����ֱ��

There has been a tremendous rise in machine learning applications that have motivated the investigation of new computing architectures dedicated to performing complex calculations. Neuromorphic integrated photonic computing aims to fabricate electrical and optical devices on the same chip to create an electro-optical circuit capable of performing neural network computations much faster than the conventional transistor-based computing architectures. There has been remarkable research going into neuromorphic photonics, but one of the significant challenges is the lack of good tools to simulate large-scale photonic integrated circuits.

Queen’s graduate student Jagmeet Singh and collaborators have presented a Verilog-A (time-domain modeling language for analog circuits) based approach to perform electro-optical co-simulation on the same platform. This work would significantly improve the efficiency of optimizing the circuits and help researchers provide an accurate simulation of the circuit performance.