scholarly journals Automated discovery of a robust interatomic potential for aluminum

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Justin S. Smith ◽  
Benjamin Nebgen ◽  
Nithin Mathew ◽  
Jie Chen ◽  
Nicholas Lubbers ◽  
...  
Keyword(s):  
Potential Energy Surfaces ◽  
Interatomic Potential ◽  
Critical Factor ◽  
Coexistence Curve ◽  
Training Dataset ◽  
Defect Energies ◽  
Automated Discovery ◽  
Non Equilibrium Molecular Dynamics ◽  
Dynamics Simulations ◽  
Energy Surfaces

AbstractMachine learning, trained on quantum mechanics (QM) calculations, is a powerful tool for modeling potential energy surfaces. A critical factor is the quality and diversity of the training dataset. Here we present a highly automated approach to dataset construction and demonstrate the method by building a potential for elemental aluminum (ANI-Al). In our active learning scheme, the ML potential under development is used to drive non-equilibrium molecular dynamics simulations with time-varying applied temperatures. Whenever a configuration is reached for which the ML uncertainty is large, new QM data is collected. The ML model is periodically retrained on all available QM data. The final ANI-Al potential makes very accurate predictions of radial distribution function in melt, liquid-solid coexistence curve, and crystal properties such as defect energies and barriers. We perform a 1.3M atom shock simulation and show that ANI-Al force predictions shine in their agreement with new reference DFT calculations.

scholarly journals Energetic Self-Folding Mechanism in α-Helixes

2019 ◽  
Author(s):  
Adolfo Bastida ◽  
José Zúñiga ◽  
Alberto Requena ◽  
Javier Cerezo
Keyword(s):  
Molecular Dynamics ◽  
Potential Energy Surfaces ◽  
Reasonable Agreement ◽  
Dihedral Angles ◽  
Native Conformation ◽  
Experimental Values ◽  
Α Helix ◽  
Dynamics Simulations ◽  
Energy Surfaces ◽  
Large Peptide

A novel energetic route driving the folding of a polyalanine peptide from an extended conformation to its α-helix native conformation is described, supported by a new method to compute mean potential energy surfaces accurately in terms of the dihedral angles of the peptide chain from extensive Molecular Dynamics simulations. The Energetic Self-Folding (ESF) route arises specifically from the balance between the intrinsic propensity of alanine residues towards the α<sub>R </sub>conformation and two, opposite, effects: the destabilizing interaction with neighbor residues and the stabilizing formation of native hydrogen bonds, with the latter being dominant for large peptide lengths. The ESF mechanism provides simple but robust support to the nucleation-elongation, or zipper models, and offers a quantitative energetic funnel picture of the folding process. The mechanism is validated by the reasonable agreement between the computed folding energies and the experimental values.

Empirical Potential-Energy Surfaces Fitting Using Feed forward Neural Networks

2012 ◽  
Author(s):  
Lionel Raff ◽  
Ranga Komanduri ◽  
Martin Hagan ◽  
Satish Bukkapatnam
Keyword(s):  
Potential Energy ◽  
Ab Initio ◽  
Potential Energy Surfaces ◽  
Force Fields ◽  
Pair Potential ◽  
Potential Surfaces ◽  
Empirical Potential ◽  
Functional Forms ◽  
Dynamics Simulations ◽  
Energy Surfaces

When the system of interest becomes too complex to permit the use of ab initio methods to obtain the system potential-energy surfaces (PES), empirical potential surfaces are frequently employed to represent the force fields present in the system under investigation. In most cases, the functional forms present in these potentials are selected on the basis of chemical and physical intuitions. The parameters of the surface are frequently adjusted to fit a very small set of experimental data that comprise bond energies, equilibrium bond distances and angles, fundamental vibrational frequencies, and perhaps measured barrier heights to reactions of interest. Such potentials generally yield only qualitative or semiquantitative descriptions of the system dynamics. Several research groups have significantly improved the accuracy of the values of the experimental properties computed using empirical potential surfaces by fitting the chosen functional form for the potential to the force fields obtained from trajectories using ab initio Car-Parrinello molecular dynamics simulations. The fitting to the force fields is usually done using a least-squares fitting approach. This method has been employed by Izvekov et al. to obtain effective non-polarizable three-site force fields for liquid water. Carré et al. have employed such a procedure to obtain a new pair potential for silica. In their investigation, the vector of potential parameters was fitted using an iterative Levenberg-Marquardt algorithm. Tangney and Scandolo have also developed an interatomic force field for liquid SiO2 in which the parameters were fitted to the forces, stresses, and energies obtained from ab initio calculations. Ercolessi and Adams have used a quasi-Newtonian procedure to fit an empirical potential for aluminum to data obtained from first-principals computations. Empirical potentials can be improved by making the parameters parameterized functions of the coordinates defining the instantaneous positions of the atoms of the system. This approach has been successfully employed by numerous investigators The difficulty with this procedure is that the number of parameters that must be adjusted increases rapidly. Appropriate fitting of these parameters requires a much more extensive database. Finally, the actual fitting process can often be tedious, difficult, and time-consuming.

scholarly journals Differentiable sampling of molecular geometries with uncertainty-based adversarial attacks

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Daniel Schwalbe-Koda ◽  
Aik Rui Tan ◽  
Rafael Gómez-Bombarelli
Keyword(s):  
Potential Energy Surfaces ◽  
Automatic Differentiation ◽  
Ground Truth ◽  
Training Data ◽  
Interatomic Potentials ◽  
Collective Variables ◽  
Molecule Interactions ◽  
Dynamics Simulations ◽  
Fast Prediction ◽  
Energy Surfaces

AbstractNeural network (NN) interatomic potentials provide fast prediction of potential energy surfaces, closely matching the accuracy of the electronic structure methods used to produce the training data. However, NN predictions are only reliable within well-learned training domains, and show volatile behavior when extrapolating. Uncertainty quantification methods can flag atomic configurations for which prediction confidence is low, but arriving at such uncertain regions requires expensive sampling of the NN phase space, often using atomistic simulations. Here, we exploit automatic differentiation to drive atomistic systems towards high-likelihood, high-uncertainty configurations without the need for molecular dynamics simulations. By performing adversarial attacks on an uncertainty metric, informative geometries that expand the training domain of NNs are sampled. When combined with an active learning loop, this approach bootstraps and improves NN potentials while decreasing the number of calls to the ground truth method. This efficiency is demonstrated on sampling of kinetic barriers, collective variables in molecules, and supramolecular chemistry in zeolite-molecule interactions, and can be extended to any NN potential architecture and materials system.

scholarly journals De novo exploration and self-guided learning of potential-energy surfaces

2019 ◽  
Vol 5 (1) ◽  
Author(s):  
Noam Bernstein ◽  
Gábor Csányi ◽  
Volker L. Deringer
Keyword(s):  
Potential Energy ◽  
Potential Energy Surfaces ◽  
Materials Science ◽  
De Novo ◽  
Interatomic Potential ◽  
Single Point ◽  
Small Unit ◽  
Wide Range ◽  
Unit Cells ◽  
Energy Surfaces

Abstract Interatomic potential models based on machine learning (ML) are rapidly developing as tools for material simulations. However, because of their flexibility, they require large fitting databases that are normally created with substantial manual selection and tuning of reference configurations. Here, we show that ML potentials can be built in a largely automated fashion, exploring and fitting potential-energy surfaces from the beginning (de novo) within one and the same protocol. The key enabling step is the use of a configuration-averaged kernel metric that allows one to select the few most relevant and diverse structures at each step. The resulting potentials are accurate and robust for the wide range of configurations that occur during structure searching, despite only requiring a relatively small number of single-point DFT calculations on small unit cells. We apply the method to materials with diverse chemical nature and coordination environments, marking an important step toward the more routine application of ML potentials in physics, chemistry, and materials science.

scholarly journals Nonadiabatic Dynamics in Multidimensional Complex Potential Energy Surfaces

2020 ◽  
Author(s):  
Fábris Kossoski ◽  
Mario Barbatti
Keyword(s):  
Potential Energy ◽  
Potential Energy Surfaces ◽  
Electron Attachment ◽  
Complex Surface ◽  
Nonadiabatic Dynamics ◽  
Surface Hopping ◽  
Molecular Systems ◽  
Out Of Plane ◽  
Dynamics Simulations ◽  
Energy Surfaces

<p>Despite the continuous development of theoretical methodologies for describing nonadiabatic dynamics of molecular systems, there is a lack of approaches for processes where the norm of the wave function is not conserved, i.e., when an imaginary potential accounts for some irreversible decaying mechanism. Current approaches rely on building potential energy surfaces of reduced dimensionality, which is not optimal for more involving and realistic multidimensional problems. Here, we present a novel methodology for describing the dynamics of complex-valued molecular Hamiltonians, which is a generalisation of the trajectory surface hopping method. As a first application, the complex surface fewest switches surface hopping (CS-FSSH) method was employed to survey the relaxation mechanisms of the shape resonant anions of iodoethene. We have provided the first detailed and dynamical picture of the p*/s* mechanism of dissociative electron attachment in halogenated unsaturated compounds, which is believed to underlie electron-induced reactions of several molecules of interest. Electron capture into the p* orbital promotes C=C stretching and out-of-plane vibrations, followed by charge transfer from the double bond into the s* orbital at the C-I bond, and, finally, release of the iodine ion, all within only 15 fs. On-the-fly dynamics simulations of a vast class of processes can be envisioned with the CS-FSSH methodology, including autoionisation from transient anions, core-ionised and superexcited states, Auger and interatomic Coulombic decay, and time-dependent luminescence.</p>

Reduced-dimensional surface hopping with offline–online computations

2021 ◽  
Author(s):  
Zachary Morrow ◽  
Hyuk-Yong Kwon ◽  
Carl Tim Kelley ◽  
Elena Jakubikova
Keyword(s):  
Molecular Dynamics ◽  
Potential Energy ◽  
Molecular Dynamics Simulations ◽  
Potential Energy Surfaces ◽  
Energy Levels ◽  
Surface Hopping ◽  
Dynamics Simulations ◽  
Energy Surfaces ◽  
Nuclear Geometry ◽  
Dimensional Surface

Molecular dynamics simulations often classically evolve the nuclear geometry on adiabatic potential energy surfaces (PESs), punctuated by random hops between energy levels in regions of strong coupling, in an algorithm...

scholarly journals Nonadiabatic Dynamics in Multidimensional Complex Potential Energy Surfaces

2020 ◽  
Author(s):  
Fábris Kossoski ◽  
Mario Barbatti
Keyword(s):  
Potential Energy ◽  
Potential Energy Surfaces ◽  
Electron Attachment ◽  
Complex Surface ◽  
Nonadiabatic Dynamics ◽  
Surface Hopping ◽  
Molecular Systems ◽  
Out Of Plane ◽  
Dynamics Simulations ◽  
Energy Surfaces

<p>Despite the continuous development of theoretical methodologies for describing nonadiabatic dynamics of molecular systems, there is a lack of approaches for processes where the norm of the wave function is not conserved, i.e., when an imaginary potential accounts for some irreversible decaying mechanism. Current approaches rely on building potential energy surfaces of reduced dimensionality, which is not optimal for more involving and realistic multidimensional problems. Here, we present a novel methodology for describing the dynamics of complex-valued molecular Hamiltonians, which is a generalisation of the trajectory surface hopping method. As a first application, the complex surface fewest switches surface hopping (CS-FSSH) method was employed to survey the relaxation mechanisms of the shape resonant anions of iodoethene. We have provided the first detailed and dynamical picture of the π*/σ* mechanism of dissociative electron attachment in halogenated unsaturated compounds, which is believed to underlie electron-induced reactions of several molecules of interest. Electron capture into the π* orbital promotes C=C stretching and out-of-plane vibrations, followed by charge transfer from the double bond into the σ* orbital at the C-I bond, and, finally, release of the iodine ion, all within only 15 fs. On-the-fly dynamics simulations of a vast class of processes can be envisioned with the CS-FSSH methodology, including autoionisation from transient anions, core-ionised and superexcited states, Auger and interatomic Coulombic decay, and time-dependent luminescence.</p>

scholarly journals Branching Corrected Mean Field Method for Nonadiabatic Dynamics

2020 ◽  
Author(s):  
Jiabo Xu ◽  
Linjun Wang
Keyword(s):  
Potential Energy Surfaces ◽  
Quantum Dynamics ◽  
High Efficiency ◽  
Mean Field ◽  
Field Method ◽  
Model Systems ◽  
Nonadiabatic Dynamics ◽  
Model Base ◽  
Dynamics Simulations ◽  
Energy Surfaces

When describing nonadiabatic dynamics based on trajectories, severe trajectory branching occurs when the nuclear wave packets on some potential energy surfaces are reflected while those on the remaining surfaces are not. As a result, the traditional Ehrenfest mean field (EMF) approximation breaks down. In this study, two versions of the branching corrected mean field (BCMF) method are proposed. Namely, when trajectory branching is identified, BCMF stochastically selects either the reflected or the non-reflected group to build the new mean field trajectory or splits the mean field trajectory into two new trajectories with the corresponding weights. As benchmarked in six standard model systems and an extensive model base with two hundred diverse scattering models, BCMF significantly improves the accuracy while retaining the high efficiency of the traditional EMF. In fact, BCMF closely reproduces the exact quantum dynamics in all investigated systems, thus highlighting the essential role of branching correction in nonadiabatic dynamics simulations of general systems.

Sign in / Sign up

Export Citation Format

Share Document