An improved molecular partitioning scheme for numerical quadratures in density functional theory

Area of Science:

• Computational chemistry
• Quantum chemistry
• Molecular modeling

Background:

• Accurate calculation of molecular energies is crucial for understanding chemical reactions and properties.
• Existing molecular partitioning schemes, like Becke's, face challenges with accuracy for weakly bound systems and computational efficiency for large molecules.

Purpose of the Study:

• To present a modified molecular partitioning scheme that improves accuracy for weakly bound complexes.
• To develop a faster, linear-scaling grid generation method without cutoff errors.
• To assess the performance of the new scheme for atomization and interaction energies and its scalability for large systems.

Main Methods:

• Modification of Becke's molecular partitioning scheme.
• Development of a novel grid generation algorithm with linear scaling.
• Testing the scheme on atomization energies of small molecules and interaction energies of van der Waals complexes.
• Evaluation of computational efficiency and scaling for systems up to 1707 atoms.

Main Results:

• The modified partitioning scheme demonstrates significantly improved accuracy for weakly bound complexes.
• The new grid generation is faster and exhibits linear scaling.
• No cutoff error was introduced by the new partitioning or grid generation methods.
• The method's efficiency and scaling were validated on large molecular systems.

Conclusions:

• The presented modification offers a more accurate and efficient approach to molecular partitioning.
• This advancement is particularly beneficial for studying weakly bound complexes and large molecular systems.
• The linear-scaling grid generation contributes to computational speedup in quantum chemistry calculations.