解方程 { d i + x i + λ ϕ ( x i ) Φ ( x i ) = 0 , i = 1 , ⋯ , p ∏ i = 1 p Φ ( x i ) = r \left\{\begin{aligned} &d_i+x_i+\lambda\frac{\phi(x_i)}{\Phi(x_i)}=0,i=1,\cdots,p\\ &\prod_{i=1}^p\Phi(x_i)=r \end{aligned} \right. ⎩⎪⎪⎪⎪⎨⎪⎪⎪⎪⎧di+xi+λΦ(xi)ϕ(xi)=0,i=1,⋯,pi=1∏pΦ(xi)=r
其中 r r r, d i d_i di已知, ϕ ( ⋅ ) \phi(⋅) ϕ(⋅)标准正态密度函数, Φ ( ⋅ ) \Phi(⋅) Φ(⋅)表示标准正态分布函数.
未知参数是 x i x_i xi, λ λ λ, i = 1 , 2 , ⋯ , p i=1,2,\cdots,p i=1,2,⋯,p
Newton法程序如下:
library(MASS)
# F是以(f_1,...,f_{p+1})为梯度向量的函数
# 求(f_1,...,f_{
p+1})的根等价于求F的极小值
Phi = function(z) pnorm(z)
phi = function(z) exp(-0.5*z^2)/sqrt(2*pi)
d_phi = function(z) -z*exp(-0.5*z^2)/sqrt(2*pi)
grad_F = function(d,r,x,lambda,p){
u = 0
for(i in 1:p)
u[i] = d[i] + x[i] + lambda * phi(x[i]) / Phi(x[i])
prod = 1
for(i in 1:p)
prod = prod * Phi(x[i])
u[p+1] = prod - r
return(u)
}
Hesse_F = function(d,r,x