Large chemical systems with several centres of interest (such as a solid with multiple defects, double adsorption on a surface, or co-doped materials) are often a challenge to solve via conventional computational chemistry methods due to a requirement of a large system to be treated quantum chemically, and large active space within the system for a fairly accurate representation of the chemistry. The cost of such calculations increases exponentially with the size of active space, limiting the tractable system size. The Density Matrix Embedding Theory (DMET)[1, 2] works on the principle of isolating a small centre of interest within a large structure and using mean field potential for connecting impurity with the environment, allowing us to focus on the centre of interest and thus reducing the cost of the method when working with systems with specific centres of interest. This method has also been extended to periodic systems [3, 4]. The cost of performing DMET calculations alone would still increase with the size of center of interest, especially significant for two or more centers of interest. An extension of DMET is Localized Active Space (LAS)[5] method which works on the principle of solving two or more non-interacting active spaces separately (for example, unconjugated pi systems) and helps in getting approximate multiconfigurational solutions at reduced costs. Here, for systems with multiple centres of interest, combining DMET and LAS offers promise, especially when searching for multiconfigurational solutions. This is because DMET doesn’t require us to treat the whole system with an expensive method and adding more active spaces to LAS is a linear scaling operation, allowing us to treat more impurities at a much lower cost.

Localized active spaces within an embedded density Matrix method

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.