Optimization of fragment state spaces within the excitonic renormalization framework.

Area of Science:

• Quantum Chemistry
• Electronic Structure Theory
• Computational Physics

Background:

• The excitonic renormalization framework offers an alternative to traditional electronic structure methods for weakly interacting systems.
• Current methods rely on localized orbitals and fragment-based correlated states but lack efficient bottom-up procedures for model space optimization.

Purpose of the Study:

• To develop an efficient bottom-up algorithm for generating optimized model state spaces within the excitonic renormalization framework.
• To complete the excitonic renormalization methodology with a scalable, polynomially scaling computational framework.

Main Methods:

• An algorithm utilizing monomer gradients at three levels and determinant space pre-screening was developed.
• This approach ensures compact model state spaces and intermediates without building the full Hamiltonian.
• The method was tested on the beryllium dimer.

Main Results:

• The algorithm successfully generated compact model state spaces, closely matching optimal ones.
• Optimization at zeroth order of the Hamiltonian expansion accurately recovered first-order results.
• This enables efficient optimization at lower computational cost.

Conclusions:

• The presented algorithm completes the excitonic renormalization methodology, establishing a polynomially scaling framework.
• This advancement facilitates efficient and accurate electronic structure calculations for complex molecular systems.