Measured d-scan trace
SHG spectrum vs glass insertion (with detector noise)
Retrieved d-scan trace
Simulated from the retrieved spectral phase
Residual map
Measured minus scaled retrieved trace
Fundamental spectrum and spectral phase
True phase (dashed) vs retrieved phase (red)
Temporal intensity and phase
True, retrieved, and transform-limited pulse
Retrieval convergence
rms trace error G vs iteration (log scale)
Pulse duration vs insertion
How the wedges compress and stretch the pulse
Spectrogram of retrieved pulse
Gabor transform: time-frequency structure
Wedge glass dispersion
Refractive index and GVD of the selected material
Spectral marginals
Trace integrated over insertion: measured vs retrieved
About this simulation
The dispersion scan (d-scan) is a pulse-characterization technique in which a femtosecond pulse is sent through a pair of glass wedges (variable, known dispersion), then frequency-doubled in a nonlinear crystal. Recording the SHG spectrum as a function of glass insertion produces a two-dimensional trace that encodes the spectral phase of the pulse. Here a test pulse is defined by its spectrum and Taylor phase coefficients (GDD, TOD, FOD); its d-scan trace is simulated, detector noise is added, and the spectral phase is then retrieved from the noisy trace alone (plus the known fundamental spectrum), exactly as in a real experiment.
Retrieval: a coarse chirp scan first fits (GDD, TOD, FOD) directly to the trace; a generalized-projections algorithm then refines the full spectral phase point by point. The error G is the rms difference between measured and retrieved traces; with noise present, G converges to the noise floor. All quantities update in real time when any input changes.