Multiple
Random Variables-
Let Xi, - . > Xn be n random variables . The
vector X = ( X . . . .> Xn
)
T is calledra¥
. SoX is a function from S to R" , where I , P
)
is theunderlying
probability
space i X Cs) = ( X,G) , -.., Xnls))T .The
distribution
of X is theprobability
measure on IR"induced
by
P . If Px denotes the distribution of X ,then
for ACIR",
P, CA
)
= PC Es ES ! Xo) EA))
. Just aswith
random variableswe
usually
describe the distribution of X via thefollowing
non
negative
functions on IR"
:
① Joint distribution function Cdf
)
⑦ Joint
probability
mass functionCjpmf )
, in the casewhen
Xi , -y Xn are all discrete .
⑦ To .it
probability density
functionCj pdf
) , inthe
case whenXi, . Xn are all continuous random variables .
For our discussion we will assume , unless otherwise
specified
,that
all the Xi' ' are discrete or all the Xi's are continuous .Joint Distribution
Functions
-
Let X = CX , . . .
, Xn )' be a random vector. The
joint
distribution function Cdf
)
of X isdefined by
Fx ( K) = PC X , E K , , . .., X, E kn
)
, where KECK , , ..>KITE R"
=p
(
{ X, ex,} n {X. SKIN . . . n { X. ski)
=p
(
{ SES :X, Alex ,}
n . . . . n { SES ! Xncs) a- kn) )
=p
(
f - o, K ,]
x C - aka) x . .. x C - o, xn))
( note : to- any 2 sets A and PS ,
A x B = { Ca , b) : a EA , be B)
,
= PCA.
)
,where
Ae - C- - ok ,It . . . x C- o,xn) C IR" .Properties
of Jon't df's-
① o E Fx (x) El for all KEIR
"
.
⑦ Fx
Csc) is continuous fromthe right
ineach component
:if ya d Ki as h → o ,
then Ling
Fx (ki. .-yki- i. Yn , Kit, , ..,kn )= Fx (ki, ..,Ki-i ,Ki ,Kit, , kn
)
Jointprobabilityhassfu.ch#
If X = (X . . ...
, Xn) T is a random vector with Xi discrete
for all it, . . .
, n
, then the distribution of X can be
specified
by
itsjoint pint
. To say that X is discrete meansthat there
is a finite or countable subset of IR" , called the
support
ofthe
distribution of X,
and
which
we will denoteby
Sx ,that
satisfies P ( X E
5×1=1
. Thejoint pmf
of X is definedby
p× (x) =P ( X = K
)
for all KEIR" . Thejpmf
pick) is defined for all KEIR" andpx
Cx) soonly
if KE f .Also ,
o a- Px (x) E l and
,c⇐§Px
Ck) = I . For and ACR" ,P ( X EA
)
= Epxck
) .KE AnSx
To.it
Probability Density
functions-
If X = (X , , . , xn)T is a random vector
with
Xi continuousfor all it, . ., n , then if there exists a function
f×
! IR" → Lo,m)satisfying
PC X EA ) =af
fxck , . . .-, xn) doc, ... dxn for any ACIR"then f*
is called a jointprobability density
function of X' .Just as continuous random variables may not have a
a continuous
random
vector may not have ajpdf
. Inthe
vector case
this
is not hard to see .Exampte
Let X , be any continuousrandom
variable .Let Xa -- X ,
' and X =CX , , Xi)T . Then if
A = { Ck, ,x,) EIR' '. K '- ki
)
we have PCXEA ) = I .If a j
then
it mustsatisfy
I =P ( X E A) =
AS
fxlx , skaldic , doc. . But there does not exist anysuch function because A has 0 area in
R2
.More
generally
,if X = (X ,
, -.> Xn )T is an n -
dimensional
continuousrandom
vector whose
support
is contained in a subset of IR" of dimension m ,with
men , then X cannot have a jExample
: MultivariateHypergeometric
Distribution- -
consider
the following experiment
. We havehi
,
objects
oftype
Inn
"
objects
oftype
rLet N = hit . .. thr be the total number of
objects
.Suppose
we draw without
replacement
nobjects randomly
. Theunderlying probability
space is CS, P)
, where S is the set
of all
possible samples
of size n that we could obtain inthis
way , and P
specifies
that everysample
in S isequally likely
.Let Xi = # of
objects
oftype
i in thesample
we draw ,i =L , . ., r .
Let X = ( X , . . ., Xr
)
' . Then X is a random vector and itsdistribution is called
the
MultivariateHypergeometric distribution with parameter
n and his . ., her . To consider thejoint pmf
of X we should first consider the
possible
values of X , i.e.,
the
support Sx
of X .ear
= 2 , h , =3 , ha = 4 , N'
- 7
4th • • • S× when h =/
3 • • • • it h = 2
z• • •
•
u h =3
, •
!#
⑨ • • JC , u n -- 4° I 2 3 11 n = 5
" h = 6
" m = 7
Note that as the
sample
size n increases thesupport
of Xstarts
hitting
constraintsimposed by
the numbers hi of eachtype
ofobject
that there are in thepopulation
.In
general
, we may write the constraints on Sx as follows :£ =
{
x = (x, , ..., Kr) E Rr : ki E { 0 , --r, hi) for i =L , . ., rand X, t ... t k, = n
}
More
explicitly
, we can write Sx asSx = { K = Ck, , ..., kn) E Rr ! x, t .- ut kn = hi ,
Max(O, h - CN -hi)) E ki E min ( n , hi
)
,and Ki is an
integer
, i =L , . ., r}
,Jo-,it
pmf
Fo- ke Sx , ( K = CK. . . .>KIT
)
p ( X = x
)
=P ( X, - K, , ---, Xr = Kr)
This is a
counting problem
. Since allsamples
in 5 areequally
likely
, the aboveprobability
is# of
samples
in S that have K, type 1objects
, ..., Krtype
robj
.-
→ total # of
samples
in S .Note that
samples
in 5 are distinct if theobjects
in thesample
are distinct , i.e., 2
samples
are distinct if theobjects
in 1sample
are not all the same as theobjects
in theother sample
( even if
they
are of the sametype )
. We havePCX
, =x. . . ., X. = xn
)
=can ,
So the
joint pmf
of X isp× Cx) =
(74)__.lh
if K -- Ck, . ..,kn5
' ESx
{
O(
'Ii ) otherwise
Alternative
Description
ofthe
MultivariateHypergeometric
Dist'n-
observe that if X = ( Xi ,. ., Xr) T has a multivariate
hypergeometric
distributionwith parameters
n , ni , . .> hrthen
thecomponents
of X have the constraint thatX , t ... . t X ,
= n
, where n is the size of the
random sample
.Then we
get
P ( X , = k , . . .
, Xr-i
-
- kn- i
)
=P(
Xi = '4 . . .-, Xr-i - Kr-i , Xin -Xi. . -Kr.)
.=
( %)
. ..III. Hn
-II
. .- xn)
That is we can
equivalently
describethis distribution
as adistribution on C Xi . . ., Xr-i
)
T . Thejoint pmf
ofXXX, . . . ., Xr ..
)
T ispxlx . . . .,xr..
)
=thx : )
. ..( % : ) (
n -ng
. . . - x. ..)
when we describe
the
distribution inthis
way the support
(
possible
values of x, , . ., Xr-i)
becomes5.
×=
{
(Xi , .., Kr..) E IR"' : O E kit . .. t kn, Eh ,max
(
o , h - ( N -ni))
E ki E m in Ch , hi)
and Ki is an
integer
, i = I . .., r -I}
Alternatively
, we can write thesupport
asSx =
{
( K, , . ., Kru ,) E Rr-t ! O E K , t . ..t kn-i E nki E { o, . .., hi}
,
i = I, .
..
, r- I
max ( o , n - CN - hi)) E ki
, i = I, . .
, r- I
}
Marginal
Distributions ( Discrete case)
#
Marginal joint pmf
's# >
Let X = ( X , . ..
, Xn
)
T be a discreterandom
vectorwith
j o , it
pmf Px
( ki . . . ., kn)
. LetLi
, . .., id)
C { Is ..., n}
and
{ ji , .. n -d}
= { I , . -s n}) Iii
. . ., id}
. We areinterested
in the
joint pmf
of ( Xi. . . .,Xia )
" , and inparticular
e howto
obtain
it fromthe
fulljoint pmf pick
, , . ., xn) .The joint pmf
of (Xii , . ., Xi,)
T is calledthe marginal
pmt
andthe
distribution of ( Xi. . . ..,
Xia)
marginalisation
of ( Xi. . ..., Xia)
'. The term
marginal
refers to the situation where we areconsidering
the joint
distribution of a set of random variablesthat
isa subset of a
larger
collection of randomvariables
.The
marginal joint pmf
of ( Xi . . - r,Xia )
"gives probabilities
of
the
formP ( Xi
,
= Ki, , ...,
Xia
- kid)
=P(
Xi. - Xi , s ..,Xia
- kid ,Xj
,ER, ..,
Xjn.IR )
=
{
Px (Y, . . ., Yn)
(Yi,..,yn) Est ! I
Yi, -Ki
, s ..., Yid --kid
where
I
= {Cx, , ...,kn) EIR" :pick
.. ...,xD >o}
is the
support
of XExample-Margina.to/-theMultivariateHypergeometr#
Suppose
{ ii. ..., id}
C { ' s --i r} and consider thejoint
distribution of ( Xi
. . . ,
Xia )
T . Thesimplest
way toget this
marginal joint
distribution is to go back to the experimentgiving
rise to the MultivariateHypergeometric
distribution and relabel allobjects
that are not oftype
in , -., id astype
O .Then we have hi
,
gbjects
oftype
i,Nia
"
objects
oftype
idN - ni,- . ..- hid
objects
oftype
Oand
then
,
letting
Xo denote the number ofobjects
oftype
o in oursample
,P
(
Xi, = Ki , , . .., Xia - Kid
)
=P (
Xi,
- Xi, s - →
Xia
-- did , Xo = n - ki. - - n-Kid )
=
cail
. ..iii. KI
:: : :
:: Ii :)
-
④
for K = (Kii , .., kid ) e
Sx
CRd
, where
Sx
={ (
ki, . ..,kid)
ENd
! O E Xi, t - - tkid E nma x (o
, n - ( N - ni
.
) )
E Xin Emin(h , nie)
and kin is an