Quantum embedding of multi-orbital fragments using the Block-Householder-transformation
Recently, some of the authors introduced the use of the Householder transformation as a simple and intuitive method for the embedding of local molecular fragments (see Sekaran et. al., Phys. Rev. B 104, 035121 (2021), and Sekaran et. al., Computation 10, 45 (2022)). In this work, we present an extension of this approach to the more general case of multi-orbital fragments using the block version of the Householder transformation applied to the one-body reduced density matrix, yet unlocking the applicability to general quantum chemistry/condensed-matter physics Hamiltonians. A step by step construction of the Block-Householder transformation is presented. Both physical and numerical interest of the approach are highlighted. The specific mean-field (non-interacting) case is thoroughly detailed as it is shown that the embedding of a given $N$ spin-orbitals fragment leads to the generation of two separated sub-systems: a $2N$ spin-orbitals "fragment+bath" cluster that exactly contains $N$ electrons, and a remaining cluster's "environment" which is described by so-called core electrons. We illustrate the use of this transformation in different cases of embedding {scheme} for practical applications. We particularly focus on the extension of the previously introduced Local Potential Functional Embedding Theory (LPFET) and Householder-transformed Density Matrix Functional Embedding Theory (Ht-DMFET) to the case of multi-orbital fragments. These calculations are realized on different types of systems such as model Hamiltonians (Hubbard rings) and \textit{ab initio} molecular systems (hydrogen rings).