Time-space correlation between diffusive and structural fluctuations using LEAP. A) LENS and τSOAP time-series, computed for the three top-most layers (995 Cu atoms) of the Cu(211) FCC surface, with the corresponding KDE distributions. Univariate Onion clustering (45) is applied. The identified clusters are reported using different colors in both LENS (top) and τSOAP (bottom). The MD snapshots reported on the right, taken at the same time step ( ns), are colored according to the identified clusters, respectively. B) Left: Projection of the LEAP dataset on the 2D LENS-τSOAP phase space. LENS and τSOAP outlier domains are colored in red and blue, respectively. Right: Time-series of the LEAP magnitude (black) for two example IDs. The transit through LENS and τSOAP outlier domains (red and blue regions shown on the left) are identified with red and blue bands, respectively. C) Characteristic time scale () estimation for LENS and τSOAP fluctuations. Left: Cumulative distribution functions (CDFs) for the mean time interval (τ) between successive τSOAP fluctuations related to (i) trajectories visiting the blue region in (B) (dashed blue line), (ii) the subset of (i) crossing the red region in (B) after the blue one (solid blue line); (iii) CDF related to the time interval between LENS fluctuations occurring in the system (red line). Right: CDF for the number of τSOAP fluctuations preceding a LENS fluctuation. D) Space correlation between LENS and τSOAP fluctuations. x-axis: Distance between atoms lying in the LENS outlier region and atoms in the τSOAP outlier region, where (a) is the interatomic distance (a ∼3.6 Å in FCC Cu). E) Left: ID LENS value vs. τSOAP mean value of its neighbors. Right: ID τSOAP value vs. LENS mean value of its neighbors. Blue and red points correspond, respectively, to the blue and red domains in (B).
