Deep neural network for fitting analytical potential energy curve of diatomic molecules from ro-vibrational spectra

We present a new approach which employs deep neural network to obtain parameters of analytical representation potential energy curve diatomic molecule. test the find spectroscopic characteristics for ground X2Σ+ electronic state MgF molecule based on experimental energies ro-vibrational transitions. The result shows that can be applied in characterisation interatomic Our is competitive with those obtained using other methods tested, i.e. shallow and so-called brute force method.