Digital Signal Processing Reference
In-Depth Information
Fig. 2.53
Illustration of Chebyshev theorem (first form)
We introduce the zero mean variable shown in Fig.
2.53b
(see Sect.
2.8.2.3
)
Y ¼ X m
X
;
(2.433)
resulting in:
ð
1
e
s
2
y
2
f
Y
ðyÞ
d
yþ
y
2
f
Y
ðyÞ
d
y:
X
(2.434)
1
e
Replacing
y
by value
ðeÞ
in (
2.434
) makes this inequality even stronger:
2
3
ð
e
1
e
2
f
Y
ðyÞ
d
y; ¼ e
2
ð
e
1
4
5
;
s
X
e
2
f
Y
ðyÞ
d
y þ
f
Y
ðyÞ
d
y þ
f
Y
ðyÞ
d
y
1
e
1
e
¼ e
2
PfjYjeg¼e
2
PfjX m
X
jeg:
(2.435)
Finally, from (
2.435
), we have:
PfjX m
X
jegs
2
X
=e
2
:
(2.436)
2.8.5.2 Second Form of Inequality
Replacing
e ¼ ks
X
in (
2.430
), we can obtain the alternative form of Chebyshev
inequality
=k
2
PfjX m
X
jks
X
g
1
:
(2.437)
The inequality (
2.437
) imposes a limit of 1/
k
2
, where
k
is a constant, to the
probability at the left side of (
2.437
).
Search WWH ::
Custom Search