A Neural-Network-Optimized Hydrogen Peroxide Pairwise Additive Model for Classical Simulations DOI Creative Commons

Alvaro Ramos Peralta,

Gerardo Odriozola

Journal of Chemical Theory and Computation, Journal Year: 2023, Volume and Issue: 19(13), P. 4172 - 4181

Published: June 12, 2023

We have developed an all-atom pairwise additive model for hydrogen peroxide using optimization procedure based on artificial neural networks (ANNs). The is experimental molecular geometry and includes a dihedral potential that hinders the cis-type configuration allows crossing trans one, defined between planes two oxygen atoms each hydrogen. model's parametrization achieved by training simple ANNs to minimize target function measures differences various thermodynamic transport properties corresponding values. Finally, we evaluated range of optimized its mixtures with SPC/E water, including bulk-liquid (density, thermal expansion coefficient, adiabatic compressibility, etc.) systems at equilibrium (vapor liquid density, vapor pressure composition, surface tension, etc.). Overall, obtained good agreement data.

Language: Английский

Roadmap on machine learning glassy dynamics DOI
Gerhard Jung, Rinske M. Alkemade, Victor Bapst

et al.

Nature Reviews Physics, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 6, 2025

Language: Английский

Citations

3
Learning stochastic dynamics and predicting emergent behavior using transformers DOI Creative Commons
Corneel Casert, Isaac Tamblyn, Stephen Whitelam

et al.

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: Feb. 29, 2024

We show that a neural network originally designed for language processing can learn the dynamical rules of stochastic system by observation single trajectory system, and accurately predict its emergent behavior under conditions not observed during training. consider lattice model active matter undergoing continuous-time Monte Carlo dynamics, simulated at density which steady state comprises small, dispersed clusters. train called transformer on model. The transformer, we has capacity to represent are numerous nonlocal, learns dynamics this consists small number processes. Forward-propagated trajectories trained densities encountered training, exhibit motility-induced phase separation so existence nonequilibrium transition. Transformers have flexibility from without explicit enumeration rates or coarse-graining configuration space, procedure used here be applied wide range physical systems, including those with large complex generators.

Language: Английский

Citations

4
3-D rotation tracking from 2-D images of spherical colloids with textured surfaces DOI Creative Commons
Vincent Niggel, Maximilian R. Bailey, Carolina van Baalen

et al.

Soft Matter, Journal Year: 2023, Volume and Issue: 19(17), P. 3069 - 3079

Published: Jan. 1, 2023

Tracking the three-dimensional rotation of colloidal particles is essential to elucidate many open questions,

Language: Английский

Citations

6
Re-entrant percolation in active Brownian hard disks DOI Creative Commons
David Evans, José Martín‐Roca, Nathan J. Harmer

et al.

Soft Matter, Journal Year: 2024, Volume and Issue: 20(37), P. 7484 - 7492

Published: Jan. 1, 2024

Non-equilibrium clustering and percolation are investigated in an archetypal model of two-dimensional active matter using dynamic simulations self-propelled Brownian repulsive particles. We concentrate on the single-phase region up to moderate levels activity, before motility-induced phase separation (MIPS) sets in. Weak activity promotes cluster formation lowers threshold. However, driving system further out equilibrium partly reverses this effect, resulting a minimum critical density for system-spanning clusters introducing re-entrant as function pre-MIPS regime. This non-monotonic behaviour arises from competition between activity-induced effective attraction (which eventually leads MIPS) activity-driven breakup. Using adapted iterative Boltzmann inversion method, we derive potentials map weakly cases onto passive (equilibrium) with conservative attraction, which can be characterised by Monte Carlo simulations. While systems have practically identical radial distribution functions, find decisive differences higher-order structural correlations, threshold is highly sensitive. For sufficiently strong no pairwise potential reproduce system.

Language: Английский

Citations

1
A Neural-Network-Optimized Hydrogen Peroxide Pairwise Additive Model for Classical Simulations DOI Creative Commons

Alvaro Ramos Peralta,

Gerardo Odriozola

Journal of Chemical Theory and Computation, Journal Year: 2023, Volume and Issue: 19(13), P. 4172 - 4181

Published: June 12, 2023

We have developed an all-atom pairwise additive model for hydrogen peroxide using optimization procedure based on artificial neural networks (ANNs). The is experimental molecular geometry and includes a dihedral potential that hinders the cis-type configuration allows crossing trans one, defined between planes two oxygen atoms each hydrogen. model's parametrization achieved by training simple ANNs to minimize target function measures differences various thermodynamic transport properties corresponding values. Finally, we evaluated range of optimized its mixtures with SPC/E water, including bulk-liquid (density, thermal expansion coefficient, adiabatic compressibility, etc.) systems at equilibrium (vapor liquid density, vapor pressure composition, surface tension, etc.). Overall, obtained good agreement data.

Language: Английский

Citations

2