Scribd
Upload
205 views5 pages

Multivariate Normal Distribution Basics

1. The document describes properties of multivariate normal distributions including means, variances, covariance matrices, and conditional distributions. 2. It provides examples calculating probabilities for various events involving a multivariate normal random vector X with specified mean and covariance matrix. 3. The examples include finding probabilities that a linear combination of components of X is above or below certain values, and calculating a conditional distribution.

Uploaded by

d
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
205 views5 pages

Multivariate Normal Distribution Basics

1. The document describes properties of multivariate normal distributions including means, variances, covariance matrices, and conditional distributions. 2. It provides examples calculating probabilities for various events involving a multivariate normal random vector X with specified mean and covariance matrix. 3. The examples include finding probabilities that a linear combination of components of X is above or below certain values, and calculating a conditional distribution.

Uploaded by

d
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

STAT 410

Fall 2016

X1

X2
X =

...

X
n

2
E( X ) = X =
...

n-dimensional
random vector

Covariance matrix:

11 12

22

X = 21
...
...

n1 n 2

ij = Cov ( Xi , Xj )

... 1n

... 2n
...

... nn

X = E[ ( X ) ( X ) T ]
X symmetric, nonnegative-definite

n=1

n=2

X = (X)

X = 1
X 2


X = 1
2

X = ( )

X = ( 2 )

2
Cov(X 1 , X 2 )
Var (X 1 )
1
=
X =

(
)
(
)
Cov
X
,
X
Var
X
1
2
2

1 2

Let Y = A X + b,

where

1 2

A m n (non-random) matrix,
b Rm (non-random) vector

Then

a1

a 2
a =
...

a
n

E( Y ) = Y = A X + b

Y = A X AT

a1 X1 + a2 X2 + + an Xn = aT X
E( aT X ) = aT X

Var ( a T X ) = a T X a

22

1.

X1
5


Consider a random vector X = X 2 with mean E ( X ) = = 1
X
3
3

2 3
9

and variance-covariance matrix Cov ( X ) = = 2


4 2 .
3 2 16

Var ( X 1 ) = 9,

Then

12 =

2
9 4

1
,
3

13 =

Var ( X 2 ) = 4,
3
9 16

Var ( X 3 ) = 16,

1
,
4

23 =

2
4 16

1
.
4

X
1
Consider 2 X 1 3 X 2 X 3 = ( 2 3 1 ) X 2 .
X
3

E( 2 X1 3 X2 X3 ) =

(2

3 1 ) = 2 1 3 2 3 = 2 5 3 1 1 3 = 4.

2 3 2
9

4 2 3
Var ( 2 X 1 3 X 2 X 3 ) = ( 2 3 1 ) 2
3 2 16 1

= ( 15 6 16 ) 3 = 64.
1

OR

Var ( 2 X 1 3 X 2 X 3 ) = 4 Var ( X 1 ) + 9 Var ( X 2 ) + Var ( X 3 )


12 Cov ( X 1 , X 2 ) 4 Cov ( X 1 , X 3 ) + 6 Cov ( X 2 , X 3 )
= 36 + 36 + 16 24 + 12 12 = 64.

Multivariate Normal Distribution:


X ~ N n ( , )

fX(x) =

(2p )n 2

12

exp (x )' 1 (x )
2

Z ~ N n ( 0, In )

fZ(z) =

n 1
1

1
exp z' z =
exp zi2
(2p )n 2 2 i =1 2p 2

X = 1/2 Z +

M X ( t ) = exp t ' + t ' t


2

Amn

Y = AX + b

b Rm

Y ~ N m ( A + b, A A' )

X1

X2

X =

t1

t2
t = Rn
...

t
n

X1 is of dimension m < n
X2 is of dimension n m

1
1
X1 | X2 ~ N m 1 + 12 22
(X 2 2 ), 11 12 22
21

11

21

12

22

2.
X ~ N 3 ( , )

a)

4 1 0

= 1 4 2
0
2 9

5

= 3
7

Find P ( X 1 > 6 ).
X1 ~ N ( 5, 4 )

6 5
= P ( Z > 0.5 ) = 0.3085.
P ( X1 > 6 ) = P Z >
4

b)

Find P ( 5 X 2 + 4 X 3 > 70 ).
E ( 5 X 2 + 4 X 3 ) = 5 3 + 4 7 = 43.
Var ( 5 X 2 + 4 X 3 ) = 25 Var ( X 2 ) + 40 Cov ( X 2 , X 3 ) + 16 Var ( X 3 )
= 25 4 + 40 2 + 16 9 = 324.
OR

4 1 0 0


Var ( 5 X 2 + 4 X 3 ) = ( 0 5 4 ) 1 4 2 5 =
0
2 9 4

0

( 5 28 46 ) 5 = 324.

4

70 43
= P ( Z > 1.50 ) = 0.0668.
P ( 5 X 2 + 4 X 3 > 70 ) = P Z >
324

c)

Find P ( 3 X 3 5 X 2 < 17 ).
E ( 3 X 3 5 X 2 ) = 3 7 5 3 = 6.
Var ( 3 X 3 5 X 2 ) = 9 Var ( X 3 ) 30 Cov ( X 2 , X 3 ) + 25 Var ( X 2 )
= 9 9 30 2 + 25 4 = 121.

OR

4 1 0 0


Var ( 3 X 3 5 X 2 ) = ( 0 5 3 ) 1 4 2 5 =
0
2 9 3

0

( 5 14 17 ) 5 = 121.
3

17 6
P ( 3 X 3 5 X 2 < 17 ) = P Z <
= P ( Z < 1.00 ) = 0.8413.
121

d)

Find P ( 4 X 1 3 X 2 + 5 X 3 < 80 ).
E ( 4 X 1 3 X 2 + 5 X 3 ) = 4 1 3 2 + 5 3 = 4 5 3 3 + 5 7 = 46.

4 1 0 4
4



Var ( 4 X 1 3 X 2 + 5 X 3 ) = ( 4 3 5 ) 1 4 2 3 = ( 19 6 39 ) 3
0
5
2 9 5


= 289.
4 X1 3 X2 + 5 X3 ~ N ( 46, 289 )

80 46
= P ( Z < 2.00 ) = 0.9772.
P ( 4 X1 3 X2 + 5 X3 < 80 ) = P Z <
289

e)*

Find P ( X1 > 8 | X2 = 1, X3 = 10 ).
4 2

22 =
2 9
1
12 22
=

1
22
=

1 9 2

32 2 4

9 2 1
1
= ( 9 2 )
( 1 0)
32
2 4 32

1 + 12 221 (X 2 2 ) = 5 +
11 12 221 21 = 4

1 3
1
= 5.75.
( 9 2)
32
10 7

1
1
( 9 2) = 3.71875.
32
0

X1 | X2 = 1, X3 = 10 ~ N ( 5.75, 3.71875 )

8 5.75
= P ( Z > 1.17 ) = 0.1210.
P ( X1 > 8 | X2 = 1, X3 = 10 ) = P Z >

3
.
71875

You might also like