cartesius fort - object fortran Library for Chemistry and Materials Science

Abstract

Modeling of structure and properties of molecules and materials (crystals/solids) on the basis of their electronic structure is one of the most important consumers of computer resources (processor time, memory and storage). The known attempts to improve its efficiency reduce to massive parallelization. This approach ignores enormous diversity of types of structures and behaviors of molecules and materials. Moreover, this diversity is by no means reflected in the paradigm currently dominating the field of molecular/material modeling. Much more efficient is, of course, a thorough analysis of the physical conditions occurring in different molecules/materials. On this way we could successfully build a series of efficient methods targeted upon specific classes of molecules/materials: inorganic ones with open d-shells and organic ones featuring local two-center bonds and developed conjugated$$\uppi $$ -systems (generalized chromophores). The experience gained formulates as a new concept of semi-empirism: that is selecting the electronic wave function of a system under study as a product of the wave functions of the chromophores present in the system. This called for a new development: of a library of objects representing different types of chromophores to be freely combinable to represent an arbitrary molecule/material so that its respective parts (chromophores) are modeled by the most efficient method suitable for the specific type of the chromophore and taking into account the interactions between them. Apparently, the deep segmentation of the system achieved within the new concept of semi-empirism allows for the efficient parallelization and more efficient usage of the HPC software.