Research
Quantum-Chemical Method Developments
Solving the many-electron correlation problem in molecules is central to the field of Molecular Quantum Mechanics. Electron correlation plays a crucial role in the accurate theoretical prediction of molecular energies, spectroscopic properties and noncovalent interactions in molecular clusters and extended systems.
We develop wave function based ab initio quantum theories to obtain highly accurate, yet computationally affordable solutions to the many-electron correlation problem. We work primarily in the expanse of Coupled-Cluster Theory (CCT), Many-Body Perturbation Theory (MBPT), and allied many-body methods. These methods are combined with molecular fragmentation techniques and local-correlation approaches to extend applicability to large molecular systems.
We apply our methods to:
- Molecules with complex electronic structure, such as those containing one or more unpaired electrons, e.g., radicals generated in photochemical reactions as well as transition metal complexes containing metal ions with partially filled d orbitals.
- Chemical processes in excited electronic states, short-lived intermediates.
- Prediction of spectroscopic parameters, such as, EPR, Mössbauer, NMR, etc.
Quantum Chemistry Program Development
We develop scalable High-Performance Computing (HPC) algorithms for the existing and new quantum-chemical theories for applications to present-day challenging problems in chemistry. Real-world chemical problems involve large molecules consisting of up to a few hundred atoms. To enable applications of accurate electron correlation approaches for quantum-chemical simulations involving such systems, we develop parallel algorithms that can effectively utilize the distributed multi-node architectures of modern supercomputers, thereby scaling down the computational costs and reducing the computing time. In addition, we develop algorithms to program graphics processing units (GPUs) to achieve many-folded acceleration (speedup) of chemical computations.
Our programs are implemented within the GAMESS (General Atomic and Molecular Electronic Structure System) quantum-chemistry package (click here for information on obtaining GAMESS)
as well as within the CFOUR (Coupled-Cluster techniques for Computational Chemistry) program package (click here for information).
Programs are written primarily in Fortran (95/03) and Python.