Scribd
Upload
23 views

17 Random Vectors 2 Lecture

Uploaded by

nay33n
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
23 views

17 Random Vectors 2 Lecture

Uploaded by

nay33n
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 49

Spring 2024 / 430.

314

Random vectors
Part II

Jonghyun Choi
Associate Professor, SNU ECE

Slides referred from Youngmin Kim and Sunghyun Choi with partial permission
Today
• Functions of random vectors
• Expected value of the random vectors and correlation matrix
• Gaussian random vectors
Functions of random vectors
• Theorem 5.6
Example 5.8
Example 5.8
Min and Max function of a random vector
• Theorem 5.7
Min and Max function of a random vector
• Theorem 5.7
Expectation of a function of a random vector
• Theorem 5.8
Expectation of a function of an independent
random variable
• Theorem 5.9
CDF and PDF of a derived random vector 1
Derivation is for each 𝑘th 𝑋 and 𝑌

• Theorem 5.10
CDF and PDF of a derived random vector 2
Derivation is for 𝑌 having the relationship across all components in 𝑋

• Theorem 5.11
Quiz 5.5 (A)
Quiz 5.5 (A)
Quiz 5.5 (A)
Quiz 5.5 (B)
Quiz 5.5 (B)
Expectation of a random vector
• Theorem 5.11
Expectation of a random matrix
• Definition 5.12 For a random matrix 𝑨 with the random variable 𝐴$% as
its 𝑖, 𝑗th element, 𝐸[𝑨] is a matrix with 𝑖, 𝑗th element of 𝐸[𝐴$% ].
Second order statistics of a random vector
1. Vector (auto-)correlation
Definition 5.13 The correlation of a random vector 𝑿 is an 𝑛 × 𝑛 matrix
𝑹𝑿 with 𝑖, 𝑗th element 𝑹𝑿 (𝑖, 𝑗) = 𝐸[𝑋$ 𝑋% ]. In vector notation,
𝑹𝑿 = 𝐸[𝑿𝑿’].
2. Vector (auto-)covariance
Definition 5.14 The covariance of a random vector 𝑿 an 𝑛 × 𝑛 matrix 𝑪𝑿
with 𝑖, 𝑗th element 𝐶' (𝑖, 𝑗) = 𝐶𝑜𝑣[𝑋$ , 𝑋% ]. In vector notation,
𝑪𝑿 = 𝐸[(𝑿 − 𝝁𝑿 )(𝑿 − 𝝁𝑿 )’].
Relationship between vector correlation and
vector covariance
• Theorem 5.12 For a random vector 𝑿 with correlation matrix 𝑹𝑿 ,
covariance matrix 𝑪𝑿 , and vector expectation 𝜇' ,
𝑪𝑿 = 𝑹𝑿 − 𝝁𝑿 𝝁(𝑿 .
Remind
Relationship between vector correlation and
vector covariance (cont’d)
• Theorem 5.12 For a random vector 𝑿 with correlation matrix 𝑹𝑿 ,
covariance matrix 𝑪𝑿 , and vector expectation 𝜇' ,
𝑪𝑿 = 𝑹𝑿 − 𝝁𝑿 𝝁(𝑿 .
Example 5.12
Example 5.12
Example 5.12
Vector cross-correlation / cross-covariance
• Vector cross-correlation
Definition 5.15 The cross-correlation of random vectors, 𝑿 with 𝑛 components
and 𝒀 with 𝑚 components, is an 𝑛 × 𝑚 matrix 𝑹𝑿𝒀 with 𝑖, 𝑗th element
𝑅𝑿𝒀 (𝑖, 𝑗) = 𝐸[𝑋# 𝑌$ ], or, in vector notation
𝑹𝑿𝒀 = 𝐸 𝑿𝒀% .
• Vector cross-covariance
Definition 5.16 The cross-covariance of a pair of random vectors, 𝑿 with 𝑛
components and 𝒀 with 𝑚 components, is an 𝑛 × 𝑚 matrix 𝑪𝑿𝒀 with 𝑖, 𝑗th
element 𝐶𝑿𝒀 (𝑖, 𝑗) = 𝐶𝑜𝑣[𝑋# , 𝑌$ ], or, in vector notation
𝑪𝑿𝒀 = 𝐸 (𝑿 − 𝝁𝑿 )(𝒀 − 𝝁𝒀 )′ .
Derived random vectors
• Theorem 5.13
Derived random vectors
• Theorem 5.13
Example 5.13
𝑪𝑿 =
Example 5.13
𝑪𝑿 =
If 𝒀 is an affine function of 𝑿
• Theorem 5.14
Example 5.14
Example 5.14 𝑪𝑿 =

From example 5.13

Quiz 5.6 is left for your exercise.


Gaussian random vector
• Definition 5.17
Independent Gaussian and diagonal 𝑪𝑿
• Theorem 5.15 A Gaussian random vector 𝑿 has independent
components if and only if 𝑪𝑿 is a diagonal matrix
Independent Gaussian and diagonal 𝑪𝑿
• Theorem 5.15 A Gaussian random vector 𝑿 has independent
components if and only if 𝑪𝑿 is a diagonal matrix
Example 5.15
Example 5.15
Example 5.15
Example 5.15
If 𝒀 is an affine function of a Gaussian 𝑿
• Theorem 5.16
If 𝒀 is an affine function of a Gaussian 𝑿
• Theorem 5.16
If 𝒀 is an affine function of a Gaussian 𝑿
• Theorem 5.16
Example 5.16
Example 5.16
Standard normal (Gaussian) random vector
• Definition 5.18 The 𝑛-dimensional standard normal random vector 𝒁 is
the 𝑛-dimensional Gaussian random vector with 𝐸[𝒁] = 0 and 𝑪𝒁 = 𝑰.
If 𝑿 is an affine function of a standard Gaussian 𝒁
• Theorem 5.18
Factorization of a covariance matrix of
Gaussian vector 𝑿
• Theorem 5.19 For a Gaussian vector 𝑿 with covariance 𝑪𝑿 , there always
exists a matrix 𝑨 such that 𝑪𝑿 = 𝑨𝑨’.
Quiz 5.7
Summary
• Functions of random vectors
• Expected value of the random vectors and correlation matrix
• Gaussian random vectors

Next: Sums of random variables


• Moment generating functions (MGF)
• Expected values of sums
• PDF of the sum of two random variables
• MGF of the sum of independent random varaibles

You might also like