Direct evaluation of the force constant matrix in quantum Monte Carlo

Area of Science:

• Computational Chemistry
• Quantum Mechanics
• Materials Science

Background:

• Accurate prediction of molecular properties like atomic positions and vibrational frequencies is crucial in chemistry and materials science.
• Existing computational methods often face limitations in accuracy and efficiency for these calculations.

Purpose of the Study:

• To develop a novel formalism for directly evaluating the matrix of force constants within Quantum Monte Carlo (QMC) calculations.
• To apply this formalism to accurately relax atomic positions and determine vibrational modes in molecules.

Main Methods:

• Utilized a combination of variational and diffusion Monte Carlo methods.
• Developed a formalism to directly compute the matrix of force constants.
• Applied the method to four test molecules: hydrogen, hydrogen chloride, carbon dioxide, and methane.

Main Results:

• Computed bond lengths showed excellent agreement with experimental results, with deviations less than 0.007 Å.
• Fundamental vibrational frequencies for H2 and HCl were within 0.1% of experimental values, approximately 10 times more accurate than leading methods.
• For CO2 and CH4, vibrational frequencies were within 1.1% of experimental values, significantly outperforming RHF and DFT (PBE) and comparable or better than DFT (semi-empirical).

Conclusions:

• The developed QMC formalism provides a highly accurate and efficient method for calculating molecular force constants.
• This approach enables precise prediction of molecular geometry and vibrational spectra, advancing computational chemistry.
• The method offers a significant improvement over traditional computational techniques for molecular property prediction.