Distribution Functions

Definitions
Suppose that X is a real-valued random variable. The (cumulative) distribution function of X is
the function F given by

F(x)=P(Xx),xR
This function is important because it makes sense for any type of random variable, regardless of
whether the distribution is discrete, continuous, or even mixed, and because it completely
determines the distribution of X. In the picture below, the light shading is intended to represent a
continuous distribution of probability, while the darker dots represents points of positive
probability; F(x) is the total probability mass to the left of (and including) x.

We will abbreviate various limits of F as follows:

F(x+)=limtxF(t),F(x)=limtxF(t),F()=limtF(t),F()=limtF(t)

Basic Properties
The properties in the following exercise completely characterize distribution functions.

F is increasing: if xy then F(x)F(y).

F(x+)=F(x) for xR. Thus, F is continuous from the right.
Proof:

Fix xR. Let x1>x2> be a decreasing sequence with xnx as n. The intervals

(,xn] are decreasing in n and have intersection (,x]. The result now follows from the
continuity theorem for decreasing events.

F(x)=P(X<x) for xR. Thus, F has limits from the left.

Proof:

Fix xR. Let x1<x2< be an increasing sequence with xnx as n. The intervals

(,xn] are increasing in n and have union (,x). The result now follows from the
continuity theorem for increasing events.

F()=0.
Proof:

Let x1>x2> be a decreasing sequence with xn as n. The intervals (,xn] are

decreasing in n and have intersection . The result now follows from the continuity theorem for
decreasing events.

F()=1.
Proof:

Let xa<x2< be an increasing sequence with xn as n. The intervals (,xn] are

increasing in n and have union R. The result now follows from the continuity theorem for
increasing events.

The following exercise shows how the distribution function can be used to compute the
probability that X is in an interval. Recall that a probability distribution on R is completely
determined by the probabilities of intervals; thus, the distribution function determines the
distribution of X.

If a,bR and a<b then

1. P(X=a)=F(a)F(a )

2. P(a<Xb)=F(b)F(a)
3. P(a<X<b)=F(b )F(a)

4. P(aXb)=F(b)F(a )

5. P(aX<b)=F(b )F(a )

Proof:

These results follow from Theorems 1-5 and the difference rule: P(BA)=P(B)P(A) if

A,B are events and AB.

Conversely, if a Function F on R satisfies the properties in Theorems 1-5, then the formulas in
Exercise 6 define a probability distribution on R, with F as the distribution function. For more
on this point, read the section on Existence and Uniqueness.
If X has a continuous distribution, then the distribution function F is continuous in the usual
calculus sense. Thus, the two meanings of continuous come together.
Proof:

If X has a continuous distribution, then by definition, P(X=x)=0 so P(X<x)=P(Xx) for

xR. Hence from Theorems 2 and 3, F(x)=F(x+)=F(x).

Relation to Density Functions

There are simple relationships between the distribution function and the probability density
function.
Suppose that X has discrete distribution on a countable subset SR. Let f denote the
probability density function and F the distribution function.

1. F(x)=tS,txf(t) for xR

2. f(x)=F(x)F(x) for xS
Proof:

Part (a) follows from the definitions and the additivity of probability. Recall that

F(x)=P(Xx) and f(x)=P(X=x) for xR. Part (b) is a restatement of Theorem 6 (a).
Thus, F is a step function with jumps at the points in S; the size of the jump at x is f(x).

There is an analogous result for a continuous distribution with a probability density function.
Suppose that X has a continuous distribution on R with probability density function f (which we
will assume is piecewise continuous) and distribution function F.
1. F(x)=xf(t)dt for xR.
2. f(x)=F(x) if f is continuous at x.
Proof:

Part (a) follows from the definition of a PDF for a continuous distribution. Part (b) follows from
(a) and a theorem from calculus.

The result in the last exercise is the basic probabilistic version of the fundamental theorem of
calculus. For mixed distributions, we have a combination of the results in the last two exercises.
Suppose that X has a mixed distribution, with discrete part on a countable subset SR, and
continuous part on R. Let g denote the partial probability density function of the discrete part, h
the partial probability density function of the continuous part, and F the distribution function.

1. F(x)=tS,txg(t)+xh(t)dt for xR
2. g(x)=F(x)F(x) for xS
3. h(x)=F(x) if xS and h is continuous at x
Naturally, the distribution function can be defined relative to any of the conditional distributions
we have discussed. No new concepts are involved, and all of the results above hold.
Suppose that X has a continuous distribution on R with probability density function f that is
symmetric about a point a. That is, f(a+t)=f(at) for tR. The distribution function F
satisfies F(at)=1F(a+t) for tR.
Proof:

F(at)=P(Xat)=atf(x)dx. Using the change of variables u=2ax gives

F(at)=t+af(u)du=P(X>t+a)=1F(t+a).

Reliability
Suppose again that X is a random variable with distribution function F. A function that clearly
gives the same information as F is the right tail distribution function:

G(x)=1F(x)=P(X>x),xR
Give the mathematical properties of a right tail distribution function, analogous to the properties
in Theorems 1-5.
Suppose that T is a random variable with a continuous distribution on [0,) with probability
density function f. If we interpret T as the lifetime of a device, then the right tail distribution
function G is called the reliability function: G(t)=P(T>t) is the probability that the device
lasts at least t time units. The function h defined below is called the failure rate function:

h(t)=f(t)G(t),t0
Note that h(t)dtP(t<T<t+dtT>t) if dt is small.
Proof:

P(t<T<t+dtT>t)=P(t<T<t+dt)P(T>t)f(t)dtG(t)=h(t)dt
Thus, h(t)dt is the approximate probability that the device will fail in next dt time units, given
survival up to time t. Moreover, the failure rate function completely determines the distribution
of T.
The reliability function can be obtained from the failure rate function.

G(t)=exp(t0h(s)ds),t0

Proof:

At the points of continuity of f we have G(t)=f(t). Hence

t0h(s)ds=t0f(s)G(s)ds=t0G(s)G(s)ds=ln[G(t)]

The failure rate function h satisfies the following properties:

1. h(t)0 for t0
2. 0h(t)dt=
Proof:

Part (a) follows from the definition. Part (b) follows from Theorem 14 and the fact that G(t)0
as t.
Conversely, suppose that h is a piecewise continuous function on [0,) that satisfies the
conditions in Exercise 15. The formula in Exercise 14 defines a reliability function for a
continuous distribution on [0,)
Proof:

The function G is continuous, decreasing, and satisfies G(0)=1 and G(t)0 as t. Hence

F=1G is the distribution function for a continuous distribution on [0,)

Multivariate Distribution Functions

Suppose that X and Y are real-valued random variables for an experiment, so that (X,Y) is
random vector taking values in a subset of R2. The distribution function of X,Y) is the function

F defined by
F(x,y)=P(Xx,Yy),(x,y)R2

In the graph above, the light shading is intended to suggest a continuous distribution of
probability, while the darker dots represent points of positive probability. Thus, F(x,y) is the
total probability mass below and to the left (that is, southwest) of the point (x,y). As in the
single variable case, the distribution function of (X,Y) completely determines the distribution of

(X,Y).
Suppose that a, b, c, d are real numbers with a<b and c<d. Then

P(a<Xb,c<Yd)=F(b,d)F(a,d)F(b,c)+F(a,c)
Proof:

Note that {Xa,Yd}{Xb,Yc}{a<Xb,c<Yd}={Xb,Yd}. The intersection

of the first two events is {Xa,Yc} while the first and third events and the second and third
events are disjoint. Thus, from the inclusion-exclusion rule we have

F(a,d)+F(b,c)+P(a<Xb,c<Yd)F(a,c)=F(b,d)
In the setting of the previous exercise, give the generalizations of the results in Exercise 6. Than
is, give the appropriate formula on the right for all possible combinations of weak and strong
inequalities on the left.
The joint distribution function determines the individual (marginal) distribution functions.
Let F denote the distribution function of (X,Y), and let G and H denote the distribution
functions of X and Y, respectively.

1. G(x)=F(x,)
2. H(y)=F(,y)
Proof:

These results follow from the continuity theorem for increasing events. For example, in (a)

P(Xx)=P(Xx,Y)=limyP(Xx,Yy)=limyF(x,y)
In the context of the last exercise, X and Y are independent if and only if

F(x,y)=G(x)H(y),(x,y)R2

Proof:

If X and Y are independent then F(x,y)=P(Xx,Yy)=P(Xx)P(Yy)=G(x)H(y) for

(x,y)R2. The converse follows from the fact that a probability distribution on R2 is
completely determined by its values on sets of the form (,x](,y] for (x,y)R2. Read
the section on existence and uniqueness of positive measures for more details.
All of the results of this subsection generalize in a straightforward way to n-dimensional random
vectors.

The Empirical Distribution Function

Suppose that (x1,x2,,xn) is a data set of observed values from a real-valued random variable

X with distribution function F. The empirical distribution function, based on the data, is defined
by

Fn(x)=1n#{i{1,2,,n}:xix}=1ni=1n1(xix),xR

Thus, Fn gives the proportion of values in the data set that are less than or equal to x. The
function Fn is an approximation to F, based on the given data set. This concept is explored in
more detail in the section on the sample mean in the chapter on random samples.
http://www.math.uah.edu/stat/dist/CDF.html
Distribution Functions
Consider a real-valued random variable X. The properties of X are described by its
probability
function PX , which gives the probability that X A for any set A R. However, it is
also possible to specify the distribution of a random variable by considering Pr(X A) for a
limited class of sets A; this approach has the advantage that the function giving such
probabilities
may be easier to use in computations. For instance, consider sets of the form (, x],
for x R, so that PX {(, x]} gives Pr(X x). The distribution function of the distribution
of X or, simply, the distribution function of X, is the function F FX : R [0, 1]
given by
F(x) = Pr(X x), < x < .
Theorem 1.2. A distribution function F of a distribution onRhas the following properties:
(DF1) limx F(x) = 1; limx F(x) = 0
(DF2) If x1 < x2 then F(x1) F(x2)
(DF3) limh0+ F(x + h) = F(x)
(DF4) limh0+ F(x h) F(x) = F(x) Pr(X = x) = Pr(X < x).
Corollary 1.1. Let X denote a real-valued random variable with distribution function F.
Then, for x1 < x2,
(i) Pr(x1 < X x2) = F(x2) F(x1)

(ii) Pr(x1 X x2) = F(x2) F(x1)

(iii) Pr(x1 X < x2) = F(x2) F(x1)
(iv) Pr(x1 < X < x2) = F(x2) F(x1)
Any distribution function possesses properties (DF1)(DF4). Furthermore, properties
(DF1)(DF3) characterize a distribution function in the sense that a function having those
properties must be a distribution function of some random variable.
Theorem 1.3. If a function F : R [0, 1] has properties (DF1)(DF3), then F is the
distribution function of some random variable.

http://www.google.com.bd/url?sa=t&rct=j&q=&esrc=s&source=web&cd=14&cad=rja&ved=0CDcQFjADOAo&url=http%3A%2F%2Fmath.arizona.edu
%2F~jwatkins%2Fg-randomvariables.pdf&ei=w4kgUq_DIMGVrgedhIGwCA&usg=AFQjCNHOsAQu_Nzd8zxcCoBUrZx1LaYBcw&bvm=bv.51495398,d.bmk