Modelling the measurement#
The forward model predicts the recorded trace for a trial pulse, and the retrieval returns the pulse that best explains the data under that model. A small residual shows that the model reproduces the trace; it does not show that the model describes the instrument. For few-femtosecond and deep-ultraviolet pulses the difference is large: on simulated deep-ultraviolet TG-FROG scans the thin-medium model fits the trace to a residual below 0.54 % yet returns 1.35–2.41 fs for a 1.05 fs pulse (validation). This chapter describes the hierarchy of models croak provides and when each is needed.
Which page do I need?#
My instrument has… |
Read |
|---|---|
a nonlinear medium thick enough to disperse the pulse (PG, SD) |
|
crossed beams (BOXCARS) whose arrival times vary across the focus |
|
a collection aperture that clips the signal beam chromatically |
|
an uncertain thickness or delay zero to fit |
Key concepts#
The models form a ladder. The standard model is the thin, local interaction. The dispersive slab propagates the input and signal fields through a medium of given material and thickness (PG and SD only). The extended model adds geometric delay smearing as a kernel set by the mask geometry. The full model replaces the kernel with an explicit chromatic focal quadrature and a modelled collection aperture. Each step costs more, and the full model benefits from a GPU. The retrieved field refers to the plane the model defines; compare fields only at the same plane.
Main entry points#
ForwardModel: the forward model, with material, thickness and depth quadrature.square_boxcars_kernel(),SmearingKernel: geometric smearing from the mask geometry.focal_mixture(): the chromatic focal quadrature.mask_hole_aperture(),aperture_from_scan(): the collection aperture.
The dispersive retrieval tutorial introduces the slab model, and the instrument examples compare the models on simulated scans. Optical data for the slab are in materials and mirrors. Continue with retrieval and numerical methods.