Ultrafast Correlation Energy Estimator

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Molecular Modeling

Background:

• Accurate calculation of electron correlation energy is crucial for predicting molecular properties.
• High-accuracy methods like coupled-cluster singles, doubles, and perturbative triples (CCSD(T)) are computationally expensive.
• Developing efficient methods for correlation energy estimation remains a key challenge in computational chemistry.

Purpose of the Study:

• To introduce a virtually no-cost computational method for determining correlation energies.
• To achieve near-exact accuracy (99.5%) compared to CCSD(T)/CBS benchmarks.
• To partition correlation energy into chemically intuitive components: bonds and lone pairs.

Main Methods:

• Developed a Correlation Energy Per Bond (CEPB) approach.
• Assigned partial correlation energies to bond types and lone pairs.
• Calibrated CEPB using CCSD(T)/CBS benchmark data.

Main Results:

• The CEPB method achieves 99.5% accuracy relative to CCSD(T)/CBS values for correlation energies.
• CEPB is applicable to general organic and inorganic molecules, including conjugated systems.
• Reaction energy accuracy rivals more expensive methods like Møller-Plesset perturbation theory.
• The method is primarily suited for near-equilibrium geometries.

Conclusions:

• The CEPB method provides a highly accurate and computationally inexpensive way to estimate molecular correlation energies.
• The findings suggest the existence of compact, chemically intuitive molecular fragments for energy partitioning.
• This opens avenues for developing ultrafast correlation energy estimators for various applications.