ITR: Computational Design of Strongly Correlated Materials Based on a Combination of the Dynamical Mean Field and the GW Methods

University Of California-Davis, Davis CA

Investigators

Abstract

This award was made on a 'small' category proposal submitted in response to the ITR solicitation, NSF-02-168. It supports collaborative computational and theoretical research between groups at New Jersey Institute of Technology and Rutgers University through award #0312478 that aims for a more realistic theory of strongly correlated electron materials. The PIs aim to construct a computational approach for the study, design and visualization of properties of materials containing strongly correlated electron systems. The theoretical foundations of this work are based on a non-perturbative many-body method involving on a combination of dynamical mean field and GW theories, which can yield material-specific predictions and interpretation of properties of solids. The PIs' objectives are to: (a) implement this approach using the high-performance, all-electron, full-potential, relativistic linear-muffin-tin orbital (LMTO) code for crystals, slabs, and periodic polymers called "LMTART;" (b) enhance performance so that Green functions, self-energies, and polarization operators on the frequency axis can be handled for complicated systems with many atoms per unit cell; (c) design and implement user-friendly interfaces and visualization capabilities for calculations of correlated electronic systems, creating a fast, powerful, database enabled and Web integrated Material Information and Design Laboratory (MINDLab) for the benefit and use in physics, material science, engineering, and educational communities; (e) test and apply this information technology enabled quantum many-body theory tool by tackling frontier problems of material science such as computational design of magnetic semiconductors and interpretation of de Haas van Alphen experiments in heavy fermion systems. MINDLab would enhance the infrastructure for research and education; it has the potential to advance discovery and understanding of materials while promoting teaching, training and learning through powerful visualization techniques. %%% This award was made on a 'small' category proposal submitted in response to the ITR solicitation, NSF-02-168. It supports collaborative computational and theoretical research between groups at Rutgers University and New Jersey Institute of Technology through award #0342290 that aims for a more realistic theory of strongly correlated electron materials. Strongly correlated electron materials display unusual phenomena such as high-temperature superconductivity, colossal magnetoresistance, giant optical non-linearities and large thermoelectric coefficients. These systems are at the frontier of materials science, and the variety of behavior they exhibit as well as their complexity makes their study intellectually challenging, and the prospects for applications exciting. The PIs aim to construct a computational approach for the study, design and visualization of properties of materials containing strongly correlated electron systems To tackle the complexity of real materials new theoretical methods, algorithms, and computer programs will be developed. By means of these novel information technology tools for computation and data generation, technologically relevant compounds containing many atoms per unit cell may be studied at a fundamental level while also including important material-specific detail. Data visualization enables access to more abstract theoretical quantities required to capture the physics of electronic correlation. The PIs' objectives include the design and implementation of a computational tool for correlated electronic systems, a fast, powerful, database enabled and Web integrated Material Information and Design Laboratory (MINDLab). MINDLab would enhance the infrastructure for research and education; it has the potential to advance discovery and understanding of materials while promoting teaching, training and learning through powerful visualization techniques. ***