Image Credit: Yurchanka Siarhei/Shutterstock.com
While much research is already underway in developing quantum computing, it is likely to still be decades before the technology is scalable and reliable enough to enter into the landscape of everyday technology. To help us attain the advantages of quantum computing sooner, some scientists have focused on creating the "probabilistic computer”, which is predicted to sit between the modern conventional computer and the futuristic quantum computer.
In a paper published in the journal Nature this September, a team of researchers explains how they have developed the first hardware for the probabilistic computer, which would help to solve problems that are outside of the realm of the capabilities of the classical computer.
Getting a Step Closer to the Quantum Computer by Developing the ‘Probabilistic Computer’
An international team of researchers from Japan and the US joined forces to develop a device that will serve as the cornerstone of new probabilistic computers. These machines will be able to solve problems that the world’s classical computers cannot, which is likely to have a significant impact on various sectors, such as encryption and cybersecurity, financial services, pharmaceutical research, data analysis and supply chain logistics, and more.
The team was motivated to create a system that could process data more efficiently than classical computers, but that wouldn’t require the same extreme settings as quantum computers, such as the low temperatures that are necessary for its qubits to work at. They devised that they could build a machine that works on p-bits instead of qubits, therefore eliminating the need for a cold environment, which requires a lot of power.
In classical computers, information is stored in binary, in the form of either ones or zeros. In quantum computers, on the other hand, information is stored as both ones and zeros at the same time through the use of qubits in place of bits. However, in 2017, a Purdue research group proposed that the third kind of information processing is possible. They put forward the idea of a probabilistic computer that stores information either as a zero or a one, and can rapidly fluctuate between the two. It does this by using p-bits in place of bits or qubits.
The researchers of the current study developed the ideas put forward in 2017 to create the first hardware for use in a probabilistic computer that would be able to accomplish tasks out of the skillset of a classical computer, bridging the gap between classical computing and quantum computing.
Altering MRAM to Support p-bits
Engineers at Purdue University in the USA and Tohoku University in Japan modified a version of magnetoresistive random-access memory, known as MRAM, which are used by some classical computers to store information. The device creates specific states of resistance that correspond to a one or a zero by altering the orientation of magnets.
By altering this system, the American/Japanese team made a device that better facilitated the fluctuation of p-bits by making the system intentionally unstable. The team then combined this altered device with a transistor, resulting in a three-terminal unit that the researchers had control over determining the fluctuations. They were able to construct and interconnect eight p-bits to build the first probabilistic computer.
First Probabilistic Hardware Created
The team at Purdue University and Tohoku University were successful in constructing the first piece of hardware to be used in a probabilistic computer. Their research demonstrated how p-bits can be created and interconnected in a device. They also demonstrated that the p-bits were able to perform calculations that would usually require a quantum computer to solve.
Evidence presented in September’s paper describes how the circuit they created was successful in solving the “quantum” problem of factoring numbers into smaller numbers, known as integer factorization. While classical computers can manage this problem, it takes a lot of energy and memory space. The probabilistic computer proved that it could calculate this problem by using significantly less energy and space.
The research is promising, however, to create a probabilistic computer capable of solving bigger problems, such that pose pressing issues to the world’s industries, hundreds of p-bits would need to be connected. Fortunately, researchers believe that this isn’t too far off, and we could see probabilistic computers being available for use in the near future.
Sources and Further Reading
- Borders, W., Pervaiz, A., Fukami, S., Camsari, K., Ohno, H. and Datta, S. (2019). Integer factorization using stochastic magnetic tunnel junctions. Nature, 573(7774), pp.390-393. https://www.nature.com/articles/s41586-019-1557-9