Developing and applying machine learning approaches to understand water at interfaces
Lead Research Organisation:
UNIVERSITY COLLEGE LONDON
Department Name: London Centre for Nanotechnology
Abstract
Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
Organisations
Publications
Huang K
(2020)
Cation-controlled wetting properties of vermiculite membranes and its promise for fouling resistant oil-water separation.
in Nature communications
Rowe P
(2018)
Development of a machine learning potential for graphene
in Physical Review B
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/N509577/1 | 30/09/2016 | 24/03/2022 | |||
| 1782307 | Studentship | EP/N509577/1 | 30/09/2016 | 30/12/2020 | Patrick Rowe |
| Description | In the past there have been two available methodologies for simulating the atomic-scale dynamical behaviour of materials in computers - using classical molecular dynamics where atoms are approximated as hard spheres interacting with fixed interactions - which lacks accuracy, and using ab initio molcular dynamics where forces are computed on the fly from quantum mechanics - which is extremely expensive. Machine learning approaches have arisen recently as a way of obtaining quantum mechanical accuracy with the cost of classical molecular dynamics. We have shown for the first time that the accuracy of a machine learning potential can be high enough that the results are functionally identical to those from direct ab initio evaluation. Further, we have extended our work on pure graphene to produce a machine learning model which is accurate for a wide range of relevant phases for carbon, which maintains accuracy on the crystalline phases while simultaneously being flexible enough to treat liquid and amorphous structures. |
| Exploitation Route | database used for training made publicly and freely available, code used to train potential made publicly available, anyone can download and continue this work. |
| Sectors | Aerospace Defence and Marine Chemicals Digital/Communication/Information Technologies (including Software) Energy |
| URL | https://journals.aps.org/prb/abstract/10.1103/PhysRevB.97.054303 |
| Title | Database of Graphene Structures and Quantum Mechanical Forces |
| Description | As part of a publication: Development of a machine learning potential for graphene, we made available free of charge a database of several thousand structures of graphene, with the associated physical properties (energies, forces..) calculated from density functional theory. Such a database is suitable for use in applications such as machine learning. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2017 |
| Provided To Others? | Yes |
| Impact | N/a |
| URL | http://www.libatoms.org/pub/Home/DataRepository/graphene_gap_final.tar.gz |
| Title | Machine Learning Potential for Graphene |
| Description | This GAP model obtains a faithful representation of a density functional theory (DFT) potential energy surface, facilitating highly accurate (approaching the accuracy of ab initio methods) molecular dynamics simulations. This is achieved at a computational cost which is orders of magnitude lower than that of comparable calculations which directly invoke electronic structure methods. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2018 |
| Provided To Others? | Yes |
| Impact | Demonstrated the potential accuracy of machine learning models, helped to encourage further research |
| URL | http://www.libatoms.org/pub/Home/DataRepository/graphene_gap_final.tar.gz |