Discrete Case

Joint CDF

The joint CDF of r.v.s and is the function given by

The joint CDF of r.v.s. is defined analogously.

Joint PMF

The joint PMF of discrete r.v.s and is the function given by

The joint PMF of discrete r.v.s is defined analogously.

We require valid joint PMFs to be nonnegative and sum to , where the sum is

Marginal PMF

From the joint distribution of and , we can get the distribution of alone by summing over the possible values of . In the context of joint distributions, we call this the marginal or unconditional distribution of

For discrete r.v.s and , the marginal PMF of is

The marginal PMF of is the PMF of , viewing individually rather than jointly with . This operation is known as marginalizing out

Conditional PMF

For discrete r.v.s. and , the conditional PMF of given is

This is viewed as a function of for fixed

Using LOTP, we have another way of the marginal PMF

Continuous Case

Joint PDF

If and are continuous with joint CDF , their joint PDF is the derivative of the joint CDF with respect to and

We require valid joint PDFs to be nonnegative and integrate to

Marginal PDF

For continuous r.v.s and with joint PDF , the marginal PDF of is

Conditional PDF

For continuous r.v.s and with joint PDF , the conditional PDF of given is

for all with .

Continuous form of Bayes' rule

For continuous r.v.s and , we have the following continuous form of Bayes' rule:

Independence of continuous r.v.s

Random variables and are independent if for all and ,

If and are continuous with joint , this is equivalent to the condition

for all , and it is also equivalent to the condition

for all such that

Tip

The marginal PDF of , , is a function of only; it cannot depend on in any way. The conditional PDF can depend on in general. Only in the special case of independence is free of

Suppose that joint PDF of and as

for all and , where and are nonnegative functions. Then and are independent. Also, is either or is a valid PDF, then the other one is a valid PDF too and and are the marginal PDFs and , respectively.