Brain-inspired computers have demonstrated the ability to solve complex mathematical equations previously associated with supercomputers. This development, detailed in a study published in Nature Machine Intelligence, was led by computational neuroscientists Brad Theilman and Brad Aimone of Sandia National Laboratories.
It was reported that these “neuromorphic” computers can efficiently solve partial differential equations (PDEs). PDEs are crucial for modeling various phenomena, including fluid dynamics, electromagnetic fields, and structural mechanics, which are used in weather forecasting, material stress analysis, and nuclear simulations.
The research suggests a path toward energy-efficient supercomputers. The National Nuclear Security Administration, which uses supercomputers for nuclear system simulations and other critical applications, could potentially reduce its energy consumption significantly through this technology.
The development involved a new algorithm based on a computational neuroscience model. This algorithm allows neuromorphic hardware to address mathematical problems that traditionally require substantial computing power from energy-intensive supercomputers.
The project received funding from the Department of Energy’s Office of Science through the Advanced Scientific Computing Research and Basic Energy Sciences programs, and the National Nuclear Security Administration’s Advanced Simulation and Computing program.
Beyond engineering applications, this research contributes to understanding how the human brain performs computations. The algorithm’s structure and behavior closely resemble cortical networks, suggesting a link between neuroscience and applied mathematics. This connection could eventually aid in understanding and treating neurological disorders.
The researchers believe this work marks a significant step in neuromorphic computing, fostering collaboration across mathematics, neuroscience, and engineering for further advancements.
