Today, a significant portion of energy is utilized to process and store data, as well as to operate the necessary terminal devices and equipment. The amount of energy used for these things is expected to rise even further in the future. Innovative ideas use energy-saving techniques to address this issue, such as neuromorphic computing. This concept, known as Brownian reservoir computing, has now been achieved in a cooperative study between experimental and theoretical physicists at Johannes Gutenberg University Mainz (JGU) with the assistance of an ERC Synergy Grant. Additionally, the outcomes were recently highlighted as an editors’ highlight in the Nature Communications scientific journal’s Devices section.
Ambient thermal energy is used in Brownian computation.
A combination of two unorthodox computing techniques is known as Brownian reservoir computing. Brownian computing makes use of the fact that computer processes usually take place at room temperature, so it is possible to use the local thermal energy and reduce the amount of electricity used. The name of this computing technology, Brownian motion, indicates how the thermal energy employed in the computing machine is essentially the random motion of particles.
The use of reservoir computing allows for incredibly effective data processing.
Reservoir computing is a very resource-efficient method of processing data because it makes use of the intricate reaction of a physical system to outside stimuli. The system itself performs the majority of the computation; therefore, extra energy is not needed. Furthermore, since the solid-state system does not need to be modified to meet certain requirements, this sort of reservoir computer can be easily adjusted to carry out a variety of activities.
A prototype that combines these two computing techniques has now been created by a team led by Professor Mathias Kläui of the Institute of Physics at Mainz University, with assistance from Professor Johan Mentink of Radboud University Nijmegen in the Netherlands. This prototype can carry out Boolean logic operations, which can be utilized as benchmark tests for reservoir computing validation.
In this case, metallic thin sheets with magnetic symposia make up the solid-state system. These magnetic vortices can be accelerated by electrical currents and behave like particles. Skyrmions’ behavior is governed by both their intrinsic Brownian motion and the applied current. As the system is immediately reset after each operation and set up for the next computation, the Brownian motion of swarms can lead to greatly enhanced energy savings.
In Mainz, the first prototype was created.
Although there have been many theoretical ideas for skyrmion-based reservoir computing in recent years, the Mainz researchers were only able to combine these ideas with the idea of Brownian computing to create the first working prototype. Experimental physicist Klaus Raab remarked, “The prototype is easy to construct from a lithographic point of view and can theoretically be decreased to a scale of just nanometers.” Theoretical scientist Maarten Brems stressed, “We attribute our achievement to the strong collaboration between the experimental and theoretical physicists here at Mainz University.” Professor Mathias Kläui, the project’s coordinator, added: “We were able to work together with exceptional colleagues at the Department of Theoretical Physics in Nijmegen thanks to funding provided through a Synergy Grant from the European Research Council, and it was this collaboration that led to our success.” “Unconventional computing, an area that also receives substantial support here in Mainz through financing from the Carl Zeiss Foundation for the Emergent Algorithmic Intelligence Center, has a lot of potential, in my opinion.”
