This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Limit Theorems We begin by reviewing the result that says that the correlation between two random variables , X Y is always between 1 and 1. This derives from the socalled Cauchy Schwartz Inequality (also know as the CauchyBunykowskiSchwartz inequality or, simply, Schwartzs inequality). Theorem 1 : Let X and Y have bivariate density ( 29 , f x y , and suppose that E XY exists. Then 2 2 E XY E X E Y Proof : Since the random variable 2 2 2 X Y E X E Y  is nonnegative, we have ( 29 ( 29 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 X Y E E X E Y XY X Y E E X E Y E X E Y E X E Y E XY E X E Y E X E Y E XY E X E Y  = + = + = Hence, ( 29 2 2 1 E XY E X E Y , and the result follows. In terms of covariance, 1 ( 29 [ ] ( 29 [ ] ( 29 ( 29 ( 29 [ ] ( 29 ( 29 [ ] ( 29 [ ] ( 29 2 2 , X Y Cov X Y E X E X Y E Y E X E X Y E Y E X E X E Y E y =   so ( 29 ( 29 , , 1 X Y Cov X Y X Y = Recall also that, since ( 29 [ ] ( 29 2 2 Var X E X E X = , it follows that for any two random variables, X and Y , ( 29 ( 29 2 E X Y E X Y   . Closeness of Random Variables Recall that a random variable is a realvalued function whose domain is a sample space S . There are many ways to measure the distance between two functions. As we consider these, you should think in terms of the graphs of the functions. We start with a simple example. Example 1: Let ( 29 m g x x = and ( 29 n h x x = for 0 1 x . In this example, both g and h are continuous, so ( 29 ( 29 { } max :0 1 g x h x x exists. This is simply the largest vertical distance between the graph of g and the graph of h . It is called the L distance between g and h , and is denoted by g h  . The graph that follows shows this distance for the functions ( 29 g x x = and ( 29 3 h x x = . The maximum distance occurs for 6 2 3 x = , and 0.1481 g h  = . Notice that this distance is completely determined by a single value of x , and does not take the whole graphs of g and h into account....
View
Full
Document
This note was uploaded on 04/09/2009 for the course MATH 40051 taught by Professor Mocioalca,o during the Fall '08 term at Kent State.
 Fall '08
 Mocioalca,O
 Correlation, Probability

Click to edit the document details