A joint data and model driven method for study diatomic vibrational spectra including dissociation behavior

Achieving vibrational energies of diatomic systems with high quality by machine learning improved DFT method

“…where ħ = h/2p (h is the Planck constant), m is the reduced mass of two nucleus, J is total angular momentum quantum number, r is internuclear distance, and V(r) corresponds to the electrostatic interaction of all the particles. 23 One problem is that the exchange-correlation energy has no universal accurate form, 29 which introduces error to results. Finally, the ro-vibrational energy level E nJ can be expressed as…”

“…18,21,22 (2) Data driven algorithms, such as empirical potential energy function and direct parameter formulas for energy levels. 23,24 The accuracy of data driven algorithms are higher than ab initio methods except that they are only applicable to some molecular systems which have superior-quality experimental data.…”

Combining ab initio and machine learning method to improve the prediction of diatomic vibrational energies

“…Therefore, diatomic molecules energy structure are widely used in astrophysics, ultracold molecule, fundamental physical constants, chemical reaction manipulation and so forth [1][2][3][4][5][6][7][8][9]. From the computational perspective, there are two main strategies to obtain the vibrational energies: (1) ab initio method based on quantum mechanics principle deduction [10,11]; (2) data driven algorithm such as empirical potential energy function and direct parameter formulas for energy levels [12,13]. Ab initio method is universal, however, due to the complexity of quantum multi-body interaction (even diatomic molecules may involve hundreds of quantum objects), people have to model and approximate the wave function and Hamiltonian of the material according to the characteristics of the system, and make a subtle balance between the amount of calculation and the complexity of the model.…”

“…According to our knowledge, there is no effective method to uniformly pre evaluate the prediction performance of various ab initio calculation configurations, so the calculation for real system needs to go through a cumbersome trial and error process. For the calculation of vibrational levels of diatomic molecules, the uncertainty of ab initio results can be hundreds or even thousands cm À1 [12,17,18]. In order to improve prediction performance, traditional data-driven algorithm constructs a specific parameter model for a specific system, and extracts detail parameter information from experimental data of the research object.…”

“…In order to improve prediction performance, traditional data-driven algorithm constructs a specific parameter model for a specific system, and extracts detail parameter information from experimental data of the research object. It turns out that the data driven method can reduce the prediction error to less than 1 cm À1 [12]. However, due to the use of model and experimental data related to the target molecule, the scalability of present data driven method for diatomic vibrational energies prediction is limited and cannot be used for systems lacking experimental measurement.…”