Speaker
Description
Considering anharmonic effects is essential to accurately describe vibra-
tional, spectroscopic and thermodynamic properties of molecular systems. Re-
cently, it has been shown that introducing anharmonic formalism into the quantum
cluster equilibrium (QCE) program improves the results of the calculations [1].
It is possible to calculate anharmonic frequencies [2]. However, such calculations are
tedious and computationally expensive, therefore it is of prime interest to
apply machine learning techniques to overcome this task. In this work we study
a variety of molecular clusters of different sizes consisting of HX, CH3X, C2H5X
(X = F, Cl, Br) monomers. Our dataset features consist of normal mode coordi-
nates as well as harmonic frequencies, anharmonic frequencies and intensities of
the fundamental and first overtone modes. Symmetry and structural descriptors
such as internal coordinates are also taken into account. Anharmonicity con-
stants are extracted from the vibrational energy levels of the Morse oscillator.
Later on, various machine learning algorithms are applied for the classifica-
tion and regression purposes. Our results show that while stretching vibration
modes have positive anharmonicity constants, the low frequency bending and
torsional modes often exhibit negative values. With this we were able to classify
the anharmonicity constants according to the type of vibration such as stretch-
ing, bending, or internal rotations and translations. Regression algorithms were
able to provide an estimate of the anharmonic frequencies.
References:
[1] J. Chem. Phys. 146, 124114 (2017)
[2] J. Chem. Phys. 105, 10332 (1996)
