Abstract:

A new pattern recognition technique PCA-BPN(principal component analysis-back propagation network) has been used to assign the unknown electronic configurations of odd-parity energy levels of the first spectrum of curium (Cm I ). The obtained results show that (1) most previous predictions given by KNN(K nearest neighbours) and CPN(counter propagation network) are further confirmed;(2) several energy levels, which could not be clearly assigned by KNN etc., are predicted to be in good agreement with the assignments of the CPN;(3) two energy levels which were wrongly predicted by the CPN are now corrected using the PCA-BPN and the new assignments are supported by the traditional pattern recognition technique, PCA-NLM(principal component analysis nonlinear mapping).

Key words: Cm I odd parity spectrum, Classification of energy levels, Pattern recognition, PCA-BP neural network, Nonlinear mapping