Digital Signal Processing Reference
In-Depth Information
However, the sum of (
2.227
) does not converge
1
1
k
!1;
(2.228)
k¼
0
and consequently, the mean value of the random variable
X
does not exist.
If we consider an arbitrary function of
x
i
,
g
(
x
i
), instead of
x
i
, the expression
(
2.214
) becomes:
¼
X
N
N
1
g
ð
x
1
Þþþ
N
N
g
ð
x
n
Þ
M
N
i
M
gðx
i
Þ:
m
empg
¼
(2.229)
i¼
1
For a high
M
value (
M >>
1), the expression (
2.229
) becomes the expected value
of
g
(
X
):
EfgðXÞg ¼
X
N
gðx
i
ÞPfX ¼ x
i
g:
(2.230)
i¼
1
If we consider that the random variable
g
{
X
} is a new variable
Y
, and
y
i
¼ g
(
x
i
),
then, according to (
2.220
), we have:
EfgðxÞg ¼ EfYg¼
X
i
y
i
PfY ¼ y
i
g:
(2.231)
Generally, the number of values of the randomvariables
X
and
Y
may be different.
The expression in (
2.231
) is an alternative relation used to find the expected value of
the function of the variable
X
,
g
(
X
). Note that in (
2.231
), we have to find the values
y
i
and the probabilities
P
{
Y ¼ y
i
}.
Example 2.7.5
Consider the random variable
X
with values
x
1
¼
1,
x
2
¼
1,
x
3
¼
2,
x
4
¼
2. The corresponding probabilities are:
P
{
x
i
}
¼
1/4, for
i ¼
1,
,4.
...
Find the expected value of
X
2
.
Solution
Using (
2.230
), we have:
g¼
X
4
EfgðXÞg ¼ EfX
2
gðx
i
ÞPfX ¼ x
i
g;
i¼
1
2
4
þ
1
2
2
4
þ
2
2
¼ð
1
Þ
1
=
1
=
4
þð
2
Þ
1
=
1
=
4
¼
5
:
(2.232)
=
2
In order to use (
2.231
), we define the random variable
Y ¼ X
2
, and find the
values
x
i
2
,
i ¼
1,
,4:
...
x
1
¼ x
2
¼
1
x
3
¼ x
4
¼
4
:
(2.233)
;
Search WWH ::
Custom Search