Fast dispersion-scan retrieval: neural networks vs iterative algorithm

The dispersion-scan (d-scan) is a simple, inline pulse characterization with single-shot trace acquisition implementations. Together with fast d-scan retrieval algorithms it enables a pathway for (near) single-shot and real-time pulse retrieval. In this work, we extend current neural networks (NNs) to simultaneously retrieve the spectrum and spectral phase of a pulse from dispersion scan traces and compare their speed and retrieval quality with a highly optimized iterative algorithm.