A deep-learning artificial neural network (NN) combined with the particle swarm optimization (PSO) method has been proposed to inversely design the semiconductor laser with high accuracy and computational speed. This method is exempt from the single-solution problem of tandem NN and can be highly useful to extract the possible problematic parameters in the failure analysis of a device. The light-current curves and small signal responses have been tested against the benchmarks calculated by the traveling-wave model to demonstrate the NN’s robustness and efficiency in simulating the laser behavior for further use in the inverse design by PSO.

No abstract

No abstract

No abstract

No abstract