APS Logo

Recent Developments in the Implementation of the RT-EOM-CC Green's Function Approach

ORAL

Abstract

Many-body excitations in X-ray photoemission spectra (XPS) are difficult to simulate from first principles. We have recently developed[1] a non-linear cumulant real-time equation-of-motion coupled-cluster (RT-EOM-CC) Green's function method that provides a non-perturbative framework for treating these problems in molecular systems. Here we report the implementation of RT-EOM-CC with single and double excitations within the Tensor Algebra for Many-body Methods (TAMM) infrastructure. TAMM is a massively parallel heterogeneous tensor library designed for utilizing soon-to-be-available exascale computing resources. The two-body electron repulsion matrix elements are Cholesky-decomposed and imposed spin-explicit form while evaluating tensor contractions of matrix elements involving various operators. This new implementation also improves performance by directly supporting complex algebra unlike our earlier Tensor Contraction Engine (TCE) implementation. Other recent improvements include a novel, variable time approach for the solution of the recursive equations present in the first-order Adams-Moulton method used to propagate the time-dependent amplitudes. These improvements are demonstrated by a study of the XPS of the formaldehyde and ethyltrifluoroacetate (ESCA) molecules. Simulations of the latter are challenging, as they involve as many as 71 occupied and 649 virtual orbitals, yet the relative ionization potentials are in good agreement with experiment.

[1] "Real-Time Equation-of-Motion CCSD Cumulant Green's Function", FD Vila et al., J. Chem. Theory Comput. (doi:10.1021/acs.jctc.1c01179).

Presenters

  • Fernando D Vila

    University of Washington, U of Washington

Authors

  • Fernando D Vila

    University of Washington, U of Washington

  • Himadri Pathak

    PNNL

  • Ajay Panyala

    PNNL

  • Bo Peng

    Pacific Northwest National Laboratory, PNNL

  • Nicholas P Bauman

    PNNL

  • Erdal Mutlu

    PNNL

  • John J Rehr

    University of Washington, U of Washington

  • Karol Kowalski

    PNNL