Escolar Documentos
Profissional Documentos
Cultura Documentos
g : Rp → R;
Definition 1.
(i) A function of one or more r.v.s that does not depend on any unknown
parameter involved in joint p.m.f./ p.d.f. of r.v.s is called a statistic.
X1 +X2
In the above example X , X1 , 2 are statistics whereas, X − θ or X1 /θ
are not statistics.
For 1 ≤ x < 2,
Z1 min{1,x−x
Z 1}
FY (x) = P(Y ≤ x)
= P(min{X1 , X2 , X3 } ≤ x)
= 1 − P(min{X1 , X2 , X3 } > x)
= 1 − P(X1 > x, X2 > x, X3 > x)
= 1 − P(X1 > x)P(X2 > x)P(X3 > x)
= 1 − (1 − F (x))3
Thus,
0,
if x < 1
FY (x) = 1 − (1 − 6i )3 , if i ≤ x < i + 1, i = 1, . . . , 5
1, if x ≥ 6
Ay = {x ∈ Sx : g1 (x) ≤ y1 , . . . , gk (x) ≤ yk }
and
By = {x ∈ Sx : g1 (x) = y1 , . . . , gk (x) = yk }.
Then the r.v. Y = (Y1 , . . . , Yk ) is of discrete type with support SY , d.f.
X
FY (y ) = fX (x), y ∈ Rk
x∈Ay
(i) Clearly
SY1 = support ofY1 = {1, 2, . . .} = N
For y ∈ SY1 ,
fY1 (y ) = P(Y1 = y )
= P(min{X1 , X2 } = y )
Thus, p.m.f. of Y1 is
Thus,
θ
2−θ ,
if y = 0
2θ(1−θ)y
fY2 (y ) = , if y = 1, 2, . . .
2−θ
0, otherwise.
SY = {(y1 , y2 ) : 1 ≤ y1 ≤ y1 + y2 }
= {(y1 , y2 ) : y1 ≥ 1, y2 ≥ 0}
= N × {0, 1, 2, . . .}.
For y = (y1 , y2 ) ∈ SY
fY (y ) = P(Y1 = y1 , Y2 = y2 )
= P(min{X1 , X2 } = y1 , max{X1 , X2 } = y1 + y2 )
= P(X1 = y1 , X2 = y1 + y2 , X1 < X2 )
+ P(X1 = y1 , X1 = y1 + y2 , X1 = X2 )
+ P(X2 = y1 , X1 = y1 + y2 , X1 > X2 )
= 2P(X1 = y1 , X2 = y1 + y2 , X1 < X2 )
+ P(X1 = y1 , X1 = y1 + y2 , X1 = X2 )
fY (y ) = P(X1 = y1 , X2 = y1 )
= θ2 (1 − θ)2y1 −2 .
fY (y ) = 2P(X1 = y1 , X2 = y1 + y2 )
= 2θ2 (1 − θ)2y1 +y2 −2 .
Thus,
(
2θ2 (1 − θ)2y1 +y2 −2 , if (y1 , y2 ) ∈ N × {0, 1, 2, . . .}
fY (y ) =
0, otherwise.
Thus,
(
θ(2 − θ)(1 − θ)2y1 −2 , if y ∈ {1, 2, . . .}
fY1 (y ) =
0, otherwise.
For y2 = 0,
∞
X θ
fY2 (y2 ) = θ2 (1 − θ)2y1 −2 =
2−θ
y1 =1
For y2 ∈ {1, 2, . . .}
∞
X
fY2 (y2 ) = 2θ2 (1 − θ)2y1 +y2 −2
y1 =1
2θ(1 − θ)y2
=
2−θ
Thus,
θ
2−θ ,
if y2 = 0
2θ(1−θ)y2
fY2 (y2 ) = , if y2 ∈ {1, 2, . . .}
2−θ
0, otherwise.
Result 2. Let X = (X1 , . . . , Xp ) be an A.C. r.v. with joint p.d.f. fX (·) and
support SX . Let S1 , . . . , Sk be open subsets of Rp such that Si ∩ Sj = φ, if
k
Si = SX0 , where SX0 is the interior of SX . Suppose
S
i 6= j and
i=1
hj : Rp → R, j = 1, . . . , p, are p functions such that on each Si ,
h = (h1 , . . . , hp ) : Si → Rp is one-to-one with inverse transformation
hi−1 (t) = (h1,i−1 −1
(t), . . . , hp,i (t)) (say), i = 1, . . . , k. Further suppose that
−1
hj,i (t), j = 1, . . . , p, i = 1, . . . , k have continuous partial derivatives and
the Jacobian determinants
Let X = (X1 , . . . , Xp ) be an A.C. r.v. with joint p.d.f. fX (·) and support
SX . Suppose that hj : Rp → R, j = 1, . . . , p, are p functions such that
h = (h1 , . . . , hp ) : SX0 → Rp is one-to-one with inverse transformation
h−1 (t) = (h1−1 (t), . . . , hp−1 (t)) (say). Further suppose that hi−1 ,
i = 1, . . . , p, have continuous partial derivatives and the Jacobian
determinants
−1
∂h1 (t) ∂h1−1 (t)
∂t
−11 . . . ∂tp
∂h2 (t) ∂h2−1 (t)
...
J = ∂t. 1 ∂tp
. 6= 0.
.. ..
−1
∂hp−1 (t)
∂hp (t)
∂t1 . . . ∂tp
FY (y ) = P(Y ≤ y )
= P(X1 + X2 ≤ y )
Z1 Z1
= 4x1 x2 dx
0 0
x1 +x2 ≤y
Z } min{1,y
min{1,y Z −x1 }
FY (y ) = 4 x1 x2 dx2 dx1
0 0
For 0 ≤ y < 1,
Zx1 yZ−x1
FY (y ) = 4 x1 x2 dx2 dx1
0 0
y 4
= .
6
(b) Clearly SX = [0, 1] × [0, 1] and SX0 = (0, 1) × (0, 1). Define Z = X2 .
Then the transformation (Y , Z ) = h(X1 , X2 ) = (X1 + X2 , X2 ) → R2 is
1-1 on SX0 with inverse transformations
h1−1 (y , z) = x1 = y − z
h2−1 (y , z) = x2 = z,
Thus
h(SX0 ) = {(y , z) ∈ R2 : z < y < 1 + z, 0 < z < 1}
Z }
min{1,y
fY (y ) = 4z(y − z)dz
max{0,y −1}
y
R
4z(y − z)dz, if 0 < y < 1
0
= R1
4z(y − z)dz, if 1 < y < 2
y −1
0, otherwise.
Let M(·) be the m.g.f. of some known distribution say D1 . If we can show
that a r.v. (say X ) has m.g.f. M(·) in a neighborhood of zero, then using
uniqueness of m.g.f., we can conclude that X has distribution D1 .
e −t 1 et
M(t) = + + , t ∈ R.
4 2 4
Find the distribution of Y = X 2 .
Solution. Clearly
X
M(t) = E (e tX ) = e tX f (x)
x∈SX
(
1
fY (y ) = P({X 2 = y }) = 2, if y = 0
1
2, if y = 1.
1 x2
f (x) = √ e − 2 , −∞ < x < ∞.
2π
k
P
where n = ni .
i=1