Photoelectric hybrid incoherent complex-valued matrix operation processor and complex-valued matrix operation method

Abstract

The invention discloses a photoelectric hybrid incoherent complex-valued matrix operation processor and a complex-valued matrix operation method. The invention belongs to the field of integrated optical computing. The system comprises an electronic control unit, a wavelength division multiplexing unit, an optical complex value matrix operation unit and a data acquisition unit, wherein the optical complex-valued matrix operation unit comprises a tunable array, a transmission matrix is coded in the tunable array, a mixed optical signal coded with an input vector is input and then is divided into multiple paths of signals with equal power, and the multiple paths of signals are respectively coupled to each row of resonance devices, so that complex-valued matrix operation of a non-negative real number field is completed; the operation is divided into two matrix operations according to positive and negative elements, and then the two matrix operations are subjected to subtraction to realize full real number field matrix operation; the matrix operation is split into four matrix operations and two times of electrical addition and subtraction through separation of a real part and an imaginary part, so that complex value matrix operation is realized; finally, large-scale complex value matrix operation is realized through matrix partitioning, and signal transformation is realized. According to the invention, a photoelectric hybrid processor which is based on an incoherent architecture and can realize large-scale complex value matrix operation can be constructed.

Description

technical field [0001] The invention belongs to the field of integrated optical computing, and more specifically relates to a photoelectric hybrid incoherent complex-valued matrix operation processor and a complex-valued matrix operation method. Background technique [0002] With the rapid development of artificial intelligence technology, the growing demand for high-performance computing has driven the development of custom hardware to accelerate specific classes of computing. However, as the exponential scale of electronic transistors reaches the physical limits revealed by Moore's Law, processors based on electronic hardware have hit the bottleneck of unsustainable performance growth. Photonic processors use photons instead of electrons to perform calculations, so optical computing can significantly speed up calculations by overcoming the inherent limitations of electrons. Unlike integrated circuits, integrated optical circuits have excellent properties such as ultra-wid...

AI Technical Summary

Problems solved by technology

However, the disadvantage of the MRR array is that it is an incoherent network. In optical calculations, there is only intensity information and no phase information, so complex operations cannot be performed, which limits its application fields.

Images