Scribd
Upload
17 views

Vectors of Random Variables: Guy Lebanon January 6, 2006

This document discusses vectors of random variables, which are collections of multiple random variables. It defines a random vector as a function that maps from a sample space to a set of real numbers, representing the values of the random variables. Random vectors can be discrete, continuous, or neither. The document also discusses how to calculate probabilities of events involving random vectors using their joint cumulative distribution functions, probability mass functions, or probability density functions. It notes that independent random variables have distributions that factor into the marginal distributions.

Uploaded by

Ngọc Chiến
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
17 views

Vectors of Random Variables: Guy Lebanon January 6, 2006

This document discusses vectors of random variables, which are collections of multiple random variables. It defines a random vector as a function that maps from a sample space to a set of real numbers, representing the values of the random variables. Random vectors can be discrete, continuous, or neither. The document also discusses how to calculate probabilities of events involving random vectors using their joint cumulative distribution functions, probability mass functions, or probability density functions. It notes that independent random variables have distributions that factor into the marginal distributions.

Uploaded by

Ngọc Chiến
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Vectors of Random Variables

Guy Lebanon
January 6, 2006
So far we treated random variables X : R and probabilities associated with them P (X A). Much
of the remaining material in the course will involve multiple RV and their probabilities. We will consider for
the next several weeks what happens with a finite number of RV X1 , . . . , Xn , where each Xi is a function
Xi : R. Later on in the course, we will discuss random processes which are an infinite number of
random variables.
Definition 1. Let be a sample space, P a probability distribution P : R on it, and (X1 , . . . , Xn )
~ = (X1 , . . . , Xn ) is a vector of random
be n random variables Xi : R. We then also say that X
~ as a function X
~ : Rn as follows
variables or a random vector. Alternatively, we can think of X
n
n
~
~ A for the event { :
X()
= (X1 (), . . . , Xn ()) R . Assuming that A R we write X
~
~
~
(X1 (), . . . , Xn ()) A} and P (X A) for its probability P (X A) = P ({ : X()
A).
Random vectors are discrete, continuous or neither. The definition is very similar to the one dimensional
~ is discrete if there exists a finite or countable set K s.t. P (X
~ is continuous if
~ K) = 1 and X
case: X
~ = (x1 , . . . , xn )) = 0 for all (x1 , . . . , xn ). For example, the voltage measurement at several locations of
P (X
a circuit is a continuous random vector. The measurement of color and sex of a particular species of animals
is a discrete random vector (note that in this case we have to map the color and sex to real numbers e.g.
brown=0, white=1, black=2 and male=0, female=1). The measurements of height, weight and gender of a
person is neither discrete nor continuous.
~ A will take a simpler form:
Often (but not always), the events X
{X1 A1 , . . . , Xn An } = { : X1 () A1 , . . . , Xn () An }
= {X1 A1 } {Xn An }
and P (X1 A1 , . . . , Xn An ) denotes the probability of that event. Such an event is called a factored event
(you may visualize it as an n dimensional rectangle of sides A1 , . . . , An ).
Definition 2. The random variables X1 , . . . , Xn are independent if for all A1 , . . . , An
P (X1 A1 , . . . , Xn An ) = P (X1 A1 ) P (Xn An )
For example, let describes the result of throwing a pair of dice, X1 is a RV measuring the sum of the
two dice and X2 is a RV measuring the difference of the two dice. We have P (X1 = 2, X2 = 2) = P (X1
{2}, X2 {2}) = , P (X1 = 6, X2 = 2) = P ({(4, 2)}) = 1/36 and
P (X1 > 4, X2 < 0) = P ({(1, 4), (1, 5), (1, 6), (2, 3), (2, 4), (2, 5), (2, 6), (3, 4), (3, 5), (3, 6), (4, 5), (4, 6), (5, 6)} = 13/36.
We also have P (X1 > 4) = 1 P ({(1, 1), (1, 2), (2, 1), (1, 3), (3, 1), (2, 2)}) = 30/36 and P (X2 < 0) = 15/36.
Since P (X1 > 4, X2 < 0) 6= P (X1 > 4)P (X2 < 0), X1 , X2 are not independent.
~
As with one dimensional RV we define cdf, pmf and pdf to help us calculate probabilities. The cdf of X
is defined as
FX
~ (x1 , . . . , xn ) = P (X1 x1 , X2 x2 , . . . , Xn xn )
the pdf is defined as its multiple derivative
fX(x
~ 1 ,...,xn ) (x1 , . . . , xn ) =

n
F ~ (x1 , . . . , xn )
x1 xn X
1

if it exists and 0 otherwise and the pmf is defined as


pX
~ (x1 , . . . , xn ) = P (X1 = x1 , . . . , Xn = xn ).
~ =x
~ = (x1 , . . . , xn ) e.g. pX
~ ).
We will sometime abbreviate the above formulas by denoting x
x) = P (X
~ (~
~
As in the one dimensional case, the pdf (for continuous X) has to be non-negative and integrate to 1
R
R
~ has to be non-negative and sum to one P P p ~ (~
fX
x)d~
x = 1, and the pmf (for discrete X)
~ (~
x1
x2 X x) =
1.
From the fundamental theorem of calculus, we have that for continuous RV
Z xn
Z x1 Z x2

fX
FX
~ (t1 , t2 , . . . , tn ) dt1 dt2 dtn .
~ (x1 , x2 , . . . , xn ) =

~ A), A Rn are similar to the one-dimensional case, and their proofs


The formulas for computing P (X
carry over without much problems
(P
~ is a discrete random vector
x)
X
~ (~
~ A) = R ~xA pX
P (X
.
~
fX
x) d~
x X is a continuous random vector
~ (~
~
XA
The marginal pmf, pdf and cdf are obtained as follows:
X
XX
X
X
pXi (xi ) =

pX
pX1 ,X2 (x1 , x2 )
~ (x1 , . . . , xn ) e.g. pX1 (x1 ) =
x1

xi1 xi+1

Z
fXi (xi ) =

xn

x2 R

fX
~ (x1 , . . . , xn ) dx1 dxi1 dxi+1 dxn

FXi (xi ) = FX
~ (, , , xi , , , )

e.g. fX1 (x1 ) =

fX1 ,X2 (x1 , x2 ) dx2

e.g. FX1 (x1 ) = FX1 ,X2 (x1 , )

~ A) or
The following result is useful when the variables are independent and we want to compute P (X
if we want to check whether some random variables are indeed independent.
~ are independent if and only if the cdf factors F ~ (x1 , . . . , xn ) = FX (x1 ) FX (xn ).
Theorem 1. The vector X
1
n
X
~
X is also independent if and only if the pmf factors (for discrete random vectors) pX
~ (x1 , . . . , xn ) = pX1 (x1 ) pXn (xn )
or the pdf factors (for continuous random vectors) fX
~ (x1 , . . . , xn ) = fX1 (x1 ) fXn (xn ).
2

1 (x +y )/2
e
. Since the
Example: Suppose that the vector (X, Y ) has the joint-normal pdf fX,Y (x, y) = 2
pdf factors we know that fX,Y (x, y) = fX (x)fY (x) where X, Y are independent normal distribution with
1 x2 /2
= 0, = 1 fX (x) = 2
e
. Because of this factorization of the pdf we can compute probabilities of
factored sets (rectangles) more easily:

P (3 < X < 6, Y < 9) =

fX,Y (x, y) dxdy =

Z
fX (x)fY (y) dxdy =

fX (x)dx
3

fY (y)dy.

Example: Let (X, Y ) be a random vector where X, Y are independent exponential random variables with
parameters 1 , 2 . Then
fX,Y (x, y) = fX (x)fY (y) = 1 2 e1 x e2 y = 1 2 e1 x2 y
and the joint cdf is
Z x Z
FX,Y (x, y) =

1 x2 y

1 2 e

dxdy =

1 x

1 e

2 e2 y = (1e1 x )(1e1 x ) = FX (x)FY (y).

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy