Files include the data presented in the manuscript entitled: "Tailoring a highly nonlinear fiber amplifier for the emission in the 1565-1675 nm spectral range with machine learning" by M. Krakowski, A. Kwaśny, D. Stachowiak and G. Soboń.
Figure 2a presents the amplifier output spectrum produced by the amplifier using a 1560 nm seed wavelength
Figure 2b presents the amplifier output spectrum produced by the amplifier using a 1540 nm seed wavelength
Figure 4 shows the best settings of the system for the SHG-based optmization:
Figure 4a presents the best fitness (cost function) results over the iterations of the optimization process.
Figure 4b presents the spectral and spectral phase profiles of the best algorithm setting, calculated by the algorithm, overlaid on the measured seed spectrum at the preamplifier's output
Figure 4c presents the recorded spectrum of the best fitness for the cost function at the system's output
Figure 5 shows the measurements of the best results obtained from OSA-based optmization:
Figure 5a presents the FROG temporal profile of the pulse along with its temporal phase profile
Figure 5b presents the recorded autocorrelation of the pulse, with the Gaussian fit fitting the autocorrelation function
Figure 5c presents the FROG-retrieved spectrum of the pulse along with the spectral phase profile
Figure 5d presents the measured FROG spectrogram
Figure 5e presents the retrieved FROG spectrogram
Figure 6 shows the best settings of the system for the SHG-based optmization:
Figure 6a presents the best fitness (cost function) results over the iterations of the optimization process.
Figure 6b presents the spectral and spectral phase profiles of the best algorithm setting, calculated by the algorithm, overlaid on the measured seed spectrum at the preamplifier's output
Figure 6c presents the recorded spectrum of the best fitness for the cost function at the system's output
Figure 7 shows the measurements of the best results obtained from SHG-based optmization:
Figure 7a presents the FROG temporal profile of the pulse along with its temporal phase profile
Figure 7b presents the recorded autocorrelation of the pulse, with the Gaussian fit fitting the autocorrelation function
Figure 7c presents the FROG-retrieved spectrum of the pulse along with the spectral phase profile
Figure 7d presents the measured FROG spectrogram
Figure 7e presents the retrieved FROG spectrogram
Figure 8 shows the comparison of the amplifier output spectra from the best results from both optimizations in the logarithmic domain recorded with OSA
Figure 9 shows the spectral and temporal evolution (autocorrelation traces) of the output pulse for: a) optimized and b) unoptimized cases in relation to pump power
(2026-05)
