Fig. 2
(a) Euclidean distance matrices and the corresponding fBm trajectories, generated by the unconditional diffusion model, for the three values of Hurst exponent: \(H=1/3\) (subdiffusion), \(H=1/2\) (normal diffusion) and \(H=2/3\) (superdiffusion). The trajectories were obtained using gradient optimization of three-dimensional coordinates to match the generated distance matrices. The color changes from red to blue along the trajectory. (b) Scaling of the typical distances as a function of the contour length s for \(H=1/3\) (red), \(H=1/2\) (green) and \(H=2/3\) (blue). EDM samples from the generated ensemble fill the grey intervals and thick color lines correspond to the ensemble-averaged curves. Black lines correspond to the training databases. (c) Collapsed probability densities of the diffusion-generated pairwise distances between two points on the trajectory separated by contour distance s. The black curve corresponds to the Maxwell distribution \(4\pi r^2 \times P(r|s)\), where P(r|s) is the standard Gaussian distribution.
