Bulik Ireneusz W, Frisch Michael J, Vaccaro Patrick H
Department of Chemistry , Yale University , New Haven , Connecticut 06511 , United States.
Gaussian, Inc. , 340 Quinnipiac St. Bldg. 40 , Wallingford , Connecticut 06492 , United States.
J Chem Theory Comput. 2018 Mar 13;14(3):1554-1563. doi: 10.1021/acs.jctc.8b00016. Epub 2018 Mar 2.
We present a new approach to importance sampling in diffusion Monte-Carlo (DMC) simulations of vibrational excited states whereby the trial wave functions for low-energy states are incorporated into the diffusion equations so as to enforce their orthogonality. For the model systems examined here, simple variational wave functions based on the vibrational self-consistent field (VSCF) and the simplest vibrational configuration interaction (VCI) are effective in importance sampling provided that internal coordinates used in the underlying one-particle functions have been variationally optimized. The resulting model yields results comparable in accuracy to the best unguided DMC calculations without requiring an a priori choice of coordinates to specify nodal hyperplanes.
我们提出了一种在振动激发态的扩散蒙特卡罗（DMC）模拟中进行重要性抽样的新方法，即将低能态的试探波函数纳入扩散方程，以确保它们的正交性。对于此处研究的模型系统，基于振动自洽场（VSCF）和最简单的振动组态相互作用（VCI）的简单变分波函数在重要性抽样中是有效的，前提是基础单粒子函数中使用的内坐标已进行变分优化。由此产生的模型所得到的结果在精度上与最佳的无引导DMC计算相当，且无需事先选择坐标来指定节点超平面。
