Fig. 1: Schematic of the experimental setup and machine learning workflow for x-ray photon correlation spectroscopy (XPCS) data.
A In rheo-XPCS, a rheometer is placed in the beam path so that coherent X-rays scatter off the relaxing sample. B XPCS two-time correlations, C2(q, ϕ) are calculated by correlating intensity in a specific scattering region over time experimental times t1 and t2, where q represents the radial scattering wavenumber and ϕ describes the azimuthal scattering bin over which C2 is calculated. C shows a sample of experimental C2 to illustrate the wide variation in dynamics seen in non-equilibrium XPCS. The time scale bar in the left-most C2 applies to all other images. D The autoencoder is trained to reproduce raw C2, and the learned latent representation is used to cluster and classify data points (E).
