach quantum computing. To solve quantum problems would not necessarily mean going quantum. In a diagram Purdue provided, the probabilistic computer sits between a PC network and a quantum computer.
A probabilistic computer can store and use information in the form of zeros and ones at room temperature, while quantum machines typically operate at near absolute zero.
Also, a probabilistic computer could process multiple states of zeros and one at one time, but a p-bit would actually rapidly fluctuate between zero and one, while a qubit is actually a superposition of zero and one. On a chip, these fluctuations would be correlated between p-bits but entangled in qubits.
The Purdue team wants to tweak common memory technology to make it purposely unstable so that p-bits can fluctuate in devices called magnetic tunneling junctions.
The researchers have a patent through Purdue’s Office of Technology Commercialization and have used current silicon technology that is publicly available through AWS to emulate a probabilistic computer with thousands of p-bits. The patent, granted in the U.S. in 2020, is for part of a probabilistic computer, specifically a stochastic switching device with adjustable randomness.
Jeorg Appenzeller, also a professor of electrical and computer engineering at Purdue, said there’s not a final decision on the best implementation of a p-bit. “We’re showing what works so we can figure it out along the way,” he said in a statement.
The work has support from Semiconductor Research Corp and the National Science Foundation as well as the Defense Advanced Research Project Agency.
The research team has published several papers on their work, including modeling of how to make such a p-bit system work on a large scale with energy efficiency.