求状态转移概率矩阵

大海3米深

已于 2022-01-20 09:08:52 修改

阅读量915

点赞数

分类专栏：数学文章标签：矩阵

于 2022-01-17 15:19:55 首次发布

本文链接：https://blog.csdn.net/weixin_41440200/article/details/122538672

版权

数学专栏收录该内容

1 篇文章 0 订阅

订阅专栏

和式

$\arg \min \varepsilon =\sum_{i=1}^{N}\left \| Tx_{i}-x_{i+1} \right \|^{2}$

$d\varepsilon =d\sum_{i=1}^{N}\left ( Tx_{i}-x_{i+1} \right )\left ( Tx_{i}-x_{i+1} \right )^{T}$

$d\varepsilon =Trd\varepsilon =Trd\sum_{i=1}^{N}\left ( Tx_{i}x_{i}^{T}T^{T}-Tx_{i}x_{i+1}^{T}-x_{i+1}x_{i}^{T}T^{T}+x_{i+1}x_{i+1}^{T} \right )$

$=\sum_{i=1}^{N}\left \{ Trd\left (Tx_{i}x_{i}^{T}T^{T}-Tx_{i}x_{i+1}^{T}-x_{i+1}x_{i}^{T}T^{T}+x_{i+1}x_{i+1}^{T} \right ) \right \}$

$=\sum_{}^{} \left \{ 2x_{i}x_{i}^{T}dT-Tr\left ( dTx_{i}x_{i+1}^{T}\right )-Tr\left ( x_{i+1}x_{i}^{T}d\left ( T^{T} \right ) \right ) \right \}$

$=\sum_{}^{} \left \{ 2x_{i}x_{i}^{T}dT-x_{i}x_{i+1}^{T}dT-Tr\left ( x_{i+1}x_{i}^{T}d\left ( T \right )^{T} \right ) \right \}$

$=\sum_{}^{} \left \{ 2x_{i}x_{i}^{T}dT-x_{i}x_{i+1}^{T}dT-Tr\left ( dTx_{i}x_{i+1}^{T} \right ) \right \}$

$=\sum_{}^{} \left \{ 2x_{i}x_{i}^{T}dT-2x_{i}x_{i+1}^{T}dT \right \}$

令偏导数等于0，则

$0=\frac{\partial \varepsilon }{\partial T}=\sum_{}^{} \left \{ 2x_{i}x_{i}^{T}-2x_{i+1}x_{i}^{T} \right \}=\sum_{}^{} \left \{ 2x_{i}x_{i}^{T} \right \}-\sum_{}^{} \left \{ 2x_{i+1}x_{i}^{T} \right \}$

矩阵式

$\varepsilon =\left \| TX-Y \right \|^{2}$

$\varepsilon =\left ( TX-Y \right )\left ( TX-Y \right )^{T} =TXX^{T}T^{T}-TXY^{T} -YX^{T}T^{T} +YY^{T}$

$\frac{\partial \varepsilon }{\partial T}=2XX^{T}T-2Y^{T}X=0$

$T=\left ( XX^{T} \right )^{-1}Y^{T}X$

大海3米深

关注

0
点赞
踩
0

收藏

觉得还不错? 一键收藏
0
评论
求状态转移概率矩阵

和式矩阵式
复制链接

扫一扫

专栏目录

求状态转移概率矩阵

和式

矩阵式

“相关推荐”对你有帮助么？