Atomistic simulations in materials science and nanotechnology play a key role in many scientific and industrial applications. However,the predictive capability of these approaches hinges on the accuracy of the mathematical models used to describe atomic interactions. Modern models are optimized (fit) to reproduce quantum mechanical values for the forces and energies of representative atomic configurations deemed important for the problem of interest. However, no standardized approach currently exists for quantifying the range of applicability of an interatomic model or estimating the accuracy of its predictions. The result is that the field of atomistic modeling struggles with the unknown and uncontrolled capacity of its models to predict phenomena outside the fitting database, keeping this field from fully realizing its scientific and technological potential. This problem will be addressed by creating the Knowledge-base of Interatomic Models (KIM): an interactive, self-extending, database of interatomic models, self-contained simulation codes that test the predictions of these models, and reference data. This online resource will allow users to rapidly compare model predictions with reference data, to generate new predictions by uploading their own tests, and to download models conforming to community standards developed as part of this project. The critical mass of models and tests gathered in KIM from diverse scientific disciplines will then be used to develop a quantitative theoretical framework for evaluating the accuracy and precision of interatomic models, which together define their transferability. These transformative advances will provide, for the first time, rational guidelines for selecting appropriate interatomic models for given applications and will define a fundamentally new atomistic simulation methodology that provides error estimates for computed properties.
This project aims to answer the question: When and to what extent can we believe the results of atomistic simulations of materials? The project's objectives are of central concern to an unusually large cross-section of the industrial and scientific communities who are interested in understanding materials from their basic building blocks; this includes physicists, materials scientists, chemists, and engineers from academia, government, and industry. The KIM project should initiate a transformative shift in the way researchers think about and perform atomistic materials simulations. The result will be more precise and accurate predictions of materials behavior that will allow for faster and cheaper discovery, design, and optimization of new, specialized technologically-useful materials. The creation of the KIM system will provide unprecedented standardized access to the state-of-the-art in atomistic modeling and simulation. This access will break down the barrier-to-entry to this field for traditionally underrepresented groups and institutions around the world and facilitate the efforts of industry in the U.S. and internationally to use interatomic models to advance their technological goals. Furthermore, the development of a rigorous methodology of assessing the transferability and accuracy of interatomic models will bring about a paradigm shift in how models are developed, selected, and used. The KIM project will help to train the next generation of scientists and engineers by providing educational experience for post-doctoral, graduate, and undergraduate students at the PIs' home institutions, as well as by conducting educational tutorials and workshops at popular materials conferences and other venues. Finally, the KIM project will strive to recruit and engage minority and traditionally underrepresented scientists and engineers.
