Tensor operations are a type of superior arithmetic that help many trendy applied sciences, particularly synthetic intelligence. These operations go far past the easy calculations most individuals encounter. A useful method to image them is to think about manipulating a Rubik’s dice in a number of dimensions without delay by rotating, slicing, or rearranging its layers. People and conventional computer systems should break these duties into sequences, however mild can carry out all of them on the identical time.
At present, tensor operations are important for AI methods concerned in picture processing, language understanding, and numerous different duties. As the quantity of knowledge continues to develop, typical digital {hardware} corresponding to GPUs faces growing pressure in pace, vitality use, and scalability.
Researchers Reveal Single-Shot Tensor Computing With Mild
To deal with these challenges, a world workforce led by Dr. Yufeng Zhang from the Photonics Group at Aalto College’s Division of Electronics and Nanoengineering has developed a essentially new strategy. Their technique permits complicated tensor calculations to be accomplished inside a single motion of sunshine by way of an optical system. The method, described as single-shot tensor computing, features on the pace of sunshine.
“Our technique performs the identical sorts of operations that at present’s GPUs deal with, like convolutions and a focus layers, however does all of them on the pace of sunshine,” says Dr. Zhang. “As a substitute of counting on digital circuits, we use the bodily properties of sunshine to carry out many computations concurrently.”
Encoding Info Into Mild for Excessive-Velocity Computation
The workforce completed this by embedding digital data into the amplitude and part of sunshine waves, reworking numerical information into bodily variations throughout the optical area. As these mild waves work together, they robotically perform mathematical procedures corresponding to matrix and tensor multiplication, which kind the premise of deep studying. By working with a number of wavelengths of sunshine, the researchers expanded their method to help much more complicated, higher-order tensor operations.
“Think about you are a customs officer who should examine each parcel by way of a number of machines with totally different features after which type them into the fitting bins,” Zhang says. “Usually, you’d course of every parcel one after the other. Our optical computing technique merges all parcels and all machines collectively — we create a number of ‘optical hooks’ that join every enter to its appropriate output. With only one operation, one move of sunshine, all inspections and sorting occur immediately and in parallel.”
Passive Optical Processing With Vast Compatibility
Probably the most putting advantages of this technique is how little intervention it requires. The mandatory operations happen on their very own as the sunshine travels, so the system doesn’t want lively management or digital switching throughout computation.
“This strategy could be applied on virtually any optical platform,” says Professor Zhipei Solar, chief of Aalto College’s Photonics Group. “Sooner or later, we plan to combine this computational framework instantly onto photonic chips, enabling light-based processors to carry out complicated AI duties with extraordinarily low energy consumption.”
Path Towards Future Mild-Based mostly AI {Hardware}
Zhang notes that the final word goal is to adapt the method to present {hardware} and platforms utilized by main know-how firms. He estimates that the tactic may very well be included into such methods inside 3 to five years.
“It will create a brand new era of optical computing methods, considerably accelerating complicated AI duties throughout a myriad of fields,” he concludes.
The examine was printed in Nature Photonics on November 14th, 2025.
