Results 191 to 200 of about 159,157 (335)

3D investigation and modeling of the geometric effects on porosity in packed beds

open access: yesAIChE Journal, EarlyView.
Abstract In porous beds, physical boundaries restrict particle arrangement, leading to inhomogeneous porosity. This paper reports on the porosity profiles that are the result of geometric effects on monodisperse packed beds in cylindrical and cubic arrangements. Special focus is given to the influence of edges and corners in cubic geometries.
Bastian Oldach   +3 more
wiley   +1 more source

A Comprehensive Assessment and Benchmark Study of Large Atomistic Foundation Models for Phonons

open access: yesAdvanced Intelligent Discovery, EarlyView.
We benchmark six large atomistic foundation models on 2429 crystalline materials for phonon transport properties. The rapid development of universal machine learning potentials (uMLPs) has enabled efficient, accurate predictions of diverse material properties across broad chemical spaces.
Md Zaibul Anam   +5 more
wiley   +1 more source

A Physics Constrained Machine Learning Pipeline for Young's Modulus Prediction in Multimaterial Hyperelastic Cylinders Guided by Contact Mechanics

open access: yesAdvanced Intelligent Discovery, EarlyView.
A physics‐guided machine learning framework estimates Young's modulus in multilayered multimaterial hyperelastic cylinders using contact mechanics. A semiempirical stiffness law is embedded into a custom neural network, ensuring physically consistent predictions. Validation against experimental and numerical data on C.
Christoforos Rekatsinas   +4 more
wiley   +1 more source

Gaussian Process Regression–Neural Network Hybrid with Optimized Redundant Coordinates: A New Simple Yet Potent Tool for Scientist's Machine Learning Toolbox

open access: yesAdvanced Intelligent Discovery, EarlyView.
A machine learning method, opt‐GPRNN, is presented that combines the advantages of neural networks and kernel regressions. It is based on additive GPR in optimized redundant coordinates and allows building a representation of the target with a small number of terms while avoiding overfitting when the number of terms is larger than optimal.
Sergei Manzhos, Manabu Ihara
wiley   +1 more source

Home - About - Disclaimer - Privacy