*5.7The Gamma Distribution 183
we see that
(n)=(n−1)!
The function (α) is called thegammafunction.
It should be noted that whenα=1, the gamma distribution reduces to the exponential
with mean 1/λ.
The moment generating function of a gamma random variableXwith parameters (α,λ)
is obtained as follows:
φ(t)=E[etX]
=
λα
(α)
∫∞
0
etxe−λxxα−^1 dx
=
λα
(α)
∫∞
0
e−(λ−t)xxα−^1 dx
=
(
λ
λ−t
)α
1
(α)
∫∞
0
e−yyα−^1 dy [byy=(λ−t)x]
=
(
λ
λ−t
)α
(5.7.2)
Differentiation of Equation 5.7.2 yields
φ′(t)=
αλα
(λ−t)α+^1
φ′′(t)=
α(α+1)λα
(λ−t)α+^2
Hence,
E[X]=φ′(0)=
α
λ
(5.7.3)
Var(X)=E[X^2 ]−(E[X])^2
=φ′′(0)−
(α
λ
) 2
=
α(α+1)
λ^2
−
α^2
λ^2
=
α
λ^2
(5.7.4)
An important property of the gamma is that ifX 1 andX 2 are independent gamma
random variables having respective parameters (α 1 ,λ) and (α 2 ,λ), thenX 1 +X 2 is a
gamma random variable with parameters (α 1 +α 2 ,λ). This result easily follows since
φX 1 +X 2 (t)=E[et(X^1 +X^2 )] (5.7.5)
=φX 1 (t)φX 2 (t)