Acceleration of Graph Neural Network-Based Prediction Models in Chemistry via Co-Design Optimization on Intelligence Processing Units.

Atomic structure prediction and associated property calculations are the bedrock of chemical physics. Since high-fidelity ab initio modeling techniques for computing the structure and properties can be prohibitively expensive, this motivates the development of machine-learning (ML) models that make these predictions more efficiently. Training graph neural networks over large atomistic databases introduces unique computational challenges, such as the need to process millions of small graphs with variable size and support communication patterns that are distinct from learning over large graphs, such as social networks.

Geometric parameters of isotropic ensembles of right cylinders from the small-angle-scattering correlation function.

The scattered intensity of ensembles of right homogeneous quasi-diluted cylinders with constant oval right section (RS) and volume fraction phi are analyzed using the small-angle-scattering (SAS) correlation function (CF) gamma(r) = gamma(r, phi) in the isotropic two-phase approximation. A relation between the CF of the cylinder RS, beta(0)(r), and the CF of the single cylinder of height H, gamma(0)(r, H), allows the calculation of the explicit cylinder parameters of height, surface area, RS surface area, RS perimeter and volume. This is accomplished by evaluating the first two derivatives of gamma(0)(r) at r = 0.