Estimates of electron correlation based on density expansions.

Area of Science:

• Quantum Chemistry
• Computational Chemistry
• Molecular Modeling

Background:

• Estimating correlation energy is crucial for accurate molecular electronic structure calculations.
• Existing methods often lack efficiency or require extensive computational resources.
• Partitioning correlation energy into atomic contributions offers a promising avenue for simplification.

Purpose of the Study:

• To develop and evaluate two new methods for estimating molecular correlation energy.
• To assess the accuracy of these methods across a diverse range of molecules.
• To explore an extension of these methods for studying molecular dissociation.

Main Methods:

• Partitioning molecular correlation energy into atomic contributions.
• Method 1: Expanding electron density using atomic contributions and electron repulsion bounds.
• Method 2: Associating correlation contributions with basis function pairs, determining atom-specific correlation factors via configuration interaction (CI) calculations.

Main Results:

• The proposed methods provide accurate correlation energy estimates with an average error of 2.6% compared to high-level CI calculations.
• The methods were tested on 27 molecules with varied bonding environments.
• An extension involving truncated CI and atomic correlation factors was successfully applied to molecular dissociation.

Conclusions:

• The atomic partitioning approach offers a reliable and efficient way to estimate molecular correlation energy.
• The developed methods demonstrate good predictive power for various molecular systems.
• The extension shows potential for studying dynamic correlation effects in chemical processes like dissociation.