Molecular dynamics with on-the-fly machine learning of quantum-mechanical forces.

By Zhenwei Li, James R. Kermode, A. De Vita. 2015

514 citations recorded by the source corpus.

We present a molecular dynamics scheme which combines first-principles and machine-learning (ML) techniques in a single information-efficient approach. Forces on atoms are either predicted by Bayesian inference or, if necessary, computed by on-the-fly quantum-mechanical (QM) calculations and added to a growing ML database, whose completeness is, thus, never required. As a result, the scheme is accurate and general, while progressively fewer QM calls are needed when a new chemical process is encountered for the second and subsequent times, as demonstrated by tests on crystalline and molten silicon.

Subfields: Atomic and Molecular Physics, and Optics, Materials Chemistry, Molecular Biology, AI for Physical Sciences

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Atomic and Molecular Physics, and Optics

1. Yuri Mishin
2. Evgeny V. Podryabinkin
3. Alexander V. Shapeev
4. Rohit Batra
5. G. P. P. Pun
6. Rampi Ramprasad
7. Aldo Glielmo
8. A. De Vita

In Materials Chemistry

1. Yuri Mishin
2. Alexander V. Shapeev
3. Evgeny V. Podryabinkin
4. Rohit Batra
5. Rampi Ramprasad
6. G. P. P. Pun
7. Konstantin Gubaev
8. Ivan S. Novikov

In Molecular Biology

1. Yuri Mishin
2. Alexander V. Shapeev
3. Rohit Batra
4. Evgeny V. Podryabinkin
5. G. P. P. Pun
6. Rampi Ramprasad
7. Dongkun Zhang
8. G. Em Karniadakis

In AI for Physical Sciences

1. Dongkun Zhang
2. Alexander V. Shapeev
3. Evgeny V. Podryabinkin
4. Yuri Mishin
5. Rampi Ramprasad
6. Rohit Batra
7. Adrian E. Roitberg
8. G. P. P. Pun

Citations in the corpus

• The MLIP package: moment tensor potentials with MPI and active learning (2020, background): “These are symbolic-regression potentials [29], non-conservative force fields [8,38], force-fields for a fixed-size molecular system (typically applied to one or two organic molecules) [9, 42], on-lattice potentials [66], and cheminformatics-type models [10, 14,19,21,28,30,62,64,71], see also a recen”
• Outsmarting Quantum Chemistry Through Transfer Learning (2018, background): “Recent chemistry related work reports on ML models for chemical reactions 21,22 , potential energy surfaces 23–27 , forces 28–30 , atomization energies 31,32 , atomic partial charges 32– 35 , molecular dipoles 26,36,37 , materials discovery 38–40 , and protein-ligand complex scoring 41 .”
• Accurate interatomic force fields via machine learning with covariant kernels (2016, background): “Both linear and quadratic kernels are significantly more accurate than a classical Stillinger-Weber (SW) potential [55] fitted to reproduce the DFTB lattice parameter and bulk modulus [26].”