Many-body Green's function GW and Bethe-Salpeter study of the optical excitations in a paradigmatic model dipeptide

TL;DR

This study applies many-body Green's function GW and Bethe-Salpeter methods to accurately compute optical excitations in a model dipeptide, achieving results comparable to high-level quantum chemistry methods.

Contribution

The paper demonstrates a parameter-free GW and Bethe-Salpeter approach that accurately predicts excitation energies in a dipeptide, with improved agreement over standard DFT methods.

Findings

Bethe-Salpeter excitation energies agree within 0.07 eV with CASPT2 data.

Discrepancy of up to 0.5 eV for one charge-transfer state.

Method achieves 0.1 eV mean absolute error compared to LC-BLYP and CAM-B3LYP.

Abstract

We study within the many-body Green's function GW and Bethe-Salpeter formalisms the excitation energies of a paradigmatic model dipeptide, focusing on the four lowest-lying local and charge-transfer excitations. Our GW calculations are performed at the self-consistent level, updating first the quasiparticle energies, and further the single-particle wavefunctions within the static Coulomb-hole plus screened-exchange approximation to the GW self-energy operator. Important level crossings, as compared to the starting Kohn-Sham LDA spectrum, are identified. Our final Bethe-Salpeter singlet excitation energies are found to agree, within 0.07 eV, with CASPT2 reference data, except for one charge-transfer state where the discrepancy can be as large as 0.5 eV. Our results agree best with LC-BLYP and CAM-B3LYP calculations with enhanced long-range exchange, with a 0.1 eV mean absolute error. This has been achieved employing a parameter-free formalism applicable to metallic or insulating extended or finite systems.