Taylor A. Barnes

Embedded density functional theory for covalently bonded and strongly interacting subsystems.

Embedded density functional theory (e-DFT) is used to describe the electronic structure of strongly interacting molecular subsystems. We present a general implementation of the Exact Embedding (EE) method [J. Chem. Phys. 133, 084103 (2010)] to calculate the large contributions of the nonadditive kinetic potential (NAKP) in such applications. Potential energy curves are computed for the dissociation of Li(+)-Be, CH(3)-CF(3), and hydrogen-bonded water clusters, and e-DFT results obtained using the EE method are compared with those obtained using approximate kinetic energy functionals. In all cases, the EE method preserves excellent agreement with reference Kohn-Sham calculations, whereas the approximate functionals lead to qualitative failures in the calculated energies and equilibrium structures. We also demonstrate an accurate pairwise approximation to the NAKP that allows for efficient parallelization of the EE method in large systems; benchmark calculations on molecular crystals reveal ideal, size-independent scaling of wall-clock time with increasing system size.

Density functional theory embedding for correlated wavefunctions: Improved methods for open-shell systems and transition metal complexes

Density functional theory (DFT) embedding provides a formally exact framework for interfacing correlated wave-function theory (WFT) methods with lower-level descriptions of electronic structure. Here, we report techniques to improve the accuracy and stability of WFT-in-DFT embedding calculations. In particular, we develop spin-dependent embedding potentials in both restricted and unrestricted orbital formulations to enable WFT-in-DFT embedding for open-shell systems, and develop an orbital-occupation-freezing technique to improve the convergence of optimized effective potential calculations that arise in the evaluation of the embedding potential. The new techniques are demonstrated in applications to the van-der-Waals-bound ethylene-propylene dimer and to the hexa-aquairon(II) transition-metal cation. Calculation of the dissociation curve for the ethylene-propylene dimer reveals that WFT-in-DFT embedding reproduces full CCSD(T) energies to within 0.1 kcal/mol at all distances, eliminating errors in the dispersion interactions due to conventional exchange-correlation (XC) functionals while simultaneously avoiding errors due to subsystem partitioning across covalent bonds. Application of WFT-in-DFT embedding to the calculation of the low-spin/high-spin splitting energy in the hexaaquairon(II) cation reveals that the majority of the dependence on the DFT XC functional can be eliminated by treating only the single transition-metal atom at the WFT level; furthermore, these calculations demonstrate the substantial effects of open-shell contributions to the embedding potential, and they suggest that restricted open-shell WFT-in-DFT embedding provides better accuracy than unrestricted open-shell WFT-in-DFT embedding due to the removal of spin contamination.

Accurate and systematically improvable density functional theory embedding for correlated wavefunctions

We analyze the sources of error in quantum embedding calculations in which an active subsystem is treated using wavefunction methods, and the remainder using density functional theory. We show that the embedding potential felt by the electrons in the active subsystem makes only a small contribution to the error of the method, whereas the error in the nonadditive exchange-correlation energy dominates. We test an MP2 correction for this term and demonstrate that the corrected embedding scheme accurately reproduces wavefunction calculations for a series of chemical reactions. Our projector-based embedding method uses localized occupied orbitals to partition the system; as with other local correlation methods, abrupt changes in the character of the localized orbitals along a reaction coordinate can lead to discontinuities in the embedded energy, but we show that these discontinuities are small and can be systematically reduced by increasing the size of the active region. Convergence of reaction energies with respect to the size of the active subsystem is shown to be rapid for all cases where the density functional treatment is able to capture the polarization of the environment, even in conjugated systems, and even when the partition cuts across a double bond.

Accurate basis set truncation for wavefunction embedding

Density functional theory (DFT) provides a formally exact framework for performing embedded subsystem electronic structure calculations, including DFT-in-DFT and wavefunction theory-in-DFT descriptions. In the interest of efficiency, it is desirable to truncate the atomic orbital basis set in which the subsystem calculation is performed, thus avoiding high-order scaling with respect to the size of the MO virtual space. In this study, we extend a recently introduced projection-based embedding method [F. R. Manby, M. Stella, J. D. Goodpaster, and T. F. Miller III, J. Chem. Theory Comput. 8, 2564 (2012)] to allow for the systematic and accurate truncation of the embedded subsystem basis set. The approach is applied to both covalently and non-covalently bound test cases, including water clusters and polypeptide chains, and it is demonstrated that errors associated with basis set truncation are controllable to well within chemical accuracy. Furthermore, we show that this approach allows for switching between accurate projection-based embedding and DFT embedding with approximate kinetic energy (KE) functionals; in this sense, the approach provides a means of systematically improving upon the use of approximate KE functionals in DFT embedding.

Analyzing Performance of Selected NESAP Applications on the Cori HPC System

NERSC has partnered with over 20 representative application developer teams to evaluate and optimize their workloads on the Intel® Xeon Phi™Knights Landing processor. In this paper, we present a summary of this two year effort and will present the lessons we learned in that process. We analyze the overall performance improvements of these codes quantifying impacts of both Xeon Phi™architectural features as well as code optimization on application performance. We show that the architectural advantage, i.e. the average speedup of optimized code on KNL vs. optimized code on Haswell is about 1.1\(\times \). The average speedup obtained through application optimization, i.e. comparing optimized vs. original codes on KNL, is about 5\(\times \).