ProgramMaster Logo
Conference Tools for 2018 TMS Annual Meeting & Exhibition
Register as a New User
Submit An Abstract
Propose A Symposium
Presenter/Author Tools
Organizer/Editor Tools
About this Abstract
Meeting 2018 TMS Annual Meeting & Exhibition
Symposium Computational Method and Experimental Approaches for Model Development and Validation, Uncertainty Quantification, and Stochastic Predictions
Presentation Title Machine Learning Based Atomistic Force Fields
Author(s) Rampi Ramprasad, Venkatesh Botu, Rohit Batra, James Chapman , Huan Doan Tran
On-Site Speaker (Planned) Rampi Ramprasad
Abstract Scope A general strategy to create machine learning (ML) based atomistic force fields is outlined. In contrast to conventional force fields, ML force fields can be systematically improved by adding progressively new training data (representing new atomic environments). Our ML scheme involves (1) preparing a big reference dataset of atomic environments and forces with sufficiently low noise, e.g., using density functional theory or higher-level methods, (2) utilizing a generalizable class of structural fingerprints for representing atomic environments, (3) optimally selecting diverse and non-redundant training datasets from the reference data, and (4) suitable learning approaches to predict atomic forces directly (and rapidly) from atomic configurations. From the atomistic forces, accurate potential energies can then be obtained by appropriate integration along a reaction coordinate or along a molecular dynamics trajectory. The concept is demonstrated for the case of Al, including extensive tests and molecular dynamics studies of diverse phenomena.
Proceedings Inclusion? Planned: Supplemental Proceedings volume


Bayesian Linear Regression and Kriging Methods for Uncertainty Quantification in Process-structure-property Linkages of Low Carbon Steels and Superalloys
Benchmarking Density Functional Theory Based Methods to Predict Optical and Electronics Properties of 2H-TaX2 (X=S, Se)
Calibration of a Titanium Modified Embedded Atom Method Potential to High Temperature Behavior
Correlations of Numerical Precision in Material Properties Derived from Density Functional Theory
Development of a Semi-empirical Potential for Simulation of Ni Solutes Segregated in Ag Grain Boundaries
Dynamic Failure of High Energy Materials: Uncertainty Quantification and Stochastic Predictions
Errors of Molecular Dynamics Simulations, and Development of “Accurate” Analytical Bond Order Potentials for Al-Cu-H and Mg-H Systems
Extending the Angular-embedded Atom Method (A-EAM) Framework to an Al-Mg-Si Ternary System
Extending the Reach of DFT to Molecular Simulations Using Neural Networks
It's a SNAP: Automated Generation of High-accuracy Interatomic Potentials Using Quantum Data
Large Scale Sensitivity of Uncertain Parameters on Optimal Control Solutions: An Example in Additive Manufacturing
Lattice Thermal Conductivity: Uncertainty Quantification in First Principles Predictions and Experimental Validation
Linear Scaling, Quantum-accurate Interatomic Potentials with SNAP; Reaching those Hard-to-reach Places in Classical Molecular Dynamics
Machine Learning Based Atomistic Force Fields
Machine Learning Methods for Interatomic Potentials: Application to Boron Carbide
Machine Learnt Interatomic Potentials for Stanene and Germanene to Study Thermal Conductivity and Growth
New Advances in Semi-empirical Interatomic Potentials - the Modified Embedded Atom Method (MEAM)
Overcoming Singularities within Rate-independent Crystal Plasticity to Enable Realistic Latent Hardening
Parametrically Homogenized Models of Deformation and Failure of Metals and Alloys with Uncertainty-quantification
Property Localization: Quantifying the Uncertainty of Inferred Constitutive Models for Grain Boundaries
The Current State of Phase Field Benchmark Problems Developed by CHiMaD/NIST
The OpenKIM Testing Framework for Interatomic Potentials
The Role of Data Analysis in Uncertainty Quantification: Examples from Materials Science
Uncertainty Quantification for Additive Manufacturing Applications across Scales
Uncertainty Quantification for Solute Transport Modeling
Uncertainty Quantification in Materials Strength Models Using Bayesian Inference
Uncertainty Quantification of the Effect of Charge Noise on Silicon Quantum Dots
Utilizing Error in First-principle Lattice Constants to Discover Novel Low-dimensional Materials

Questions about ProgramMaster? Contact