Probability Theory

1
3. Random variables (r.v.)
Let (O, K, P) a probability field (or probability o-field).

Definition
A function X: OR with the property that: xeR, {eeOX(e)<x}e K is called a
random variable on this field.
Notations
- X(e) is the value of r.v. for event e;
- {eeOX(e)<x}={X<x};
Properties
Consider (O, K, P) a probability field (or probability o-field) and X, Y: OR two
random variables. Then
1) {eeOX(e)>Y(e)}e K;
2) {eeOX(e)sY(e)}e K;
3) {eeOX(e)=Y(e)}e K.

Definition
Consider (O, K, P) a probability field (or probability o-field) and X: OR a random
variable. The function F:RR given by:
F(x)=P({eeOX(e)<x})=P(X<x)
is named cumulative distribution function (c.d.f.) of X.
Properties
The cumulative distribution function F:RR, of a random variable X, has the following
properties:
1) 1 ) ( F lim =

x
x
and lim F( ) 0
x
x

= ;
2) F(x)e[0, 1] xeR;
3) If a<b P(asX<b)=F(b)-F(a)
4) F monotonically nondecreasing function: asbF(a) sF(b);
5) F is left continuous at any x
0
eR: ) ( F ) ( F lim
0
0
0
x x
x x
x x
=
<

Probability Theory
2
3.2 Discrete random variables. Operations with discrete random variables.
Consider (O, K, P) a probability field (or probability o-field) and X: OR a random
variable.

Definition
If the set X(O)={X(e)eeO} is at most countable (that is, a finite or countable set),
then X is a discrete random variable.
Remark
If X is a discrete random variable then X(O) is either finite - X(O)={x
1
, x
2
, , x
n
}- or
countable - X(O)={x
1
, x
2
, , x
n
,}.

Definition
The distribution of a discrete random variable is defined by the formula P(X=x
i
)=p
i

x
i
eX(O) and can be represented by the following distribution table:
X:
|
|
.
|

\
|
... ...
... ...
2 1
2 1
n
n
p p p
x x x

(or equivalently, X:
I e
|
|
.
|

\
|
i
i
i
p
x
with IcN).
Properties
Consider the distribution of a discrete random variable X, given by P(X=x
i
)=p
i
,
ieIcN. Then:
1) p
i
>0, ieI.
2)

eI
i
p
i
=1.
Proof
1) p
i
is a probability, so p
i
>0, ieI.
2) O= } ) X( {
I
i
i
x = e e
e

I
A
e i
i
and A
i
A
j
=u i=j, i, jeI. Then
1=P(O)=P(

I
A
e i
i
)=

eI
) P(A
i
i
=

eI
i
p
i
.
Example
The distribution table of the random variable X representing the face obtained when a
die is rolled is X:
|
|
|
.
|

\
|
6
1

6
1

6
1

6
1

6
1

6
1
6 5 4 3 2 1
.

Probability Theory
3
Remark
The c.d.f. of a discrete random variable having the distribution table X:
I e
|
|
.
|

\
|
i
i
i
p
x
with IcN can
be computed by the formula: F(x)=

< e } {

x x j i
i
j
p .
Definition
Let X, Y: OR two random variables defined on the same probability field (O, K, P). We
say that X, Y are independent iff
P((X<x)(Y<y)) =P(X<x) P(Y<y) (or P(X<x, Y<y)= P(X<x) P(Y<y))
Consider (O, K, P) a probability field (or probability o-field) and X, Y: OR two random
variables. The function Z: OR
2
defined by Z(e)=(X(e), Y(e)) eeO is called
two-dimensional (or bivariate) random variable (or random vector) and the random
variables X, Y are called marginal random variables (notation Z=(X, Y)).
If the random variables X and Y are discrete, then Z=(X, Y) is a discrete two-dimensional
random variable and the set Z(O) is at least countable.
The c.d.f. of Z is defined by:
F(x, y)=P(X<x, Y<y)=P({eeOX(e)<x, Y(e)<y}) (x, y) e R
2
.
Remark
Using the c.d.f. of random vector Z=(X, Y) we can write
X, Y independent iff F
Z
(x, y)=F
X
(x) F
Y
(y).

The distribution function of a discrete two-dimensional random variable Z=(X, Y).
Let X:
I e
|
|
.
|

\
|
i
i
i
p
x
, IcN and Y:
J i
j
j
q
y
e
|
|
.
|

\
|
, JcN the marginal random variables of
Z=(X, Y). We denote A
i
={eeOX(e)=x
i
}, B
j
={eeOY(e)=y
j
},
C
ij
=A
i
B
j
={eeOX(e)=x
i
, Y(e)=y
j
}, p
ij
=P(C
ij
) for every ieI, jeJ.
Remark
- The set (C
ij
)
(i,j)eIJ
is a partition of O.
- It can be proved that the probabilities p
ij
have the following properties:
- p
ij
>0, (i, j)eIJ;
-

eJ j
ij
p =p
i
, ieI and

eI i
ij
p =p
j
, jeJ;

e J I j i
ij
p
) , (
=1.

Probability Theory
4
As a consequence, the random variable Z=(X, Y) can be given by the matrix:
Z:
J I ) , (

) , (
e
|
|
.
|

\
|
j i
ij
j i
p
y x
.
In the case of finite marginal random variables (I=n, J=m), the random variable Z=(X, Y)
is suitable to be represented by the following table:
Y
X
y
1
y
2
y
j
y
m
P(X=x
i
)
x
1
x
2
p
11
...
.....
p
1j
p
1m

.

p
1
p
2

x
i


...

p
ij



p
i

x
n
p
n1
........ p
nj
.. p
nm
p
n
P(Y=y
j
) q
1
q
2
........

q
j
q
m



Remark
The discrete two-dimensional random variables X, Y are independent iff
p
ij
=p
i
q
j
, (i, j)eIJ.


Proposition 1 (operations with random variables)
Consider X : OR a random variables defined on the probability field (or probability o
field) (O, K, P). Then:
1) aX, aeR,
2) X,
3) X
k
, keN*,
4)
X
1
if 0eX(O),
are also random variables.



Probability Theory
5
Proof
1) If a=0 then xeR we have
{eeO(aX)(e)<x}=

< e > e e e
> e < e e e
0 a if , }
a
x
) ( X {
0 a if , }
a
x
) ( X {
K
K
.
If a=0 then {eeO(aX)(e)<x}=

< e
> e u
0 x if ,
0 x if ,
K
K
. Therefore, (aX) is a random variable.
2) For xeR we can write
{eeO X(e)<x}=

> e < e < e e

s e u
0 x if , } x ) ( X x {
0 x if ,
K
K
.
So X is a random variable.
3) If k=2p, peN* then, xeR,
{eeOX
k
(e)<x}=

> e < e < e e

s e u
0 x if , } x ) ( X x {
0 x if ,
k k
K
K
.
If k=2p+1, peN* then, xeR, {eeOX
k
(e)<x}={eeOX(e)<
k
x }eK.
So X
k
is a random variable.

4) For xeR, we have
{eeO
) ( X
1
e
<x}=

> e < e < e

< e < < e
= e < e
0 if , } 0 ) ( { )} (
1
{
0 x if , } 0 ) (
1
{
0 x if , } 0 ) ( {
x X X
x
X
x
X
K
K
K
e e e e
e e
e e

Proposition 2 (operations with random variables)
Consider X, Y:OR random variables defined on the probability field (or probability o
field) (O, K, P). Then:
5) X-Y, X+Y, XY,
6)
Y
X
if 0eX(O),
7) max{X,Y}, min{X, Y},
are also random variables.
Probability Theory
6
3.3 Continuous random variables.
Definition
Consider (O, K, P) a probability o field. A random variable X: OR is said to be continuous
iff - f: R[0, ) integrable on R such that:
F(x)=
}

x
dt ) t ( f ,
where F is the cumulative distribution function of X.
In this case f is called probability density function (p.d.f.).
Properties (of the p.d.f.)
Let f: R[0, ) a probability density function of a random variable X. Then
1) f(x)>0, xeR.
2)
}

dx ) x ( f =1.
3) F(x)=f(x) provided that F(x) exists.
4) P(asX<b)=P(a<Xsb)=P(asXsb)=P(a<X<b)=
}
b
a
dx ) x ( f with a<b.
Exercise 1.
Consider a continuous random variable X with the p.d.f. of given by: f(x)=

e
otherwise , 0
b] [a, x if , k
.
Compute k and the c.d.f. F(x).

Exercise 2.
Consider a continuous random variable X with the p.d.f. of given by: f(x)=

e
otherwise , 0
3] [0, x if ,
2
kx
.
Compute k, c.d.f. F and P(1<X<2).

3.4. Numerical values associated to random variables.
Definition (expectation, mean value, expected value)
Consider X: OR a random variable on a probability field (or probability o field).
The expectation of r. v. X, denoted E(X), is defined by:
Probability Theory
7
- E(X)=

eI i
i i
p x if X:
I i
i
i
p
x
e
|
|
.
|

\
|
a discrete random variable;
- E(X)=
}

dx ) x ( xf if X: R e
|
|
.
|

\
|
x ,
) x ( f
x
is a continuous random variable.

Properties (expectation)
Consider X, Y: OR discrete random variables defined on the same probability field.
1) E(a)=a, aeR.
2) E(aX)=aE(X), aeR.
3) E(X+Y)=E(X)+E(Y).
4) E(XY)=E(X)E(Y) if X, Y are independent.
Proof
1) We have a:
|
|
.
|

\
|
1
a
, aeR, so E(a)=a1=a.
2) If X:
I i
i
i
p
x
e
|
|
.
|

\
|
a discrete random variable, then aX:
I i
i
i
p
ax
e
|
|
.
|

\
|
so
E(aX)=

eI i
i i
p ax =a

eI i
i i
p x =aE(X).
If X: R e
|
|
.
|

\
|
x ,
) x ( f
x
is a continuous random variable, then aX: R e
|
|
.
|

\
|
x ,
) x ( f
ax
so
E(aX)=
}

dx ) x ( axf =a
}

dx ) x ( xf =aE(X).
We prove the properties 3 and 4 only for discrete case.
3) Let X, Y: OR discrete random variables, X:
I i
i
i
p
x
e
|
|
.
|

\
|
, Y:
J j
j
j
q
y
e
|
|
.
|

\
|
, I, JcN and
X+Y:
J I ) j , i (
ij
j i
p
y x
e
|
|
.
|

\
| +
. Then we can write
E(X+Y)=

e
+
J I ) j , i (
ij j i
p ) y x ( =

e
+
J I ) j , i (
ij j ij i
p y p x =

e e I i J j
ij i
) p x ( +

e e J j I i
ij j
) p y ( =
=

eI i
i i
p x +

eI j
j j
q y =E(X)+E(Y).
Probability Theory
8
4) Let X, Y: OR independent discrete random variables, X:
I i
i
i
p
x
e
|
|
.
|

\
|
, Y:
J j
j
j
q
y
e
|
|
.
|

\
|
,
I, JcN and XY:
J I ) j , i (
j i
j i
q p
y x
e
|
|
.
|

\
|
. Therefore E(XY)=

e J I ) j , i (
j i j i
q p y x
=

e J I ) j , i (
j j i i
) q y )( p x ( =

eI i
i i
) p x (

eJ j
j j
) q y ( =E(X)E(Y).
Definition
The r
th
(initial) moment of a random variable X is defined by:
M
r
=E(X
r
)=

e
|
|
.
|

\
|
|
|
.
|

\
|
}

e
e
r.v. continuous , x ,
) (
: X if ) (
variable random discrete
p
x
: X if
i
i
R
x f
x
dx x f x
p x
r
I i
I i
i
r
i
, with reN*.
It is obvious that m
1
=E(X).

Another important typical value of a random variable is the variance, which measures the
dispersion of the values of X around its mean.

Definition (variance)
The variance of a random variable X: OR is defined by:
Var(X)=E[(X-E(X))
2
].

Remark
If E(X
2
)< then the variance of r. v. X can be computed as follows:
Var(X)=E(X
2
)-E
2
(X)
because Var(X)=E[(X-E(X))
2
]=E[X
2
-2XE(X)+E
2
(X)]=E(X
2
)-2E(X)E(X)+E(E
2
(X))=
=E(X
2
)-2E
2
(X)+E
2
(X)=E(X
2
)-E
2
(X).

Properties (variance)
Consider X, Y: OR, random variables defined on the same probability (o) field
1) Var(a)=0, aeR ;
2) Var(aX)=a
2
Var(X), aeR.
3) Var(XY)=Var(X)+Var(Y) if X, Y are independent random variables.
Probability Theory
9
Proof
1) Var(a)=E(a
2
)-E
2
(a)=a
2
-a
2
=0.
2) Var(aX)=E(a
2
X
2
)-E
2
(aX)=a
2
E(X
2
)-(aE(X))
2
=a
2
[E(X
2
)-E
2
(X)]=a
2
Var(X).
3) As X, Y are independent r. v. we have E(XY)=E(X)E(Y) (*).
Var(XY)=E[(XY)
2
]-E
2
(XY)= E(X
2
+Y
2
2XY)-(E(X) E(Y))
2
=
=E(X
2
)+E(Y
2
) 2E(XY)-(E
2
(X)+ E
2
(Y) 2E(XY))=
=E(X
2
)-E
2
(X)+E(Y
2
)-E
2
(Y) 2[E(XY)-E(X)E(Y)]= Var(X)+Var(Y) from (*).
Definition
Consider X: OR a random variable. The central moment of r
th
order of X is defined by:

r
=E[(X-E(X))
r
] with reN*.
The square root of Var(X) is called the standard deviation of X and we write: o
X
= ) X ( Var .
Remark
1)
1
=0;
2)
2
=Var(X).

Definition (central moment)
Consider X, Y:OR univariate random variables defined on the probability field (or
probability o - field) (O, K, P) and Z=(X, Y) bivariate (two-dimensional) random variable.
The (r, s) order central moment of Z is defined by:
r,s
=E[(X-E(X))
r
(Y-E(Y))
s
]
If Z=(X,Y) is discrete, X:
I i
i
i
p
x
e
|
|
.
|

\
|
, Y:
J j
j
j
q
y
e
|
|
.
|

\
|
, I, JcN then

r,s
=

e

J I ) j , i (
ij
s
j
r
i
p )) Y ( M y ( )) X ( M x ( where p
ij
=P(X=x
i
, Y=y
j
), (i, j)eIJ.
Particular cases
Let Z=(X, Y) bidimensional random variable then:
1)
2,0
=E[(X-E(X))
2
]=Var(X)
2)
0,2
=E[(Y-E(Y))
2
]=Var(Y)
3)
1,1
=E[(X-E(X))(Y-E(Y))]
not
= cov(X, Y) covariance or random variables X and Y.




Probability Theory
10
Properties (covariance)
1) cov(X, Y)=E(XY)-E(X)E(Y);
2) X, Y independent cov(X, Y)=0;
3) cov(X, Y)=cov(Y, X)
4) cov(X, X)=Var(X).
5) cov(X, a)=0 aeR
6) cov(aX, bY)= ab cov(X, Y)
7) cov(X+X, Y)=cov(X, Y)+cov(X, Y).
Consequence
Var(XY)=Var(X)+Var(Y)2cov(X, Y)

Definition (correlation coefficient)
Consider X, Y:OR random variables defined on the probability field (or probability o field)
(O, K, P). The correlation coefficient of random variables X and Y is defined by
(X, Y)=
Y X
Y) , X cov(
o o
=
) Y ( Var ) X ( Var
) Y ( M ) X ( M ) XY ( M
if o
X
, o
Y
=0.
Properties (correlation coefficient)
1) X, Y independent random variables (X, Y)=0;
2) (X, Y)e[-1, 1], X, Y;
3) If Y=aX+b (X, Y)=1;
4) If o
X
=0 then (X, X)=1 and (X, -X)=-1.

Exercise
Consider the discrete bidimensional random variable Z=(X, Y) given by the table:

Probability Theory
11
a)
Y
X
0 y p
i

x a
12
1
b
1
2
1
c
3
2

q
j

4
3
d

b)
Y
X
0 y p
i

x a
6
1
b
1
12
7
c
3
2

q
j

4
3
d

Find x, y a, b, c, d such as E(X)=
3
1
, E(Y)=
4
1
and for the obtained values, compute X, Y, the
cdf F
X
and F
(X, Y)
(0, 2), XY, cov(X, Y), (X, Y), o
X
, o
Y
.










Probability Theory
12
3.5 Several Classical Distributions

3.5.1 Binomial distribution

We say that a random variable X: OR follows a binomial distribution of parameters n and
p, and we write XeBi(n, p) if P(X=k)=
k
n
C p
k
(1-p)
n-k
, k= n , 0 or X:
|
|
.
|

\
|
k n k k
n
q p C
k
, k= n , 0 ,
q=1-p, pe[0, 1].

Remarks
1) p
k
=
k
n
C p
k
(1-p)
n-k
>0 k= n , 0 .
2)

=

n
0 k
k n k k
n
q p C =(q+p)
n
=1
Expectation: E(X)=np
Variance: Var(X)=npq

3.5.2 Poisson distribution
A random variable X: OR is said to be Poisson distributed with parameter >0
- we write XeP()) iff P(X=k)=e
-
! k
k

, keN , or X:
N e

|
|
|
.
|

\
|

k
k
! k
e
k
.
Remarks
1) p
k
=P(X=k)= e
-
! k
k

>0, keN.
2)

e

e

=
N N k
k
k
k
! k
e p = e
-
e

=1.
Expectation: E(X)=
XeP() E(X)=

>


0 k
k
! k
ke =e
-

>

1 k
k
! k
k = e
-


>

1 k
1 k
)! 1 k (
= e
-

>

0 k
k
! k
= e
-
e

=.
Variance: Var(X)=
Probability Theory
13
E(X
2
)=

>


0 k
k
2
! k
e k = e
-

>

1 k
k
)! 1 k (
k = e
-

>

1 k
1 k
)! 1 k (
k =
= e
-
(

>

1 k
1 k
)! 1 k (
) 1 k ( +

>

1 k
1 k
)! 1 k (
)= e
-
(

>

2 k
1 k
)! 2 k (
+ e

)= e
-
( e

+ e

)=(+1).
Hence Var(X)=E(X
2
)-E
2
(X)= (+1)-
2
=.

3.5.3 Hypergeometric distribution
A random variable X: OR has a hypergeometric distribution of parameters n, a and b (it is
written XeH(n, a, b)) iff P(X=k)=
k n-k
a b
n
a+b
C C
C

, ke
0
0, n where a, b, neN,
0
n min{ , n} a = and
ns a+b (or X:
0
k n-k
a b
n
a+b
k 0, n
k
C C
C
e
| |
|

|
|
\ .
).
Remarks
1) p
k
=P(X=k)=
k n-k
a b
n
a+b
C C
C

>0, ke
0
0, n .
2)
0 0 0
n n n
k n-k
a b k n-k n
k a b a+b
n n n
a+b a+b a+b
k=0 k=0 k=0
C C 1 1
p C C C 1
C C C

= = = =

.

Expectation: E(X)= n
a
a b +

Variance: Var(X)= n
1
a b a b n
a b a b a b
+
+ + +
.

3.5.4 Normal distribution

We say that a random variable X: OR follows a normal distribution of parameters m and o
- we write XeN(m, o) - if its p.d.f. has the form:
f(x)=
2
2
2
) m x (
e
2
1
o

t o
, meR, o>0.
Remarks
1) f(x)>0 xeR.
Probability Theory
14
2)
}

dx ) x ( f =
t o 2
1
}

dx e
2
2
2
) m x (
=
t o 2
1
}

o dt 2 e
2
t
=
t
1
}

dt e
2
t
=2
t
1
}

0
t
dt e
2
=
=2
t
1
2
t
=1 where (x-m)=to 2 .

Expectation: E(X)=m
E(X)=
}

dx ) x ( xf =
t o 2
1
}

dx xe
2
2
2
) m x (
t
2
m x
=
o

=
t o 2
1
}

o + o dt 2 e ) m t 2 (
2
t
=
=
t o 2
1
(o 2 )
2
}

dt te
2
t
+m
t o 2
1
}

dt e
2
t
=m

Variance: Var(X)=o
2

Var(X)=
}

dx ) x ( f ) m x (
2
=
}

t o
dx e
2
1
) m x (
2
2
2
) m x (
2
=
=
t o 2
1
}

dx e ) m x (
2
2
2
) m x (
2
t
2
m x
=
o

=
t o 2
1
}

o o dt 2 e t 2
2
t 2 2
=
t
o
2
}

dt )' e ( t
2
t
= =
t
o
2
(t
2
2
t
e

-
}

dt e
2
t
)=
t
o
2
2
}

0
t
dt e
2
=
t
o
2
2
2
t
=o
2
.
Standard deviation (o
X
=o)
The c.d.f. values of a normal distributed random variable are computed using Laplace
function defined by: | :RR, |(z)=
t 2
1
}

z
0
t
2
1
dt e
2
.
Properties
1) |(0)=0;
2) |(-z)=- |(z) (odd function);
3) ) z ( lim
z
|

=0.5;
4) ) z ( lim
z
|

=-0.5;

Probability Theory
15
Cumulative distribution function (F(z)=
2
1
+|(
o
m z
))
F(z)=
}

o

t o
z
2
) m x (
dx e
2
1 2
2
t
m x
=
o

=
t o 2
1
}
o

o
m z
t
2
1
dt e
2
=
=
t 2
1
}

0
t
2
1
dt e
2
+
t 2
1
}
o

m z
0
t
2
1
dt e
2
u
2
t
=
=
t 2
1
}

0
u
du ) 2 ( e
2
+|(
o
m z
)=
=
t
1
}

0
u
du e
2
+|(
o
m z
)=
2
1
+|(
o
m z
).
Remark
It is easy to prove that F(-z)=1-F(z), zeR.

The most important normal distribution is the standard normal N(0,1), with m=0 and o
2
=1:
- we write XeN(0, 1) - if its p.d.f. has the form: f(x)=
2
2
2
1
x
e

t
, xeR
3.5.5 Gamma distribution

We say that a random variable X: OR follows a gamma distribution of parameters a and b
- we write Xe(a, b) - if its p.d.f. has the form:
f(x)=
1
1
, x>0
( )
0 , x 0
x
a
b
a
x e
a b

, a, b>0.

Remarks
1) f(x)>0 xeR.
2)
}

dx ) x ( f =
a
b ) a (
1
I
}

0
b
x
1 a
dx e x
t
b
x
=
=
a
b ) a (
1
I
}

0
t 1 a
bdt e ) bt ( =
a
b ) a (
1
I
}

0
t 1 a a
dt e t b =
=
a
b ) a (
1
I
b
a
I(a)=1
Expectation: E(X)=ab
Probability Theory
16
E(X)=
}

dx ) x ( xf =
a
b ) a (
1
I
}

0
b
x
a
dx e x
t
b
x
=
=
a
b ) a (
1
I
}

0
t a
bdt e ) bt ( =
) a (
b
I
}

0
t a
dt e t =
=
) a (
b
I
I(a+1)=
) a (
b
I
aI(a)=ab.

Variance: Var(X)=ab
2

E(X
2
)=
a
b ) a (
1
I
}

+
0
b
x
1 a
dx e x
t
b
x
=
=
a
b ) a (
1
I
}

+
0
t 1 a
bdt e ) bt ( =
) a (
b
2
I
}

+
0
t 1 a
dt e t =
=
) a (
b
2
I
I(a+2)=
) a (
b
2
I
a(a+1) I(a)=ab
2
(a+1)
Var(X)= ab
2
(a+1)-a
2
b
2
=ab
2
.

3.5.6 Exponential distribution

We say that a random variable X: OR follows a negative exponential distribution of
parameter >0 - we write XeExp() - if its p.d.f. has the form: f(x)=

s
>

0 x , 0
0 x , e
x
.

Remark
XeExp()XeI(1,

1
) so E(X)=

1
and Var(X)=
2
1

.
3.5.7 Beta distribution
We say that a random variable X: OR follows a Beta distribution of parameters a and b
and we write Xe(a, b), if its pdf is:
f(x)=

, a, b>0.
Show that the expectation and variance of X are E(X)=
b a
a
+
and Var(X)= .
) 1 ( ) (
2
+ + + b a b a
ab




Probability Theory
17

3.5.8 Chi-square distribution

We say that a random variable X: OR follows a chi-square distribution of parameters n
and o - we write Xe_
2
(n, o) iff Xe(
2
n
, 2o
2
) . Then the p.d.f. of X is:
f(x)=
2
1
2 2
2
2
1
, x>0
( )(2 )
2
0 , x 0
x n
n
x e
n

o

I o

, neN*, o>0 and E(X)=no

2
, Var(X)=2no
4
.
3.5.9 Student distribution

A random variable X: OR follows a Student distribution of parameter n - we write
XeS(n) if its p.d.f. has the formula
f(x)=
)
2
n
( n
)
2
1 n
(
I t
+
I
2
1 n
2
)
n
x
1 (
+

+ , xeR, neN\{0,1,2}.
It can be proved that E(X)=0 and Var(X)=
2 n
n