Complete pulse characterization through optical correlations and genetic algorithm

Summary form only given. There is a great deal of work on using autocorrelations to characterize the amplitude and phase of pulses. Recently, Koumans and Yariv presented an iterative algorithm to recover the amplitude and phase from autocorrelations of a pulse after propagation through different lengths of dispersive material. They coin this approach DP-TROG, since it is similar to FROG except that the role of time and frequency are interchanged. Their inversion algorithm assumes that the fiber behaves linearly. In a real system using fiber as the dispersive medium, high power is needed to obtain enough SHG signal to obtain a clean autocorrelation trace, and since the effective area of fiber is small, nonlinear effects can be important. In our experiments, nonlinear pulse shaping effects were clearly visible in the autocorrelation when 3 ps pulses from a 5 GHz actively modelocked laser diode were used. Therefore, the problem of recovering the pulse amplitude and phase from clean autocorrelation traces is inherently a nonlinear problem. One way to recover the pulse amplitude and phase is by the use of a genetic algorithm.