Neural Networks Reveal the Impact of the Vibrational Dynamics in the Prediction of the Long-Time Mobility of Molecular Glassformers

Abstract

Two neural networks (NN) are designed to predict the particle mobility of a molecular glassformer in a wide time window ranging from vibrational dynamics to structural relaxation. Both NNs are trained by information concerning the local structure of the environment surrounding a given particle. The only difference in the learning procedure is the inclusion (NN A) or not (NN B) of the information provided by the fast, vibrational dynamics and quantified by the local Debye–Waller factor. It is found that, for a given temperature, the prediction provided by the NN A is more accurate, a finding which is tentatively ascribed to better account of the bond reorientation. Both NNs are found to exhibit impressive and rather comparable performance to predict the four-point susceptibility χ4(t) at τα, a measure of the dynamic heterogeneity of the system.

Glass Transition and Related Phenomena

“…Papers [1][2][3][4], based on the results of computer simulations, constitute an interesting collection of successful applications of isotropic and anisotropic models in equilibrium and non-equilibrium molecular dynamics (MD) computations, which are even extended to the study of dynamic properties using artificial intelligence methods [4] that become unavoidably necessary research tools. Lagogianni and Varnik [1] studied metallic glasses and their deformations, called shear bands, that hamper the application of such glasses as structural components.…”

“…The latter finding makes an important contribution to the long-standing discussion on the dynamic heterogeneity role in molecular dynamics of complex systems approaching the glass transition as well as our understanding of the primary and secondary relaxations and the coupling of relaxation processes. Going further in the second paper [4], Leporini's group introduced us to the modern era of artificial intelligence (AI) applications in a simulation study of disordered condensed matter systems. The authors first performed MD simulations of a three-dimensional polymer melt of fully-flexible (i.e., without bond-bond bending and torsional potentials) linear pentamers with a total number of monomers N = 10,000, where non-bonded monomers interact via an LJ potential and bonded monomers interact via a harmonic potential.…”