A5037628479- Advanced Thermodynamics and Statistical Mechanics
- Generative Adversarial Networks and Image Synthesis
- Theoretical and Computational Physics
- Spectroscopy and Quantum Chemical Studies
- Model Reduction and Neural Networks
- Protein Structure and Dynamics
- Machine Learning in Materials Science
- Cell Image Analysis Techniques
- Field-Flow Fractionation Techniques
- Domain Adaptation and Few-Shot Learning
- Material Dynamics and Properties
- Quantum, superfluid, helium dynamics
- Advanced biosensing and bioanalysis techniques
- Lipid Membrane Structure and Behavior
- Molecular Junctions and Nanostructures
- Computational Physics and Python Applications
- Superconducting and THz Device Technology
- Evolutionary Algorithms and Applications
- Characterization and Applications of Magnetic Nanoparticles
- Magnetic and Electromagnetic Effects
- Tropical and Extratropical Cyclones Research
- Micro and Nano Robotics
- Hydrological Forecasting Using AI
- Advanced Physical and Chemical Molecular Interactions
- Enhanced Oil Recovery Techniques
DeepMind (United Kingdom)
2020-2023
Google (United Kingdom)
2023
University of Cambridge
2014-2021
Global medium-range weather forecasting is critical to decision-making across many social and economic domains. Traditional numerical prediction uses increased compute resources improve forecast accuracy but does not directly use historical data the underlying model. Here, we introduce GraphCast, a machine learning-based method trained from reanalysis data. It predicts hundreds of variables for next 10 days at 0.25° resolution globally in under 1 minute. GraphCast significantly outperforms...
Global medium-range weather forecasting is critical to decision-making across many social and economic domains. Traditional numerical prediction uses increased compute resources improve forecast accuracy, but cannot directly use historical data the underlying model. We introduce a machine learning-based method called "GraphCast", which can be trained from reanalysis data. It predicts hundreds of variables, over 10 days at 0.25 degree resolution globally, in under one minute. show that...
Free energy perturbation (FEP) was proposed by Zwanzig [J. Chem. Phys. 22, 1420 (1954)] more than six decades ago as a method to estimate free differences and has since inspired huge body of related methods that use it an integral building block. Being importance sampling based estimator, however, FEP suffers from severe limitation: the requirement sufficient overlap between distributions. One strategy mitigate this problem, called Targeted FEP, uses high-dimensional mapping in configuration...
We propose a new algorithm for non-equilibrium molecular dynamics simulations of thermal gradients. The is an extension the heat exchange developed by Hafskjold et al. [Mol. Phys. 80, 1389 (1993); 81, 251 (1994)], in which certain amount added to one region and removed from another rescaling velocities appropriately. Since same between velocity steps Hamiltonian, expected conserve energy. However, it has been reported previously that original version exhibits pronounced drift total energy,...
Abstract We present a machine-learning approach, based on normalizing flows, for modelling atomic solids. Our model transforms an analytically tractable base distribution into the target solid without requiring ground-truth samples training. report Helmholtz free energy estimates cubic and hexagonal ice modelled as monatomic water well truncated shifted Lennard-Jones system, find them to be in excellent agreement with literature values from established baseline methods. further investigate...
In recent years, there has been a growing interest in using machine learning to overcome the high cost of numerical simulation, with some learned models achieving impressive speed-ups over classical solvers whilst maintaining accuracy. However, these methods are usually tested at low-resolution settings, and it remains be seen whether they can scale costly high-resolution simulations that we ultimately want tackle. this work, propose two complementary approaches improve framework from...
Biological membranes typically contain a large number of different components dispersed in small concentrations the main membrane phase, including proteins, sugars, and lipids varying geometrical properties. Most these do not bind cargo. Here, we show that such "inert" can be crucial for precise control cross-membrane trafficking. Using statistical mechanics model molecular dynamics simulations, demonstrate presence inert isotropic curvatures dramatically influences cargo endocytosis, even...
Abstract We present a machine-learning model based on normalizing flows that is trained to sample from the isobaric-isothermal ensemble. In our approach, we approximate joint distribution of fully-flexible triclinic simulation box and particle coordinates achieve desired internal pressure. This novel extension flow-based sampling ensemble yields direct estimates Gibbs free energies. test NPT -flow monatomic water in cubic hexagonal ice phases find excellent agreement energies other...
Using non-equilibrium molecular dynamics simulations, it has been recently demonstrated that water molecules align in response to an imposed temperature gradient, resulting effective electric field. Here, we investigate how thermally induced fields depend on the underlying treatment of long-ranged interactions. For short-ranged Wolf method and Ewald summation, find peak strength field range between 2 × 107 5 V/m for a gradient 5.2 K/Å. Our value is therefore order magnitude lower than...
Electrical charges are conserved. The same would be expected to hold for magnetic charges, yet monopoles have never been observed. It is therefore surprising that the laws of non-equilibrium thermodynamics, combined with Maxwell's equations, suggest colloidal particles heated or cooled in certain polar paramagnetic solvents may behave as if they carry an electrical/magnetic charge [J. Phys. Chem. B $\textbf{120}$, 5987 (2016)]. Here we present numerical simulations show field distribution...
We present a machine-learning model based on normalizing flows that is trained to sample from the isobaric-isothermal ensemble. In our approach, we approximate joint distribution of fully-flexible triclinic simulation box and particle coordinates achieve desired internal pressure. This novel extension flow-based sampling ensemble yields direct estimates Gibbs free energies. test NPT-flow monatomic water in cubic hexagonal ice phases find excellent agreement energies other observables...
We present a mean-field theory to explain the thermo-orientation effect in an off-center Stockmayer fluid. This is underlying cause of thermally induced polarization and monopoles, which have recently been predicted theoretically. Unlike previous theories that are based either on phenomenological equations or scaling arguments, our approach does not require any fitting parameters. Given equation state assuming local equilibrium, we construct effective Hamiltonian for computation Boltzmann...
Learning dynamics is at the heart of many important applications machine learning (ML), such as robotics and autonomous driving. In these settings, ML algorithms typically need to reason about a physical system using high dimensional observations, images, without access underlying state. Recently, several methods have proposed integrate priors from classical mechanics into models address challenge reasoning images. this work, we take sober look current capabilities models. To end, introduce...
Summary If secondary hydrocarbon recovery methods fail because of the occurrence gravity override or viscous fingering one can turn to an enhanced oil method like injection foam. The generation foam be described by a set partial differential equations with strongly nonlinear functions, which impose challenges for numerical modeling. To analyze effect on fingering, we study dynamics simple model based Buckley-Leverett equation. Whereas flux is smooth function water saturation, will cause...
Denoising Score Matching estimates the score of a noised version target distribution by minimizing regression loss and is widely used to train popular class Diffusion Models. A well known limitation Matching, however, that it yields poor at low noise levels. This issue particularly unfavourable for problems in physical sciences Monte Carlo sampling tasks which clean original known. Intuitively, estimating slightly should be simple task such cases. In this paper, we address shortcoming show...
When fluids of anisotropic molecules are placed in temperature gradients, the may align themselves along gradient: this is called thermo-orientation. We discuss theory effect a fluid particles that interact by spherically symmetric potential, where particles' centres mass do not coincide with their interaction centres. Starting from equations motion molecules, we show how simple assumption local equipartition energy can be used to predict thermo-orientation effect, recovering result...