Remote dispersion scan: transformer-network retrieval of ultrafast pulses after non-linear propagation

Accurate and rapid characterization of broadband electric fields is essential for all ultrafast applications and remains an active field of research. In this work, we introduce remote dispersion scan, a transformer neural network enabled dispersion scan based pulse characterization method that can characterize femtosecond laser pulses. A local scan of the non-linear spectral phase before several linear and nonlinear processes, including amplification, compression, and self phase modulation, allows for the field retrieval remotely at the interaction region. We show that the reconstruction accuracy obtained from a single measurement of the fundamental and second harmonic is comparable to that of a full two dimensional scan. We confirm the technique experimentally by compressing a 300 W, 1.3 ps, 1030 nm pulse in a hollow core fiber to 100 fs and measuring the fundamental and second harmonic spectra while scanning the second order phase in a pulse shaper before power amplification. These results establish a simple, robust, alignment insensitive live-view pulse reconstruction modality.