Capturing intensive and extensive DFT/TDDFT molecular properties with machine learning
Capturing intensive and extensive DFT/TDDFT molecular properties with machine learning
复制标题
DOI：
10.1140/epjb/e2018-90148-y
复制
发表时间：
2018-08-06
影响因子：
1.6
通讯作者：
Mueller, Klaus-Robert
中科院分区：
文献类型：
--
作者：
Pronobis, Wiktor;Schuett, Kristof T.;Mueller, Klaus-Robert
机器学习已成功应用于预测小有机分子的化学性质，如能量或极化率。与这些性质相比，电子激发能提出了一个更具挑战性的学习问题。在这里，我们研究了两种现有的机器学习方法的适用性，从含时密度泛函理论的激发能的预测。为此，我们系统地研究了各种2体和3体描述符以及深度神经网络SchNet的性能，以预测广泛和密集的性质，例如从基态到第一和第二激发态的跃迁能。正如预期的那样，当前最先进的机器学习技术更适合预测广泛而不是密集的数量。我们推测需要开发全球性的描述符，可以描述广泛和密集的属性在平等的基础上。
Machine learning has been successfully applied to the prediction of chemical properties of small organic molecules such as energies or polarizabilities. Compared to these properties, the electronic excitation energies pose a much more challenging learning problem. Here, we examine the applicability of two existing machine learning methodologies to the prediction of excitation energies from time-dependent density functional theory. To this end, we systematically study the performance of various 2- and 3-body descriptors as well as the deep neural network SchNet to predict extensive as well as intensive properties such as the transition energies from the ground state to the first and second excited state. As perhaps expected current state-of-the-art machine learning techniques are more suited to predict extensive as opposed to intensive quantities. We speculate on the need to develop global descriptors that can describe both extensive and intensive properties on equal footing.