Scribd
Upload
41 views

Math Review Revised August 30 2012

1. The document provides an overview of summation notation and its uses, including summing values with indices, double sums, and weighted sums. 2. It also discusses derivatives, including taking derivatives of simple functions, linear combinations of functions, and composite functions using the chain rule. 3. Partial derivatives of functions with multiple variables are explained as taking the derivative with respect to one variable while holding the others constant.

Uploaded by

Larry Johnson
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
41 views

Math Review Revised August 30 2012

1. The document provides an overview of summation notation and its uses, including summing values with indices, double sums, and weighted sums. 2. It also discusses derivatives, including taking derivatives of simple functions, linear combinations of functions, and composite functions using the chain rule. 3. Partial derivatives of functions with multiple variables are explained as taking the derivative with respect to one variable while holding the others constant.

Uploaded by

Larry Johnson
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

Mathematics Review

A. Summation Notation
For example:

1. Overall points: a. b. c. d. this is always just bookkeeping not necessary for adding up a handful of values extremely useful when things get more complicated and/or when N is large sum is like a DO loop in FORTRAN: sumvar = 0. DO i = 1,N sumvar = sumvar + y(i) ENDDO e. choice of running index is up to you, but make sure you pick a letter not in use for something else in whatever problem you are working on. f. whenever summation notation confuses you:

re-write the expression for N = 2 or 3 usually this will make the situation obvious

2. Single Sums: a. where g is a constant (i.e., g does not depend on the running index of the sum):

b. where k is a constant (i.e., k does not depend on the running index of the sum):

c. The summation of the sum of two variables equals the sum of the summations of the two variables:

d. Thus,

and, where xi = 1 for all values of i,

e. More explicitly, let y1 ... yN be observations on some economic variable (e.g., N observations on quarterly GDP) and let c1 ... cN be N given numbers. Then the weighted sum of the N observations can be written:

This is called a linear combination of the yis (This is important because our estimator of in a model like Yi = + xi + Ui is of this very form.)

3. Double Sums: Consider the particular double sum:

Note that there are two distinct running indices here, one for each sum. This expression can be thought of as instructions to form the single sum of the twelve terms:

or, equivalently, as the sum of the four sums:

A good albeit wordy way to think about the double sum

is to think of initially setting the value of i (the running index for the outer, or leftmost, sum) to its lower limit (which is one here) and then adding up a1,j for all of the values of j. Then one increments the value of i by one and adds up a2,j for all of the values of j. One continues in this manner until the running index of the outer sum has been used with the value specified by its upper limit. It is worth noting that this process is essentially equivalent to accumulation over a pair of nested loops in a typical programming language like FORTRAN: sumvar = 0. DO i = 1,N DO j = 1,N sumvar = sumvar + y(i) ENDDO ENDDO

We will need to consider the square of a linear combination of N observations:

This is important because showing that our estimator of in a model like Yi = + xi + Ui is a good estimator and testing a (perhaps economically crucial) hypothesis like Ho: = 0 both require an estimator of the noisiness i.e., the sampling variance of our estimator of , and it is of this very form. Note that:

{It is easier to see that equations * and ** are equivalent by observing that equation ** implies equation *. This is because the expression ciyi is a constant as far as the sum over j is concerned since it does not depend on the value of j. Consequently, it doesnt matter whether the term ciyi is inside the sum over j, multiplying each term in the sum before adding them up (as in equation **) or outside the sum over j, multiplying the value of the entire sum of the cjyjs, as in equation *.}

OR:

then

What would this look like for N = 2?

and

B. Differential Calculus

1. Derivatives of simple functions a. if f(x) = xk then df(x)/dx = kxk-1 b. if f(x) = x-1 then df(x)/dx = - x-2 c. if f(x) = log(x) then df(x)/dx = 1 / x

2. Derivative of a linear combination of functions

or

3. Derivative of a composite function (chain rule) for example:

4. Derivatives of a function of many variables (partial derivatives) If f is a function of 3 variables i.e., f(x1, x2, x3) then the partial derivative of f with respect to x1 is just the derivative of f(x1, x2, x3) with respect to x1, treating x2 and x3 as constants, and similarly for the partial derivatives with respect to x2 and x3: I.e., Mf(x1, x2, x3)/Mx2 = partial derivative of f(x1, x2, x3) with respect to x2 = derivative of f(x1, x2, x3) with respect to x2 holding x1 and x3 constant For example, if

then

You might also like