Dr. A. Eugene DePrince IIIAssistant Professor
PhD – University of Chicago (2009)
Research InterestResearch in our group focuses on the development and application of highly accurate many-body quantum chemistry methods, with a specific focus on the design and implementation of efficient algorithms for modern multicore CPUs and many-core GPUs.
Time-dependent quantum chemistry:
Light interactions with quantum systems are usually treated computationally by time-dependent perturbation theory, and electric fields are assumed to be of low intensity. However, a much more flexible description of light-matter interactions may be obtained, albeit at an increased computational cost, through explicitly time-dependent methods in which one propagates the electronic Schrodinger equation in the presence of a time-dependent perturbation. Such methods can describe perturbations of arbitrary strength and shape, and they correctly recover properties obtained by the usual linear-response methods when considering weak perturbations. We use such real-time time-dependent methods to describe very fast electron dynamics relevant to a number of chemical phenomena including the optical control of molecular devices and the emergence of collective behavior in quantum systems.
Low-rank tensor factorizations:
Approximate tensor factorizations are increasingly common in electronic structure theory as a means to both accelerate computations and eliminate the storage or generation of the 4-index electron repulsion integral (ERI) tensor. The most well recognized of these integral approximations is known as density fitting (DF) [or the resolution of the identity (RI)], and a similar approach involves the partial Cholesky decomposition (CD) of the ERI tensor. More exotic representations of the ERIs such as tensor hypercontraction (THC) density fitting and related Parafac/Candecomp decompositions have recently emerged, but many of these methods are in their infancy. We are interested in developing new implementations of popular electronic structure methods based on these integral factorizations. Our implementation of the DF/CD-CCSD(T) method is currently available in the open-source Psi4 electronic structure package.
GPU quantum chemistry:
Graphics processing units (GPUs) and other coprocessors (e.g. Intel MIC) are revolutionizing high-performance scientific computing. These devices deliver a huge number of floating-point operations at much lower cost (both physical cost and power consumption) than conventional multicore CPUs. To leverage this enormous potential in some meaningful way, however, one must usually abandon legacy algorithms in favor of new ones that account for the many annoying peculiarities of GPU programming. Our group develops new implementations of high-accuracy many-body methods for use in heterogenous computing environments where a standard compute node consists of a modern multicore CPU and at least one GPU.
|A. E. DePrince III and C. David Sherrill, J. Chem. Theory Comput. 9, 2687 (2013). "Accuracy and Efficiency of Coupled-Cluster Theory Using Density Fitting/Cholesky Decomposition, Frozen Natural Orbitals, and a t1-Transformed Hamiltonian"|
|A. E. DePrince III and C. David Sherrill, J. Chem. Theory Comput. 9, 293 (2013). "Accurate noncovalent interaction energies using truncated basis sets based on frozen natural orbitals"|
|A. E. DePrince III, M. Pelton, J. R. Guest, and S. K. Gray, Phys. Rev. Lett. 107, 196806 (2011). "Emergence of excited-state plasmon modes in linear hydrogen chains from time-dependent quantum mechanical methods"|
|A. E. DePrince III and J. R. Hammond, J. Chem. Theory Comput. 7, 1287 (2011). "Coupled cluster theory on graphics processing units I: The coupled cluster doubles method"|