Machine learning-assisted approximation of symmetrized quantum time correlation functions

Open-chain imaginary-time path-integral sampling approach known with the acronym OPSCF (J. Chem. Phys. 148, 102340 (2018)) is an approach to the calculation of approximate symmetrized quantum time correlation functions. In OPSCF approach, the real time t is treated as a parameter, and therefore for each real time t, a separate simulation on the imaginary time axis is needed to be run, which makes the OPSCF approach quite expensive and as a result, the approach loses the advantage of being a standard path-integral sampling approach. In this study, I propose that the use of OPSCF approach in combination with machine learning can reduce the computational cost by 75% to 90% (depending on the problem at hand). Combining OPSCF approach with ML is very straight forward which gives an upper hand to OPSCF approach over the trajectory-based methods such as the centroid molecular dynamics (CMD) and the ring-polymer molecular dynamics (RPMD).

Noise reduction in X-ray photon correlation spectroscopy with convolutional neural networks encoder–decoder models

Like other experimental techniques, X-ray photon correlation spectroscopy is subject to various kinds of noise. Random and correlated fluctuations and heterogeneities can be present in a two-time correlation function and obscure the information about the intrinsic dynamics of a sample. Simultaneously addressing the disparate origins of noise in the experimental data is challenging. We propose a computational approach for improving the signal-to-noise ratio in two-time correlation functions that is based on convolutional neural network encoder–decoder (CNN-ED) models. Such models extract features from an image via convolutional layers, project them to a low dimensional space and then reconstruct a clean image from this reduced representation via transposed convolutional layers. Not only are ED models a general tool for random noise removal, but their application to low signal-to-noise data can enhance the data’s quantitative usage since they are able to learn the functional form of the signal. We demonstrate that the CNN-ED models trained on real-world experimental data help to effectively extract equilibrium dynamics’ parameters from two-time correlation functions, containing statistical noise and dynamic heterogeneities. Strategies for optimizing the models’ performance and their applicability limits are discussed.