Fig. 3: Comparison of measured and simulated time-dependent difference pair distribution functions (ΔPDFs).
Panel a shows experimental and simulated (solid and dotted black lines) ΔPDFs at different delays in a time window of 0.5 picoseconds (ps) around time-zero. Simulations use the ab initio multiple spawning (AIMS) methods (see the “Methods” section). Analogous to Fig. 2, α, β, and γ regions are highlighted with different colors. Panel b shows the integrated experimental (solid) and simulated (dotted) time-dependent intensity of the three regions from panel a with identical color-coding. We fit error functions to the temporal onsets of the experimental signals (see Supplementary Note 3). The delay values of the error function centers are marked with arrows and as black dots with error bars. Additionally, the widths of the signal onsets according to the error function fits are marked by color-coded bars in panels (a) and (b). The shaded areas (simulation) and error bars (measurement) of the line plots indicate the uncertainty (s.e.m.) obtained from bootstrapping analysis (68% confidence interval). For the simulations, these error bars reflect convergence with respect to initial condition sampling. The temporal evolution of three representative carbon–carbon distances in the AIMS simulations, the (C3–C4), (C3,C5), and (C3,C10) distances, (labeling according to Fig. 2a) is plotted in panel (c). Additionally, snapshots of the molecular geometry evolution based on a representative AIMS trajectory are shown with the three representative carbon–carbon distances marked. Note the alignment of the vertical axes in all three panels.
