优化量子Monte　Carlo计算中的波函数

Optimized Wavefunctions for Quantum Monte Carlo Calculations

提出了一个优化量子Ｍｏｎｔｅ Ｃａｒｌｏ波函数的新方法，与前人的方法相比，它不使用变分原理而是极小化薛定谔方程的剩余量；不按ψ２取样而是按ψ２（ＥＬ─Ｅ）２取样；不用差分而是用分析导数；不使用传统的速降法而是使用一个步长自动调节的下降法，它具有拟牛顿性质，因而是超线性收敛的．Ｈ２Ｏ，ＣＨ劝和Ｆ２分子的检验结果表明，本文提出的这一优化量子Ｍｏｎｔｅ Ｃａｒｌｏ波函数的新方法是非常成功的．

in this paper we have proposed a new algorithm for optimizing quantum Monte Carlo wave function. Unlike previous methods the surplus of the Schrodinger equation is minimized instead of using variational principle; sampling according to the distribution ψ2 (EL-E)2 instead of ψ2; the improved steepest descent technique is adopted, in which the step size is automatically adjustable. The procedure is quasi-Newton and converges superlinearly; in the procedure we use an analytical derivatives as the coefficients instead of the differences. The test results of H2O, CH4 and F2 show that the novel optimization procedure proposed in the present paper is very successful.