Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Xiguang Wu; Wenjiang Zhou; Haikuan Dong; Penghua Ying; Yanzhou Wang; Bai Song; Zheyong Fan; Shiyun Xiong

doi:10.1063/5.0213811

Paper

Atlas literature

Research paper

Computational MLIPComputational MDComputational Aimd

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Xiguang Wu, Wenjiang Zhou, Haikuan Dong, Penghua Ying et al.

arXiv·2024·10.1063/5.0213811·arXiv:2401.11427

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics | Matter42 Literature

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Correcting force error-induced underestimation of lattice thermal conductivity in machine learning molecular dynamics

Samples

Crystalline silicon structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

GaAs structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

Graphene structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

PbTe structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

Amorphous silicon configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Cubic germanium configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Si-Ge alloy configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

(10,10) carbon nanotube configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Why these are connected

Samples

Crystalline silicon structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

GaAs structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

Graphene structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

PbTe structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

Amorphous silicon configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Cubic germanium configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Si-Ge alloy configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

(10,10) carbon nanotube configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Why these are connected

Samples

Crystalline silicon structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

GaAs structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

Graphene structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

PbTe structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

Amorphous silicon configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Cubic germanium configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Si-Ge alloy configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

(10,10) carbon nanotube configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Why these are connected

Samples

Crystalline silicon structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

GaAs structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

Graphene structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

PbTe structures used to generate DFT/AIMD training data and benchmark MD thermal conductivity calculations.

Amorphous silicon configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Cubic germanium configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Si-Ge alloy configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

(10,10) carbon nanotube configurations used in Langevin-noise sensitivity analysis of HNEMD thermal conductivity.

Why these are connected