0 o m e Specia

l Probability
Distributions
(i) Binomial d 179
istribution is a
n legitimate /Jroha
n bility distribution
1 P(X tj = = since
1 nC, q"- r pr
r= O
r= O
= (q + p)'1 = I
( ) r br na!lle 'b
2 i110 111ia I dislribu lio
n' is
2, ... , 11) are the successive term given since the probaWities 11C, q"
) If we ass11111<1 tb s in tbe expansio - 'p' {r = 0, 1,
3 a: n t.1ials. co~ n of the binomia
( pmction of
the bwo1111al dz
stit~te ~ s~t an
d if ,ve cons~der
l e>.pression__(q ~L
•r - 0 1 2, .. stnlmtzon is y_w
en by J(r) = N
N sets; the frequ
- '' . , n. 111 otber iv p :n~
ords, tbe numbe (r) = :-- • 11C, i f p
Jtbe occurrences
~r tbe el'ent A ) = r oif sels in 1vhich
N • ,,c, q"•r p'; we get exact Iv r
successe
,
r = 0, 1, 2, . .. ..1 - s
and V a r ia n , n. _
?Jean c e o f t h e Bin
omial Dis
tribution
have already fo
\V~ the no
un d OU!_E @~!}d Var
1 ment generati~g
(X)_ fot the bin
using e shall fun~tion in Ex omial distributi
find am p le o~ B(n,__p)
ttere w th em directly usi 3 in Worked
E
ng the definitio x ample 4 (b).
ns of E (X) an
E(X)= L
X ,P r
d V ar(X).
r

n
=L r= O
r . nC r p r qn -r
... __ --

- ~£ . . i , .n \
- - - - p r qn -r
r=O r \ ( n
-r)\ ( l)
~
(n-1)\
= np •; , ( r r- 1
_ 1) \ { ( n - (n -l )- (r -1
1) - ( r - 1 )} ! p )
q
= np Ln (n - l) Cr _ 1 r _ \
• P
(n - l) - (r - 1)
r= l •q
= np (q + p)n - l
= np
(2 )
n
E(X 2 ) = L x,2 Pr= L r2
r Pr
0
n
= L {r(r - 1) + r) n\
r ! ( n - r) \
p ' qll - r
r=O

= n ( n - l) p 2 Ln (n - 2 )C ,- 2 P r- 2 n - r + no
q.
r= 2
-.:..-'
(by (1) and(2)1
= n ( n - l) p -,
- ( q + p ) n-2 + np
 Probability, Statistics and Random Processes
2
= n (n -1) P + np
2 2·
Var(X) = E (X ) - {E (X)}
2 2 2
= n (n - 1) p + np - n p
= np (1 - p)
=npq

Recurrence Formula for the Central Moments of the

Binomial DistribQtion
By definition, the kth order central moment µk is given by µk :::: E{
For the binomial distribution B(n,p), X'l:(.tJi1
n

µk = ~
~
(r - np )k nCr p t qn-r
r=O
Differentiating (I) with respect top, we get (I)

n'
dµk
L
dp = r=O net [-nk (r- nP/-'·p ' t-r + (r- npl {rp'-l qn-r

+ (n - r) p' qn-r-1
n (--I}j
=-nk µk~ 1 + L, nC (r- npl Pr-I q n-r-l {
rq-(n-r) p}
r==O r

=- nk µk- 1 + L, nC (r- nn)kpr-1 qn-r-1 (

r r r - nn) (·. p +q:: I)
r =0 r •

n
=-nkµ + 1 ~
k- J pq £. nC, p' qn-r (r _ np) k + 1
r=O

= - nk µk - I + _!_ µ
pq k+ I

• µk + 1 = pq [ -dµ t + nk
i.e., ]
. d p µk- l . (1)
.du sing recurrence relation (2)
v1 ed we kno ' we may comp t
u e moments of higher order, pro-
• k ==w moments
putt1ng
1 in (2)
of low er order.
'we get

µ2 == Pq [ ~dpµL + nµo ]
. == npq ( ·: ==
Putting k == 2 in (2) Jlo 1 and µ 1 ::: 0)
'we get

µ3 == pq [ d ]
· dp µ2+2nµ
1
 Some Special Probability Distributions 181

d
= pq - [np (1 - p))
dp
= npq [ 1 - 2p) = npq ( q - p)
• e1 k ~ 3 in (2), we get
J'lltttlli:, [
J.14 = pq ddp J.13 + 3n I½
1
= npq [ :p {p(l - p) (1 - 2p)} + 3npq J
= npq [l-6p + 6p2 + 3npq)
= npq [1 - 6pq + 3npq]
= npq [1 + 3 pq (n - 2)]

1
• µ2 is the variance, µ3 is a measure of skewness and µ 4 is a measure of kurtosis.
Sometimes the coefficients {3 1 and /32 are used to measure skewness and kurtosis respective!)·,

µj = µ4
whereµ, == J and µ2 µ2 •
µ2 2

Poisson distribution
Definition: If Xis a discrete RV that can assume the values 0, 1, 2, ... , such tha1
its probability mass function is given by
e-').. ').t
P(X=r)=--; r=O, 1,2, ... ; A>O ~
r! ~
thenX is said to follow a Poisson distribution with parameter Aor symbolical
Xis said to follow P(A).
(Note: Poisson distribution is a legitimate probability distribution, since
oo oo -A ).:
L P(x=r)= L _e-
r=O r=O r.'
=e-l el= 1)
Poisson Distribution as Limiting Form of Binomial
Distribution /
~oisson distribution is a limiting case of binomial distribution under the fi
ing conditions:
~!) n, the number of trials is indefinitely large, i.e., n ➔ 00 •
(u) P, the constant probability of success in each trial is very smi
P ➔ O.

(iii) np(::: A) is finite or p =!. and q =1 - ~. where A. is a posi1

n n
number.
 &L

'

I..- (.h f\,\ \

,\
~ht) 'l"' ,i,\'
~

f
= b f (x) dx +t
a
f''
o
i\f(.\') cl.\' + '"

l'J I

t' II . .
Obviously, the coefficient'of ;j' in (2) ht Uau /" 1 mouwut, uliouti Uu~ HI tnt11,

d' v,
Also - M<t) = - r ! + v,, ♦ J t f, ,, , I;; v ' ' If ,
' dt' t:.()
r! ~1
_ 1•
I
'

Thus, vr about origin= ,.th d~rivatlw~ ,,I' M(t) w1U1 t - 0,

Although the moment gtn,~ratlng fur,dA,,u (111.u.l',J Jua~ f u1t111 ,lt 1fltu 1,l lot' U,u ~,., 1,,t,lu •
only, tte dt~njtion can ~.1€ g'::r11~raHZ1~ l'J> ttwt a
J,,,J,J,:, I',,,, u vu, jut,lu, wl11 11u, ;,;f ,, ,,,,,~ I 1,,,,
ofx. e.g., uz = z - rn lrn i.E tfJJ:anJ, th,: ,1J, 1,,,,,!,,~11t 1,fJt,11'1 i wUI ui \II• I 111 ,,,,,,,, •• ,,,, ,,,· • Hl,,,,,I. IJ,,,
mtG!. '"·
1~:::::.ce:.! g';~t;r..~(,;rJ ;tit,n fi,r z wHJ d,~suJy t,,, ~jy,,-, 111:t

'JJ/JJ. £ ,./#f'(i)tft

11 1
UJs--«fJJs f ~/}//INl/'ltJdt -'1 ' J\/,/1111/1 ,,.µ1tllJA/IJ.
J~ N
 270
A T~XTBOOK OF ENGINEERING
MA,H~M
3.13.2 In Case of Dis cre te Distributio
We know that, for a variable x,
n of the Variable x 7' . /\lie~

V = rxr . P·
r ·,
where Pis the probability tha t the vari able takes on the value x.
If z is any function of x, we get r th 111omen
t for z by the relation
vr = u"P
and the n1on1ent generating function is
given by
Mz(t) = r.e'·2 p
To verify tha t this function generates mom
ents, we will expand etz and then
~~, ·•.(l)
~~
... M (t) =
z_
L ( 1 + tz +-2!2 z •+ ...JP = tP + ttzP + -
t 2 t2
t 2
2! z -P+ ...
t2
=v +tv +- v +"·
0 1 2! 2

.
In this case, we can also show tha t v = dr
-M z(t )
r dt'
t=O
M(t)' is clearly the expected value
of etx and hence can be written as E(etx)
the moment generating function in case which .
of discrete as well as conti~uous variable
Expectation of an;Y function $(x) is given s. gives
by
E{$(x)} = L $(x) f (x) I for discrete distribution
or, E{c!>(x)} = [ cl>(x) f(x) dx
I for continuous distribution
Eqn. (1) can also be rewritten as

Mx-a(t) = E [et(x-a)] = L/<xi

i
-a) pi = e-at ·r
i
~txi pi = e-at Mo(t)
Therefore the moment generating functio
n about the point 'a' is equal to e-«t times
moment generating function about the orig the
in.
Note. m.g.f. is not always defined since
E{ I etx I} does not always exist for all values oft.
e.g., if ftx) = : 2,x = 1, 2, 3, ... then m.g.f. does not exist.
1t X

m.g.f. always exists fort = 0 since Mx=oCO)

= 1.
3.13.3 Pro per ties of Moment Ge ner
atin g Functionl'-
(M. T. U. 2013)
(1) The moment generg,ting function
of the sum of two independent chance vari
product of their respe~tive ables is the
moment generating functions.
Symbolically, Mx+y(t) = Mx(t) x MY(t) pro
vided that x and y are independent random
variables~ • •
 E:vCH:_:N~fQ:_U_E_S_-_1_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
.,r,cAL r -
'1AT1cs
Y f. Lett
andY be two independent random variables so that x + y is also a random
. . .
,;,,,The ro.g. f• of the sum x + y w.r.t.
.
ongin 1s
M
x+y
(t) = E{e'U + Y>} = E(et.x . ety) =E(etx) . E(ety)
-· . "
.
~rexa ndy are independent variables . and so are etx and ety. i.
.•. Mx+y (t) = M.X (t) . MY(t) II
14
Hence the theo·rem . •
"
•
(1) !Effect ofchange of origin and scale on m.g.f. •
•
•
:-Ill
M u(t). = e-a1 11a M% (ti h) where u = x-a
h •
ltto! ~t u be. 8 .. mndom
.new . . en by u = x - a so that x -_ a + hu
. vanable.giv
JJ... I h_
 M u (t ) :::: E (e tu )

== E et (- x~ a) = e- ~ Mx (~J
McX(t) =M x( ct ), c b ei n g a constant.

deflllitiorl, L ll S == Mex (t )
~~f
= E(el<x) = Mx (c t) = RHS.
./ ult.
" tbe re 5
~ence [1 L L U S T R A T IV
E EXAMPLES \
.• d th e m o m en t g en
e 1. F in er a ti n
g fu n ct io n o f th e ex po
1 ne nt ia l di st ri bu ti on
~jJllll'l f(x) = -;; e- xl c; 0 <
x < oo , c > 0
a n a n d st a n d a rd d ;>J (M.T .U. 2014
e fi,r1,d its me ra ti n g fu n ev ia ti o n .
eenC ent gene ct io n a b o u t th e o ri g in
.
is g iv en b y
~1. ?Joitl
~
M (t ) = . Joo tx 1 -x lc d
X
e •- e x
O C

00

- -
1 Joo (t - l \
e\ c) d x = -1
X •
Jt-~}
- C O
C
~-~) 0
= (1 - ct)- 1 = 1 + ct
+ c2 t2 + c3 t3 + ....
. _f.!!:._ M /t ) \ =(c + 2c t + 3c
I 2 3 2
Ldt
\! 1 -
Jt=O
t + •· ·\ = o = c
 ~
~272 "'
OK OF ENGINEERING MATHEMAT1cs
Arexra o~.=---- ---~ r
,f/
.-c ::: Joou 2 (u - 1)-t du Where 1 -
1 +e i::: u
and 1
~ - e-z d
No~, mean

v 2 = [:,: M:c<t>]
r=O
=2c2
0
:::
•
1 -t
v (1- v) t du
V-I --
Where
~ d
1
- 1 z::: du
x = v1 = c_ = 2c 2 -c2 = c2 ::: ~(1 - t, 1 + t) ; 1- t > o u v::: l du
Vari ance µ2 = v - x 2 = v2 - 2
2 v1 ::: nt cosec nt, t < 1.
:. Stan dard devia tion =
Exam ple 2. Obta in the mom ent=gener
c. . fi Ji;
,,·on of the rand om variable x havinn Find the moment generatin g function
,ahng unc Je 5. - '}.,e-kc; x, A > 0. of the negative
prob abili ty distr ibuti on ."l5
p9"'p f(x) - "° oo
exponential funct'
x, for O< x < 1
r.1
r
(t) ::: A e'x •e-AX dx = Ar e(t - ).)x
Jo ,1•. - 1 r - (A.K.r.u. 2~:;)
f(x) = 2 - x, for 1 ~ x < 2 [G.B.T.U. 2013) 901•
Jo u-t - "'Jn e <A-th: dx
{ o, elsewhere x A ( t )-1 • t )' o
Also deter mine mean v , v an.d variance
1 2 µ2.
= H = 1-A. = ; ;bi
00 (
I
Sol. M/t) = E(e'x)
6 Find the moment generating function of the d' .
= 1: X. etx dx, + Ii2 (2 - X) etx dx, + fo O. etx dx,
~pi e
~,
•
PX() -_ ncX i' qn-x (where q = 1-p)
"'rete b'11omial distribution
1 2 ~· fi st and second moments about the mean
xerx etx) (2etx _ xeb: et:c)
·d the ir . . ·. . . .
= --- + - ftlSO fin enerating function about the ongin
( t t2 + 2 r.{oJlleOt g 18 given by (A.K.T.U. 2021)
t t t SO.I
0 1
M(t) = ~ etX . nex • p Xqn - X
e1 e' % = ~ ncx (pet}' qn-x = (q +pet)1'
- --- +-1+ i(2e2
-t
1 _ 2e2t ~)- (2e t _{ +{)] = e2t _ 2e' + 1
+ t2
- t t2 t2 t
2

t t t
2
t
2
V
1
= [!£ Mx
dt
(t)]
t=o
= [n(q + pet)1'- 1 . pet ]t: =np
0 ISince q +p =1
t2 t3
2 (t+ -+- + ...
e1 -1) 2! 3! )
V2 =[ d22 Mx( t)]
= - t-
( = t2 = l + t + t + ...
2
dt
r~
t=O
= [np{et. (n -1) (q +pe'Y'- 2pet+ (q ~pe'ya-1,
Mean= vl = et}]t=O
Mx(t )] =1 = [np (q + petY' - 2 . e' {~n - l)pe' + (q +pe')}], =
dt t=O
0
Similarly, v = 2, µ = v - x2 = v - v/ = 2.:.... 2 = [np(q + pe'Y' - 2 . e'(q + npe')]t. o
2 2 2 2 (1) = 1 = Variance.
Example 3. Find the moment generating funct = np(q + np)
v, = (r + 1) ! 2'.
ion ofthe random variable whose moments 1I I·: q +P= 1
= npq + n2p2
Hence first and second moments about the mean are
L etx P(X = x)
00
Sol. M_/t) = E(etx) =
given by
µ1 = 0
X =0
oo t' Since x = v1 = np
= L- r.1
oo
v, = "!... (r+l )!.2 ' =
L-r t
r
L (r+ 1)(2tY.
oo
~ = V2 - x2 = V2- vf = npq + n'1p2 - n2p2 = npq
r= O r=O - ' r=0
= 1 + 2.2t + 3 . (2t) 2 + .... = (1- 2t)-2. • Hence, mean = np, S.D. = .Jµ; = .[npq •
Exam ple 4. Find the moment generating funct Example 7•Find the moment generating funct ion of the discrete Poisson distribution
ion ofthe probabil~tydistrii~tion function
f(z) = e-z (1
+ e-z;-2, - oo < z < oo. ~ P(x) = e-l · ')..x . A/,so find the first and second moments about
Sol. Mz (t) = E(e'z) the mean.
x! (M.T.U. 2013)
= J:ooetz. e-zo + e-zr2 dz,
SoL Moment generating function about • • 1s
the ongin • given
• bY
')..x ()Ji l )i • A(e' - ll
M/t) = r.ea . e_,. . _ = e_,. I ,- =e-A • e = e
X! X.
-
~
274 /'
!I( •
A ~
TEXTBOOK OF ENGINEEFl!No "'~r~ 'flflll!7';i;J
~ l'~
• • -
1<C
ttNIOU
ES-I------ --------~
.
the random variable }( <18su,,,. the •ali.e 'n' tvith t•-
10. Let 2 3, •••• Find the "''""ent ge11erati11g fu11etiondhehprobability lciw

----
~
IIIIP!lill
<75
d M ( )] I A(r 1
- ll 1 _t] / flll'te/ " = 1, ' an enc, l>lean •lld
v1 = [ -d x t = e f\K. t=O =A ) ' pQ' - '•,ren distn"buti·on is a discrete
.t t=O ; ,i•~-
P..., • d1stnbution
p• . · <A,JtT.\J. 2022)
v2 = [ d: M x (l)] = [A {e' . e"<r' - n • 'A.e' + eA<e' - 1) e')] / 'ii£· ']'b• gt M (t) = L e•• pq• - t = - l: (e' q)" =E. (1- e'q)" =-:::-£___
dt t=O '=O I
' 901• n
ating function. q q 1 q(l-qe')
= ~
= {)..e•1•'- 11 e' (A e' + 1)],=0 =A(A + 1) oment gener [d ] . P[ qet ] 1
Hence, the first and second moments about the mean are given by 1b' e J1l _ - M. (t) -o =-:--!'._ =-E.. =_
µl =0 ·ob jS tb vt - dt •=o q (1- qe')2 •=o (1- q) p p
Since v 1 = x =A 2 2
µ2 = V2 - x2 =V2 -
Example 8. Find the moment generating function of the continuous n
v/ = A (A+ 1) - ).
2
=A V2 = [:t: M.(t)L p[¾{~}L
=
I • orrnal distnb .
J --. <.r-µJ· lltto,i (l-qet)2 .et ~et .2(1~qet)(-qe,2]
given by f(x) = ./2ir. e 2
cr ; - oo < x < oo •
=p [ •• (1-qe')'
o 21t 1=0
Sol. Moment generating function about the origin is defined as
(1-q)2 + 2q (1-q)] =.!.+~
1 (x-µ)
2
=p [ (l-q)4 P P
M/t) = E (e0 ) = f°" ix·
1
e-
2 ~ dx
_, cr✓
2n Mean = x = v1 = p1
eµt Joo
= ../2ii, ___ e- 21 z2 . e-tcrz dz x-µ Van•ance =µ2 = v 2 - -2 2q -l..:.!L
1 P
where z::--=.... x = -+2
P P2 P2
0
2 2
= -1- e(µ1+.!.1
2 1
cr ) Joo --(z-tcr)2
./2n 2
--00
e dz \TESTYOURKNOWLEOOEI
=e
µt+.!.t2er2
2 .l=e
µt+.!.t2cr2
2 •• •100 e:-z2 dz= Ji Defi~e moment generating function. Find the moment generating' functi f
1· Xwhose prob a b'l't
11 y
fun t· · · b
c 10n 1s given y:
d •b •
on o a ran om vana le
0 2
P(X =x) =p(l - p't, x =0, 1, 2, ..., 00 (G.B.T.U. 2012)
Example 9. The random variable X assuming only non-negative values has a Gamma
2. Define moment generating function and two properties of moment generating function with proof.
probability distribution if its probability distribution is given by (M.T.U. 2013)
a p P-1 -ax . . • } 01 00
f(x) = r~ x
00
e , x > 0, a > 0, f3 > 1 3. The probability density function of the random variable Xis~)= :a exp (- \x ; ).- <x< •
{
0, elsewhere Find moment generating function of X. Hence find the mean E(X) and variance V(X). •
- (M.T.U. 2013)
Find the moment generating function of Gamma probability distribution.
Sol. MX (t) =E(etx)
=i e oo tx aP P-1 -ax dx
·-·x e aP
=-
loo
x P-1 e -x(a-t) d,x [
Hint:Mx(t) =
· 28
1 J8
_ exp -
8 e dx+
( 8 - X)
1 t:c
28 8 exp - 8
%- )ea d,x1 r(
• f · d m variable X having the p.d.f.

8
o r~ rf3 o 4. Show that the moment generating function o ran °2t -t
= __,.,,P Uv__
ioo
yP-1 e-P d I wherey = x(a-t) so that dy = (a-t)dx 1 e -e t *o
(a-t)P r~ o Y
f(x)
-
= {3'
-l<x<2
•}
. _ -
is Mx<t> - { a; ' t =o
1 1 ( )-P · 0, elsewhere . ' defined by
= (l-:r ·r~fll= 1-!
. .· ar distribution
; It I <a. 0· Find the moment generating function for trtangul . (M.T,U- 2013)_
x, osxsl}
f(x) = {2 - ,x, 1 S XS 2