In a development that could revolutionize the digital memory industry and reduce the high energy consumption of artificial intelligence technologies, a research team from the University of Edinburgh has developed a new mathematical framework based on “optimal control theory” to reduce the energy required to write data in future magnetic memories by up to thousands of times.

This innovation addresses the growing energy cost of data processing and storage. Large language models, scientific simulations, and digital recommendations rely on repeated retrieval and rewriting processes that consume vast amounts of electrical power across global data centers.

Led by Dr. Elton Santos from the Institute of Condensed Matter Physics and Complex Systems, the study, published in Advanced Materials, focuses on finding an accurate mathematical path to guide the natural motion of a magnet to flip the magnetic states representing digital “bits” (0s and 1s) instead of increasing the applied force.

This approach eliminates the need for applying constant and strong magnetic pulses that consume high energy and face challenges in scaling down with device miniaturization. It offers an advanced alternative to current memory technologies such as spin transfer torque magnetic random access memory (STT-MRAM) and orbital angular momentum (SOT-MRAM).

Computer simulations of three ultra-thin “van der Waals” magnetic materials (Fe₃GaTe₂, Fe₃GeTe₂ and CrSBr) showed that mathematically designed pulses can reverse magnetization in just 1 to 10 picoseconds, with magnetic fields ten times weaker than standard methods.

The switching consumed around 0.94 nanojoules of energy compared to 91.2 nanojoules for conventional methods, and the energy consumption could be reduced further to femtojoule levels in future by adjusting magnetic properties such as damping and magnetic anisotropy.

The researchers believe that bringing energy levels closer to the “Landauer limit”, which is the fundamental thermodynamic limit of the minimum amount of energy wasted when erasing data, opens up possibilities for applying this mathematics in other media.

This strategy is not limited to magnetic fields; it can be adapted for high-speed electrical currents and laser pulses, providing engineering flexibility to integrate light, electricity, and magnetism to develop a new generation of highly efficient rotational electronics.