scholarly journals Guidelines for creating artificial neural network empirical interatomic potential from first-principles molecular dynamics data under specific conditions and its application to α-Ag2Se

2019 ◽  
Vol 151 (12) ◽  
pp. 124303 ◽  
Author(s):  
Kohei Shimamura ◽  
Shogo Fukushima ◽  
Akihide Koura ◽  
Fuyuki Shimojo ◽  
Masaaki Misawa ◽  
...  
Keyword(s):  
Neural Network ◽  
Molecular Dynamics ◽  
Artificial Neural Network ◽  
First Principles ◽  
Interatomic Potential ◽  
Artificial Neural ◽  
First Principles Molecular Dynamics

Artificial neural network assisted by first-principles calculations for predicting transformation temperatures in shape memory alloys

2019 ◽  
Vol 33 (08) ◽  
pp. 1950055 ◽  
Author(s):  
Daichi Minami ◽  
Tokuteru Uesugi ◽  
Yorinobu Takigawa ◽  
Kenji Higashi
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Linear Regression ◽  
Shape Memory Alloys ◽  
Shape Memory ◽  
First Principles ◽  
Accurate Method ◽  
First Principles Calculations ◽  
Simple Linear Regression ◽  
Artificial Neural

A key property for the design of new shape memory alloys is their working temperature range that depends on their transformation temperature T0. In previous works, T0 was predicted using a simple linear regression with respect to the energy difference between the parent and the martensitic phases, [Formula: see text]E[Formula: see text]. In this paper, we developed an accurate method to predict T0 based on machine learning assisted by the first-principles calculations. First-principles calculations were performed on 15 shape memory alloys; then, we proposed an artificial neural network method that used not only computed [Formula: see text]E[Formula: see text] but also bulk moduli as input variables to predict T0. The prediction error of T0 was improved to 49 K for the proposed artificial neural network compared with 188 K for simple linear regression.

scholarly journals A systematic approach to generating accurate neural network potentials: the case of carbon

2021 ◽  
Vol 7 (1) ◽  
Author(s):  
Yusuf Shaidu ◽  
Emine Küçükbenli ◽  
Ruggero Lot ◽  
Franco Pellegrini ◽  
Efthimios Kaxiras ◽  
...  
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Structure Prediction ◽  
Interatomic Potential ◽  
Interatomic Potentials ◽  
Amorphous Phases ◽  
Network Training ◽  
Materials Modeling ◽  
Wide Range ◽  
Artificial Neural

AbstractAvailability of affordable and widely applicable interatomic potentials is the key needed to unlock the riches of modern materials modeling. Artificial neural network-based approaches for generating potentials are promising; however, neural network training requires large amounts of data, sampled adequately from an often unknown potential energy surface. Here we propose a self-consistent approach that is based on crystal structure prediction formalism and is guided by unsupervised data analysis, to construct an accurate, inexpensive, and transferable artificial neural network potential. Using this approach, we construct an interatomic potential for carbon and demonstrate its ability to reproduce first principles results on elastic and vibrational properties for diamond, graphite, and graphene, as well as energy ordering and structural properties of a wide range of crystalline and amorphous phases.

Sign in / Sign up

Export Citation Format

Share Document