Fast Estimation of Møller-Plesset Correlation Energies Based on Atomic Contributions

Area of Science:

• Computational chemistry
• Quantum chemistry

Background:

• Dynamic correlation is crucial for accurate chemical compound calculations.
• Existing methods for dynamic correlation can be computationally intensive.

Purpose of the Study:

• To introduce a fast and concise model for estimating dynamic correlation energy.
• To enable practical applications like reaction energy prediction.

Main Methods:

• Decomposition of dynamic correlation energy contributions by atomic regions.
• Utilizing Møller-Plesset perturbation (MPn) theory (MP2, MP2.5, MP4) as reference.
• Calculation of correlation energy densities at multiple theory levels.

Main Results:

• A novel model for rapid dynamic correlation energy estimation.
• Demonstrated applicability across various Møller-Plesset perturbation levels.
• The model's potential as a baseline or pre-trained model for machine learning.

Conclusions:

• The proposed model offers a concise and fast method for dynamic correlation energy estimation.
• It shows promise for practical applications in computational chemistry, including reaction energy predictions.
• The model can serve as a foundation for future machine learning-driven studies.