Minimization of dispersion in an ultrafast chirped pulse amplifier using adaptive learning

Minimizing residual frequency dispersion that accompanies pulse stretching, amplification, and recompression is an important consideration in ultrashort chirped-pulse amplifiers.. Here we show how an adaptive learning algorithm can be used in conjunction with a pulse shaper to compensate for higher-order and nonlinear dispersion in a chirped-pulse amplifier. Using spectral blueshifting as a sensitive diagnostic for pulse shape, we implement a 'learning loop' comprised of the pulse shaper, strong field laser ionization, and a genetic algorithm to minimize dispersion through the amplifier. We verify our optimization results using frequency-resolved optical Sating (FROG) measurements and also show theoretically and experimentally that spectral blueshifting is indeed a sensitive diagnostic for pulse shape, and specifically, for higher-order dispersion.