-
Notifications
You must be signed in to change notification settings - Fork 25
/
Copy pathexercises.tex
131 lines (113 loc) · 8.55 KB
/
exercises.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
\documentclass[10pt]{article}
\usepackage{bm}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage[paperheight=26 true cm,paperwidth=18.4 true cm,
top=2.6 true cm,bottom=2.2 true cm,left=1.8 true cm,right=1.8 true cm]{geometry}
\title{Exercises of State Estimation for Robotics}
% put your name here
\author{Xiang Gao}
\begin{document}
\maketitle
\section{Chapter 1}
Chapter 1 does not have any exercises.
\section{Chapter 2}
\subsection{ex 2.1}
\begin{equation}
{\mathbf{u}^T}\mathbf{v} = \sum\limits_{i = 1}^n {{u_i}{v_i}} = \mathrm{tr}\left( {\mathbf{v}{\mathbf{u}^T}} \right)
\end{equation}
\subsection{ex 2.2}
\begin{equation}
\begin{split}
H\left( {\mathbf{x},\mathbf{y}} \right) &= - \iint {p\left( {\mathbf{x},\mathbf{y}} \right)\ln }p\left( {\mathbf{x},\mathbf{y}} \right)\mathrm{d}\mathbf{x}\mathrm{d}\mathbf{y} \\
&= - \iint {p\left( \mathbf{x} \right)p\left( \mathbf{y} \right)\left( {\ln p\left( \mathbf{x} \right) + \ln p\left( \mathbf{y} \right)} \right)\mathrm{d}\mathbf{x}\mathrm{d}\mathbf{y}} \\
&= - \int {p\left( \mathbf{x} \right)\ln p\left( \mathbf{x} \right)\underbrace {\left( {\int {p\left( \mathbf{y} \right)\mathrm{d}\mathbf{y}} } \right)}_1\mathrm{d}\mathbf{x} - \int {p\left( \mathbf{y} \right)\ln p\left( \mathbf{y} \right)\underbrace {\left( {\int {p\left( \mathbf{x} \right)\mathrm{d}\mathbf{x}} } \right)}_1\mathrm{d}\mathbf{y}} } \\
&= H\left( \mathbf{x} \right) + H\left( \mathbf{y} \right) \\
\end{split}
\end{equation}
\subsection{ex 2.3}
\begin{equation}
\begin{split}
E\left( {\mathbf{x}{\mathbf{x}^T}} \right) &= E\left( {\left( {\mathbf{x} - \boldsymbol{\mu} + \boldsymbol{\mu} } \right){{\left( {\mathbf{x} - \boldsymbol{\mu} + \boldsymbol{\mu} } \right)}^T}} \right) \hfill \\
&= E\left( {\underbrace {\left( {\mathbf{x} - \boldsymbol{\mu} } \right){{\left( {\mathbf{x} - \boldsymbol{\mu} } \right)}^T}}_{\boldsymbol{\Sigma}} + \underbrace {\boldsymbol{\mu} {{\left( {\mathbf{x} - \boldsymbol{\mu} } \right)}^T}}_{\boldsymbol{\mu} {\mathbf{0}^T}} + \underbrace {\left( {\mathbf{x} - \boldsymbol{\mu} } \right){\boldsymbol{\mu} ^T}}_{\mathbf{0}{\boldsymbol{\mu} ^T}} + \boldsymbol{\mu} {\boldsymbol{\mu} ^T}} \right) \hfill \\
&= \boldsymbol{\Sigma} + \boldsymbol{\mu} {\boldsymbol{\mu} ^T} \hfill \\
\end{split}
\end{equation}
\subsection{ex 2.4}
The integrate of an odd function is zero in the symmetric interval.
\begin{equation}
\begin{split}
E\left(\mathbf{x} \right) &= \int_{-\infty}^{+\infty}\mathbf{x}p\left(\mathbf{x}\right)\mathrm{d}\mathbf{x} \hfill \\
&= \int_{-\infty}^{+\infty}\frac{\mathbf{x}}{\sqrt{(2\pi)^N\text{det}\boldsymbol{\Sigma}}}\text{exp}\left(-\frac{1}{2}(\mathbf{x}-\boldsymbol{\mu})^T\boldsymbol{\Sigma}^{-1}(\mathbf{x}-\boldsymbol{\mu})\right)\mathrm{d}\mathbf{x} \hfill \\
&=
\int_{-\infty}^{+\infty}\frac{\mathbf{y}+\boldsymbol{\mu}}{\sqrt{(2\pi)^N\text{det}\boldsymbol{\Sigma}}}\text{exp}\left(-\frac{1}{2}\mathbf{y}^T\boldsymbol{\Sigma}^{-1}\mathbf{y}\right)\mathrm{d}\mathbf{y} \hfill \\
&=
\underbrace{\int_{-\infty}^{+\infty}\frac{\mathbf{y}}{\sqrt{(2\pi)^N\text{det}\boldsymbol{\Sigma}}}\text{exp}\left(-\frac{1}{2}\mathbf{y}^T\boldsymbol{\Sigma}^{-1}\mathbf{y}\right)\mathrm{d}\mathbf{y}}_{0} +\hfill \int_{-\infty}^{+\infty}\frac{\boldsymbol{\mu}}{\sqrt{(2\pi)^N\text{det}\boldsymbol{\Sigma}}}\text{exp}\left(-\frac{1}{2}\mathbf{x}^T\boldsymbol{\Sigma}^{-1}\mathbf{x}\right)\mathrm{d}\mathbf{x} \hfill
\\&=
\boldsymbol{\mu}\underbrace{\int_{-\infty}^{+\infty}\frac{1}{\sqrt{(2\pi)^N\text{det}\boldsymbol{\Sigma}}}\text{exp}\left(-\frac{1}{2}\mathbf{y}^T\boldsymbol{\Sigma}^{-1}\mathbf{y}\right)\mathrm{d}\mathbf{y}}_{\mathbf{y}\sim\boldsymbol{N}(0,\boldsymbol{\Sigma})} \hfill
\\&=
\boldsymbol{\mu}\underbrace{\int_{-\infty}^{+\infty}p\left(\mathbf{y}\right)\mathrm{d}\mathbf{y}}_{1} \hfill
\\&= \boldsymbol{\mu}
\end{split}
\end{equation}
\subsection{ex 2.5}
\begin{equation}
\begin{split}
E\left( {\left( {\mathbf{x}-\boldsymbol{\mu}} \right){{\left( {\mathbf{x}-\boldsymbol{\mu}} \right)}^T}} \right) &= \int_{ - \infty }^\infty {\left( {\mathbf{x}-\boldsymbol{\mu}} \right){{\left( {\mathbf{x}-\boldsymbol{\mu}} \right)}^T}\frac{1}{{\sqrt {(2\pi)^N } \det \boldsymbol{\Sigma} }}\exp \left( { - \frac{1}{2}\left( {\mathbf{x}-\boldsymbol{\mu}} \right){\boldsymbol{\Sigma} ^{ - 1}}{{\left( {\mathbf{x}-\boldsymbol{\mu}} \right)}^T}} \right)\mathrm{d}\mathbf{x}} \\
&= \frac{1}{{\sqrt {(2\pi)^N } \det \boldsymbol{\Sigma} }}\underbrace{\int_{ - \infty }^\infty d\left( { - \boldsymbol{\Sigma} \left( {\mathbf{x}-\boldsymbol{\mu}} \right)\exp \left( { - \frac{1}{2}\left( {\mathbf{x}-\boldsymbol{\mu}} \right){\boldsymbol{\Sigma} ^{ - 1}}{{\left( {\mathbf{x}-\boldsymbol{\mu}} \right)}^T}} \right)} \right)}_{\mathrm{odd, = 0}} \\
& \quad \quad + \frac{1}{{\sqrt {(2\pi)^N } \det \boldsymbol{\Sigma} }}\int_{ - \infty }^\infty {\boldsymbol{\Sigma} \exp \left( { - \frac{1}{2}\left( {\mathbf{x}-\boldsymbol{\mu}} \right){\boldsymbol{\Sigma} ^{ - 1}}{{\left( {\mathbf{x}-\boldsymbol{\mu}} \right)}^T}} \right)\mathrm{d}\mathbf{x}} \\
&= \mathbf{0} + \boldsymbol{\Sigma} = \boldsymbol{\Sigma}
\end{split}
\end{equation}
\subsection{ex 2.6}
Take $\ln()$ on both sides, the right side will be:
\begin{equation}
\begin{split}
& \ln \prod\limits_{k = 1}^K {\exp \left( {\frac{1}{2}{{\left( {{\mathbf{x}_k} - {\boldsymbol{\mu} _k}} \right)}^T}\boldsymbol{\Sigma} _k^{ - 1}\left( {{\mathbf{x}_k} - {\boldsymbol{\mu} _k}} \right)} \right)} \\
=&\sum\limits_{k = 1}^K {\exp \left( {\frac{1}{2}{{\left( {{\mathbf{x}_k} - {\boldsymbol{\mu} _k}} \right)}^T}\boldsymbol{\Sigma} _k^{ - 1}\left( {{\mathbf{x}_k} - {\boldsymbol{\mu} _k}} \right)} \right)} \\
=&\frac{1}{2}\left( {\sum\limits_{k = 1}^K {\mathbf{x}_k^T\boldsymbol{\Sigma} _k^{ - 1}{\mathbf{x}_k}} + \sum\limits_{k = 1}^K {2\boldsymbol{\mu} _k^T\boldsymbol{\Sigma} _k^{ - 1}{\mathbf{x}_k}} + \sum\limits_{k = 1}^K {\boldsymbol{\mu} _k^T\boldsymbol{\Sigma} _k^{ - 1}{\boldsymbol{\mu} _k}} } \right)
\end{split}
\end{equation}
Compare left side with the right side regarding of $\mathbf{x}$, we have:
\begin{equation}
{\boldsymbol{\Sigma} ^{ - 1}} = \sum\limits_{k = 1}^K {\boldsymbol{\Sigma} _k^{ - 1}} , \quad {\boldsymbol{\Sigma} ^{ - 1}}\boldsymbol{\mu} = \sum\limits_{k = 1}^K {\boldsymbol{\Sigma} _k^{ - 1}\boldsymbol{\mu} }
\end{equation}
\subsection{ex 2.7}
\begin{equation}
\begin{split}
\mathrm{Cov}(\mathbf{x}) &= E((\mathbf{x} - \boldsymbol{\mu} ){(\mathbf{x} - \boldsymbol{\mu} )^T})\\
&= E\left( {\sum\limits_{k = 1}^K {\left( {{w_k}{\mathbf{x}_k} - {w_k}{\boldsymbol{\mu} _k}} \right)} \sum\limits_{k = 1}^K {{{\left( {{w_k}{\mathbf{x}_k} - {w_k}{\boldsymbol{\mu} _k}} \right)}^T}} } \right)\\
&= E\left( {\sum\limits_{k = 1}^K {w_k^2\left( {{\mathbf{x}_k} - {\boldsymbol{\mu} _k}} \right){{\left( {{\mathbf{x}_k} - {\boldsymbol{\mu} _k}} \right)}^T}} + \sum\limits_{m=1,n=1,m\ne n}^K {{w_m}{w_n}\left( {{\mathbf{x}_m} - {\boldsymbol{\mu} _m}} \right){{\left( {{\mathbf{x}_n} - {\boldsymbol{\mu} _n}} \right)}^T}} } \right)\\
&= \sum\limits_{k = 1}^K {w_k^2\mathrm{Cov}\left( {{\mathbf{x}_k}} \right)} + \underbrace {E\left( {\sum\limits_{m=1,n=1,m\ne n}^K {{w_m}{w_n}\left( {{\mathbf{x}_m} - {\boldsymbol{\mu} _m}} \right){{\left( {{\mathbf{x}_n} - {\boldsymbol{\mu} _n}} \right)}^T}} } \right)}_{\mathrm{independent} , = 0}\\
&= \sum\limits_{k = 1}^K {w_k^2\mathrm{Cov}\left( {{\mathbf{x}_k}} \right)}
\end{split}
\end{equation}
\subsection{ex 2.8}
Expectation:
\begin{equation}
E\left( y \right) = E\left( {\sum\limits_{k = 1}^K {x_k^2} } \right) = K
\end{equation}
Covariance:
\begin{equation}
\mathrm{Cov}\left( y \right) = E\left( {{{\left( {y - K} \right)}^2}} \right) = E\left( {{y^2} - 2Ky + {K^2}} \right) = E\left( {{y^2}} \right) - 2{K^2} + {K^2}
\end{equation}
For $E(y^2)$, we have:
\begin{equation}
E\left( {{y^2}} \right) = E\left( {{{\left( {{\mathrm{x}^T}\mathrm{x}} \right)}^2}} \right) = E\left( {\sum\limits_{k = 1}^K {x_k^4} + \sum\limits_{m = 1,n = 1,m \ne n}^K {x_m^2x_n^2} } \right)
\end{equation}
Use Isserlis to deal with the fourth-order items:
\begin{equation}
E\left( {\sum\limits_{k = 1}^K {x_k^4} } \right) = 3\sum\limits_{k = 1}^K {\underbrace {E\left( {x_k^2} \right)}_1E\left( {x_k^2} \right)} = 3K
\end{equation}
and
\begin{equation}
\begin{split}
E\left( {\sum\limits_{m = 1,n = 1,m \ne n}^K {x_m^2x_n^2} } \right) &= \sum\limits_{m = 1,n = 1,m \ne n}^K {\underbrace {E\left( {x_m^2} \right)}_1E\left( {x_n^2} \right) + 2\underbrace {E\left( {{x_m}{x_n}} \right)}_0E\left( {{x_m}{x_n}} \right)} \\
&= K\left( {K - 1} \right) = {K^2} - K
\end{split}
\end{equation}
Take them together:
\begin{equation}
\mathrm{Cov}\left( y \right) = 3K + {K^2} - K - 2{K^2} + {K^2} = 2K
\end{equation}
\end{document}