200728
本篇是应用数学之动态最优化理论的笔记,欢迎各位交流!今天是第一部分:确定性差分方程
本篇关于动态最优化的综合学习笔记。主要包括了离散与连续时间动态规划、连续时间最优控制与变分法等主题。本章仅给出在解决实际应用问题时的基本计算方法,对于数学上更进一步深入可以参考Stokey和Lucas(1989)。
x t = a x t − 1 + b t x_t = ax_{t-1} + b_t xt=axt−1+bt
x t = a x t − 1 x_t = ax_{t-1} xt=axt−1
x t g = x ˉ + c a t x_t^g = \bar{x} + ca^t xtg=xˉ+cat
x t = a n x t − n + ∑ i = 0 n − 1 a i b t − i x_{t}=a^{n} x_{t-n}+\sum_{i=0}^{n-1} a^{i} b_{t-i} xt=anxt−n+i=0∑n−1aibt−i
x t = c a t + ∑ i = 0 ∞ a i b t − i x_{t}=c a^{t}+\sum_{i=0}^{\infty} a^{i} b_{t-i} xt=cat+i=0∑∞aibt−i
x t = c a t − 1 a ∑ i = 0 ∞ ( 1 a ) i b t + 1 + i x_{t}=c a^{t}-\frac{1}{a} \sum_{i=0}^{\infty}\left(\frac{1}{a}\right)^{i} b_{t+1+i} xt=cat−a1i=0∑∞(a1)ibt+1+i
x t + 2 = a x t + 1 + b x t + d t + 2 x_{t+2}=a x_{t+1}+b x_{t}+d_{t+2} xt+2=axt+1+bxt+dt+2
[ x t + 1 y t + 1 ] = [ 0 1 b a ] [ x t y t ] \left[\begin{array}{l}x_{t+1} \\ y_{t+1}\end{array}\right]=\left[\begin{array}{ll}0 & 1 \\ b & a\end{array}\right]\left[\begin{array}{l}x_{t} \\ y_{t}\end{array}\right] [xt+1yt+1]=[0b1a][xtyt]
[ x t y t ] = [ e 11 e 12 e 21 e x 2 ] [ c 1 λ 1 t c 2 λ 2 t ] \left[\begin{array}{l}x_{t} \\ y_{t}\end{array}\right]=\left[\begin{array}{ll}e_{11} & e_{12} \\ e_{21} & e_{x 2}\end{array}\right]\left[\begin{array}{l}c_{1} \lambda_{1}^{t} \\ c_{2} \lambda_{2}^{t}\end{array}\right] [xtyt]=[e11e21e12ex2][c1λ1tc2λ2t]
z t + 1 = A z t + b z_{t+1}=A z_{t}+b zt+1=Azt+b
[ x t y t ] = [ e 11 e 12 e 21 e 22 ] [ c 1 λ 1 t c 2 λ 2 f ] + [ x ˉ y ˉ ] \left[\begin{array}{l}x_{t} \\ y_{t}\end{array}\right]=\left[\begin{array}{ll}e_{11} & e_{12} \\ e_{21} & e_{22}\end{array}\right]\left[\begin{array}{l}c_{1} \lambda_{1}^{t} \\ c_{2} \lambda_{2}^{\mathrm{f}}\end{array}\right]+\left[\begin{array}{l}\bar{x} \\ \bar{y}\end{array}\right] [xtyt]=[e11e21e12e22][c1λ1tc2λ2f]+[xˉyˉ]
x t − x ˉ = e 11 e 12 ( y t − y ˉ ) x_{t}-\bar{x}=\frac{e_{11}}{e_{12}}\left(y_{t}-\bar{y}\right) xt−xˉ=e12e11(yt−yˉ)
x t = f ( x t − 1 , y t − 1 ) y t = g ( x t − 1 , y t − 1 ) x_{t}=f\left(x_{t-1}, y_{t-1}\right)\\ y_{t}=g\left(x_{t-1}, y_{t-1}\right) xt=f(xt−1,yt−1)yt=g(xt−1,yt−1)
x ˉ = f ( x ˉ , y ˉ ) y ˉ = g ( x ˉ , y ˉ ) \bar{x}=f(\bar{x}, \bar{y})\\ \bar{y}=g(\bar{x}, \bar{y}) xˉ=f(xˉ,yˉ)yˉ=g(xˉ,yˉ)
x i + 1 − x ˉ = f x ( x ˉ , y ˉ ) ( x i − x ˉ ) + f y ( x ˉ , y ˉ ) ( y t − y ˉ ) y t + 1 − y ˉ = g x ( x ˉ , y ˉ ) ( x t − x ˉ ) + g y ( x ˉ , y ˉ ) ( y t − y ˉ ) x_{i+1}-\bar{x}=f_{x}(\bar{x}, \bar{y})\left(x_{i}-\bar{x}\right)+f_{y}(\bar{x}, \bar{y})\left(y_{t}-\bar{y}\right)\\ y_{t+1}-\bar{y}=g_{x}(\bar{x}, \bar{y})\left(x_{t}-\bar{x}\right)+g_{y}(\bar{x}, \bar{y})\left(y_{t}-\bar{y}\right) xi+1−xˉ=fx(xˉ,yˉ)(xi−xˉ)+fy(xˉ,yˉ)(yt−yˉ)yt+1−yˉ=gx(xˉ,yˉ)(xt−xˉ)+gy(xˉ,yˉ)(yt−yˉ)
A = [ f x f y g x g y ] A=\left[\begin{array}{ll}f_{x} & f_{y} \\ g_{x} & g_{y}\end{array}\right] A=[fxgxfygy]