A5090316934- Machine Learning in Materials Science
- Electron and X-Ray Spectroscopy Techniques
- Computational Drug Discovery Methods
- Advanced Materials Characterization Techniques
- Metal-Organic Frameworks: Synthesis and Applications
- Mass Spectrometry Techniques and Applications
- X-ray Diffraction in Crystallography
- Electronic and Structural Properties of Oxides
- Polyoxometalates: Synthesis and Applications
- Perovskite Materials and Applications
- Quantum many-body systems
- Political Philosophy and Ethics
- Spectroscopy and Quantum Chemical Studies
University of Oxford
2018-2024
Science Oxford
2019
Machine learning (ML) approaches enable large-scale atomistic simulations with near-quantum-mechanical accuracy. With the growing availability of these methods, there arises a need for careful validation, particularly physically agnostic models-that is, potentials that extract nature atomic interactions from reference data. Here, we review basic principles behind ML and their validation atomic-scale material modeling. We discuss best practice in defining error metrics based on numerical...
Abstract Machine learning (ML) based interatomic potentials have transformed the field of atomistic materials modelling. However, ML depend critically on quality and quantity quantum-mechanical reference data with which they are trained, therefore developing datasets training pipelines is becoming an increasingly central challenge. Leveraging idea ‘synthetic’ (artificial) that common in other areas research, we here show synthetic data, themselves obtained at scale existing potential,...
We introduce a large “synthetic” dataset of atomistic structures and energies, generated using fast machine-learning model, we demonstrate its usefulness for supervised unsupervised ML tasks in chemistry.
Zeolitic imidazolate frameworks are widely thought of as being analogous to inorganic AB2 phases. We test the validity this assumption by comparing simplified and fully atomistic machine-learning models for local environments in ZIFs. Our work addresses central question what extent chemical information can be "coarse-grained" hybrid framework materials.
Machine learning has become ubiquitous in materials modelling and now routinely enables large-scale atomistic simulations with quantum-mechanical accuracy. However, developing machine-learned interatomic potentials requires high-quality training data, the manual generation curation of such data can be a major bottleneck. Here, we introduce an automated framework for exploration fitting potential-energy surfaces, implemented openly available software package that call autoplex (`automatic...
Zeolitic imidazolate frameworks are widely thought of as being analogous to inorganic AB$_{2}$ phases. We test the validity this assumption by comparing simplified and fully atomistic machine-learning models for local environments in ZIFs. Our work addresses central question what extent chemical information can be "coarse-grained" hybrid framework materials.
Machine-learning models are increasingly used to predict properties of atoms in chemical systems. There have been major advances developing descriptors and regression frameworks for this task, typically starting from (relatively) small sets quantum-mechanical reference data. Larger datasets kind becoming available, but remain expensive generate. Here we demonstrate the use a large dataset that "synthetically" labelled with per-atom energies an existing ML potential model. The cheapness...
We use the coarse-grained Frenkel-Holstein model to simulate relaxation, decoherence, and localization of photoexcited states in conformationally disordered π-conjugated polymers. The dynamics are computed via wave-packet propagation using matrix product time evolution block decimation method. ultrafast (i.e., t < 10 fs) coupling an exciton C-C bond vibrations creates exciton-polaron. relatively short (ca. monomers) exciton-phonon correlation length causes exciton-site which is observable on...
Machine learning (ML) approaches enable large-scale atomistic simulations with near-quantum-mechanical accuracy. With the growing availability of these methods there arises a need for careful validation, particularly physically agnostic models - that is, potentials which extract nature atomic interactions from reference data. Here, we review basic principles behind ML and their validation atomic-scale materials modeling. We discuss best practice in defining error metrics based on numerical...
Machine learning (ML) based interatomic potentials have transformed the field of atomistic materials modelling. However, ML depend critically on quality and quantity quantum-mechanical reference data with which they are trained, therefore developing datasets training pipelines is becoming an increasingly central challenge. Leveraging idea "synthetic" (artificial) that common in other areas research, we here show synthetic data, themselves obtained at scale existing potential, constitute a...