How do you view the recent hot machine learning trend?

Science Talk
1

Recently, first principles + machine learning + molecular dynamics is very hot, many research groups have been using machine learning to train the force field potential function of related materials, are these trained potential functions more accurate and reliable than the existing ones? Will the existing potential functions necessarily be retrained and replaced?

Welcome to discussion and recommend relevant literature and reviews.

2

Sorry, but this is not about LAMMPS and thus off-topic for the LAMMPS categories.
Recategorized it to “Science Talk”.

1 Like
3

i recommend a review article we covered in my MD&MC class:

https://doi.org/10.1016/j.actamat.2021.116980

4

I hope not. If everything computational start becoming machine learning related stuff I will have to quit science and go plant strawberries

1 Like