Modelling potential energy surfaces for small clusters using Shepard interpolation with Gaussian-form nodal...

Area of Science:

• Computational Chemistry
• Atmospheric Chemistry
• Physical Chemistry

Background:

• Potential energy surfaces (PESs) are crucial for understanding chemical systems.
• Analytical PESs are vital for theoretical and environmental studies of atmospheric clusters.
• Shepard interpolation is a common method for generating analytical PESs using ab initio data.

Purpose of the Study:

• To develop a new, efficient method for constructing analytical potential energy surfaces (PESs).
• To improve the accuracy and reduce the computational cost of PES generation for atmospheric clusters.
• To introduce novel nodal functions and sampling strategies within the Shepard interpolation framework.

Main Methods:

• Developed a modified Shepard interpolation technique using a combination of symmetric Gaussian and asymmetric exponential terms for nodal functions.
• Created corresponding sampling methodologies tailored to the new interpolation approach.
• Applied the method to various atmospheric bimolecular clusters, including Ar-rigid H2O, Ne-rigid CO2, and N2-rigid CO2.

Main Results:

• Achieved root mean square errors (RMSE) below 0.13 kJ mol-1 for Ar-rigid H2O and Ne-rigid CO2 using only 150 samples.
• Obtained RMSE below 0.39 kJ mol-1 for N2-rigid CO2 with 1800 samples.
• Demonstrated the method's effectiveness and efficiency in accurately representing PESs of atmospheric clusters.

Conclusions:

• The novel Shepard interpolation method offers a significant improvement for generating analytical potential energy surfaces.
• The approach provides accurate PES representations with reduced computational effort, beneficial for atmospheric chemistry research.
• This work facilitates more efficient theoretical investigations of atmospheric cluster dynamics and properties.