Building a New Platform for Significantly Improving Performance of Hartree-Fock and CCSD(T) Correlation Energy Based...

Area of Science:

• Computational chemistry
• Quantum chemistry
• Theoretical chemistry

Background:

• High computational cost in gold standard coupled cluster theory arises from slow convergence of electronic energies with basis set size.
• Extrapolation methods are crucial for achieving high-quality results cost-effectively by predicting energies at the complete basis set limit.
• Accuracy of existing extrapolation formulas depends heavily on empirical parameters fitted to limited data sets and smaller basis sets.

Purpose of the Study:

• To develop an improved benchmark platform for reoptimizing extrapolation parameters.
• To enhance the accuracy of established energy extrapolation formulas.
• To provide a valuable resource for computational chemists.

Main Methods:

• Construction of a new benchmark platform with 183 diverse species (open-shell, closed-shell, ionic, neutral).
• Inclusion of larger basis sets up to aug-cc-pV6Z for data generation.
• Reoptimization of empirical parameters for ten published extrapolation formulas using the new platform.

Main Results:

• Newly optimized parameters significantly improve the energy-predictive abilities of ten published formulas.
• All ten formulas demonstrated similar performance with reoptimized parameters on the new platform.
• Development of an online calculator for practical application of these extrapolation schemes.

Conclusions:

• The enhanced benchmark platform and reoptimized parameters address deficiencies in previous extrapolation methods.
• This work provides more accurate and reliable energy extrapolation for coupled cluster theory.
• The developed online calculator facilitates wider adoption and application of these improved methods.