Systematic discrepancies between reference methods for noncovalent interactions within the S66 dataset

Area of Science:

• Computational chemistry
• Quantum mechanics
• Electronic structure theory

Background:

• Accurate modeling of noncovalent interactions is crucial for diverse chemical applications.
• Quantum diffusion Monte Carlo (DMC) and coupled cluster theory with single, double, and perturbative triple excitations [CCSD(T)] are established methods for noncovalent interactions.
• Discrepancies exceeding 7.5 kcal/mol have been observed between DMC and CCSD(T) for larger systems, with origins and scale remaining unclear.

Purpose of the Study:

• To systematically investigate the discrepancies between DMC and CCSD(T) for noncovalent interactions in medium-sized complexes.
• To identify factors influencing the magnitude and direction of these discrepancies.
• To provide benchmark systems for future method development.

Main Methods:

• Utilized advanced Quantum diffusion Monte Carlo (DMC) to calculate interaction energies for the S66 dataset.
• Employed CCSD(T) as a reference method for comparison.
• Performed energy decomposition analysis to correlate discrepancies with electrostatic and dispersion contributions.

Main Results:

• DMC predicts stronger binding than CCSD(T) for electrostatic-dominated systems.
• DMC predicts weaker binding than CCSD(T) for dispersion-dominated systems.
• The magnitude of the discrepancy correlates with the ratio of electrostatic to dispersion interactions.

Conclusions:

• Identified specific systems, acetic acid dimer (ID 20) and uracil-cyclopentane dimer (ID 42), exhibiting prominent discrepancies.
• These findings highlight the need for further development in electronic structure methods.
• The identified model systems offer cost-effective benchmarks for validating and improving DMC and CCSD(T).