- 作者: 張志強
- 作者服務機構: 國立成功大學化學系
- 中文摘要: 用牛頓一拉法遜法尋求最優軌域時其疊代漸近過程常會碰到發散困難,而用高級組態疊加法則不會有收歛上的困難,本文分析造成發散困難的原因及指出過去研究者使用方法上的錯誤。
- 英文摘要: The reported results on the convergence behavior of the direct orbital optimization methods wereinconsistent due to the interference effect. When the interference can be inactivated, the super CI methodconverges at least as good as the Newton-Raphson method. An argument based on the perturbation theoryis given to show that if the normalization constraint has been imposed, the second order perturbed functionis not needed in calculating the energy through second order. Therefore, up to second order the energycalculated by the SCI method is as accurate as that by the Newton-Raphson method, and the SCI iterativeprocess should also be assumed as a quadratically convergent one.
- 中文關鍵字: orbital optimization; variational mothod; perturbation method
- 英文關鍵字: --