Derivatives and Integrals

The concept of Derivative is at the core of Calculus and modern mathematics. The
definition of the derivative can be approached in two different ways. One is
geometrical (as a slope of a curve) and the other one is physical (as a rate of
change). Historically there was (and maybe still is) a fight between mathematicians
which of the two illustrates the concept of the derivative best and which one is
more useful. We will not dwell on this and will introduce both concepts. Our
emphasis will be on the use of the derivative as a tool.

The Physical Concept of the Derivative

This approach was used by Newton in the development of his Classical Mechanics.
The main idea is the concept of velocity and speed. Indeed, assume you are
traveling from point A to point B, what is the average velocity during the trip? It is
given by

If we now assume that A and B are very close to each other, we get close to what is
called the instantaneous velocity. Of course, if A and B are close to each other,
then the time it takes to travel from A to B will also be small. Indeed, assume that
at time t=a, we are at A. If the time elapsed to get to B is
, then we will be at B
at time
is

. If

is the distance from A to B, then the average velocity

The instantaneous velocity (at A) will be found when

get smaller and smaller.
Here we naturally run into the concept of limit. Indeed, we have

If f(t) describes the position at time t, then

we have

. In this case,

Example. Consider a parabolic motion given by the function f(t) = t2. The
instantaneous velocity at t=a is given by

Since

we conclude that the instantaneous velocity at t=a is 2 a.

This concept of velocity may be extended to find the rate of change of any variable
with respect to any other variable. For example, the volume of a gas depends on
the temperature of the gas. So in this case, the variables are V (for volume) as a
function of T (the temperature). In general, if we have y = f(x), then the average
rate of change of y with respect to x from x = a to
is

, where

As before, the instantaneous rate of change of y with respect to x at x = a, is

Notation. Now we get to the hardest part. Since we can not keep on writing
"Instantaneous Velocity" while doing computations, we need to come up with a
suitable notation for it. If we write dxfor
small, then we can use the notation

The Geometrical Concept of the Derivative

Consider a function y = f(x) and its graph. Recall that the graph of a function is a
set of points (that is (x,f(x)) for x's from the domain of the function f). We may

draw the graph in a plane with a horizontal axis (usually called the x-axis) and a
vertical axis (usually called the y-axis).

Fix a point on the graph, say (x0, f(x0)). If the graph as a geometric figure is "nice"
(i.e. smooth) around this point, it is natural to ask whether one can find the
equation of the straight line "touching" the graph at that point. Such a straight line
is called the tangent line at the point in question. The concept of tangent may be
viewed in a more general framework.

(Note that the tangent line may not exist. We will discuss this case later on.) One
way to find the tangent line is to consider points (x,f(x)) on the graph, where x is
very close to x0. Then draw the straight-line joining both points (see the picture
below):

As you can see, when x get closer and closer to x0, the lines get closer and closer to
the tangent line. Since all these lines pass through the point (x0,f(x0)), their
equations will be determined by finding their slope: The slope of the line passing
through the points (x0,f(x0)) and (x,f(x)) (where

) is given by

The tangent itself will have a slope m, which is very close to m(x) when x itself is
very close to x0. This is the concept of limit once again!
In other words, we have

So the equation of the tangent line is

Notation. Writing "m" for the slope of the tangent line does not carry enough
information; we want to keep track of the function f(x) and the point x0 in our
notation. The common notation used is
m = f'(x0).
In this case, the equation of the tangent line becomes
y - f(x0) = f'(x0) (x-x0)
where

One last remark: Sometimes it is more convenient to compute limits when the
variable approaches 0. One way to do that is to make a translation along the x-axis.
Indeed, if we set h=x-x0, we get

Techniques of Differentiation
Maybe the easiest and most useful formulas are the ones that say that the derivative
is linear:

Combined with the formula (xn)' = n xn-1, we see that every polynomial function has
a derivative at any point.
Example. For P(x) = 1-2x + 3x4 -5 x6, we have

The next two formulas are the most powerful ones. They deal with the derivative
of a product and a quotient. They are commonly called the product rule and
the quotient rule. We have

In particular, we have

So, we have

which means that the formula (xr)' = r xr-1 is also valid for negative exponents.
Before we discuss the derivative of trigonometric functions, let us stop here and
reflect a little bit more on polynomial functions. Indeed, we saw that the derivative
of a polynomial function is also a polynomial function. So we can take another
derivative and generate a new function. This function is called the second
derivative. We can keep doing this as long as we want to. The functions obtained
are called higher derivatives. The common notations used for them are

Exercise Find the derivative of the function

e use the quotient rule. We have

which gives

The Area Problem and the Definite Integral

Integration is vital to many scientific areas. Many powerful mathematical tools are
based on integration. Differential equations for instance are the direct consequence
of the development of integration.
So what is integration? Integration stems from two different problems. The more
immediate problem is to find the inverse transform of the derivative. This concept
is known as finding the antiderivative. The other problem deals with areas and
how to find them. The bridge between these two different problems is the
Fundamental Theorem of Calculus.
We want to find the area of a given region in the plane. It is not hard to see that this
problem can be reduced to finding the area of the region
bounded above by the
graph of a positive function f (x), bounded below by the x-axis, bounded to the left
by the vertical line x = a, and to the right by the vertical line x = b.

The answer to this problem came through a very nice idea. Indeed, let us split the
region
into small subregions which we can approximate by rectangles or other
simple geometrical figures (whose areas we know how to compute). This is how it
goes: split the interval [a, b] into subintervals, preferably with the same width x,

x0 = a < x1 < x2 < ... < xn = b

with xi + 1 - xi =

x=

, for i = 0, 1, ... , n - 1.

Let
be the subregion bounded above by the graph of f (x), bounded below by
the x-axis, bounded to the left by x = xi - 1, and to the right by x = xi, for i = 1, ... , n.
Clearly we have
Area(

) = Area(

) + Area(

) + ... + Area(

) .

So we focus on the subregions

, for i = 1, ... , n. Since we want to approximate
the regions by rectangles, then we only have to worry about the upper boundary of
each region (since on the other sides we already have straight lines). Again: We are
looking for good approximations of the regions
by rectangles.
The easiest way to choose a height for our rectangles is to choose the value of the
function at the left (or right) end points of the small intervals [xi - 1, xi].

Let Li be the rectangle defined by the left-end point and Ri be the rectangle defined
by the right-end point. Then an approximation to Area( ) is given by
Area(L1) + Area(L2) + ... + Area(Ln) = xf (x0) + xf (x1) + ... +
which we will call the left-sum denoted LEFT(n), and
Area(R1) + Area(R2) + ... + Area(Rn) = xf (x1) +
which we will call the right-sum denoted RIGHT(n)
Example. Consider the function
f (x) = x2
for x

xf (x2) + ... +

xf (xn - 1)
xf (xn)

[0, 1]. Let us split the interval into 4 subintervals. We have

x0 = 0 , x1 =

We have

x=

, x2 =

, x3 =

, x4 = 1 .

and

LEFT(4) =

02 +

and

RIGHT(4) =

12 =

Note that Area( ) equals , a result which we will prove in later pages.
Indeed if the function f (x) is not too badly behaved, we will show that when n gets
larger, the numbers LEFT(n) and RIGHT(n) get closer to Area( ), i.e.
Area( ) =
LEFT(n) =
RIGHT(n) .
This is the main idea described above. The number Area( ) is called the definite
integral (or more simply the integral) of f (x) from a to b and is denoted by
f (x) dx .

Note that in the expression

variable.
Example. Let

0. Then we have

This is true since the region

Example. We have

Indeed, the region

triangle.

f (x) dx the variable x may be replaced by any other

dx = (b - a) .
is simply a rectangle.

x dx = (b2 - a2) .
is simply the union of two regions: one rectangle and one

The rectangle (depicted in red) is bounded above by x = a and its area is a(b - a).
The triangle (in blue) is determined by the points: (a, a), (a, b), and (b, b). Its area
is

(b - a)2. So we have

xdx = a(b - a) + (b - a)2 = (b2 - a2) .

A precise definition for the definite integral involves partitions and lower as well as
upper sums:
Definition. A partition P of the interval [a, b] is a sequence of numbers {xi;i = 0,
1, ... , n} such that
x0 = a < x1 < x2 < ... < xn = b
For a function f (x) defined on [a, b] and a partition P of [a, b], set
mi = inf{f (x); x [xi - 1, xi]} and Mi = sup{f (x); x [xi - 1, xi]}
for i = 1, ... , n, provided that f (x) is bounded on [a, b]. The sum
Lf(P) = m1(x1 - x0) + m2(x2 - x1) + ... + mn(xn - xn - 1)
is called the lower sum for f (x) over the partition P, and
Uf(P) = M1(x1 - x0) + M2(x2 - x1) + ... + Mn(xn - xn - 1)
is called the upper sum for f (x) over the partition P.

Theorem. We have
Lf(P)
Area( )
Uf(P)
for any partition P of [a, b]. Moreover if f (x) is continuous on [a, b], except maybe
at a finite number of points, and I is a number such that
Lf(P)
I
Uf(P)
for any partition P of [a, b], then I = Area( ).
This theorem is fundamental. Let us illustrate this with the following example.
Example. Use the above theorem to show
x2dx =

(b3 - a3)

where b
a
0. Let P = {x0, x1, ... , xn} be a partition of [a, b]. Since f (x) = x2 is
increasing on [a, b], then mi = xi - 12 and Mi = xi2. So we have
Uf(P) = x21(x1 - x0) + x22(x2 - x1) + ... + x2n(xn - xn - 1)
and
Lf(P) = x20(x1 - x0) + x21(x2 - x1) + ... + x2n - 1(xn - xn - 1) .
For each i, we have
xi - 12
since xi - 1

xi - 12 + xi - 1xi + xi2

xi 2

xi. If we multiply by xi - xi - 1, we get

xi - 12(xi - xi - 1)

xi - 12 + xi - 1xi + xi2 (xi - xi - 1)

xi2(xi - xi - 1) .

But
xi - 12 + xi - 1xi + xi2 (xi - xi - 1) = xi3 - xi - 13
which implies
xi - 12(xi - xi - 1)

(xi3 - xi - 13)

xi2(xi - xi - 1) .

Hence
Lf(P)

(b3 - a3)

Uf(P)

since
(x13 - x03) + (x23 - x13) + ... + (xn3 - xn - 13) = b3 - a3 .

Properties of the Definite Integral

The following properties are easy to check:
Theorem. If f (x) and g(x) are defined and continuous on [a, b], except maybe at a
finite number of points, then we have the following linearity principle for the
integral:
(i)
f (x) + g(x) dx =

f (x) dx +

g(x) dx;

(ii)
f (x) dx =
f (x) dx, for any arbitrary number .
The next results are very useful in many problems.
Theorem. If f (x) is defined and continuous on [a, b], except maybe at a finite
number of points, then we have
(i)
f (x) dx = 0;
(ii)
f (x) dx =

f (x) dx +

f (x) dx;

(iii)
f (x) dx = -

f (x) dx;

for any arbitrary numbers a and b, and any c [a, b].

The property (ii) can be easily illustrated by the following picture:

Remark. It is easy to see from the definition of lower and upper sums that if f (x)
is positive then
If f (x)

f (x) dx
g(x) for x

0. This implies the following

[a, b]

f (x) dx

g(x) dx .

Example. We have
(x2 - 2x)dx =
We have seen previously that
x2 dx =

x2 dx - 2

and

x dx .

x dx =

Hence
(x2 - 2x)dx =

-2

=-

Exercise 1. Given that

f (x)dx = 2 ,
find

f (x)dx = - 1 ,

f (x)dx.

We have
f (x)dx =

f (x)dx +

f (x)dx

which implies
f (x)dx =

f (x)dx -

f (x)dx = - 1 - 2 = - 3

Techniques of Integration
Integration by Parts
One of very common mistake students usually do is

To convince yourself that it is a wrong formula, take f(x) = x and g(x)=1. Therefore,
one may wonder what to do in this case. A partial answer is given by what is
called Integration by Parts. In order to understand this technique, recall the
formula

which implies

Therefore if one of the two integrals

and

is easy to

evaluate, we can use it to get the other one. This is the main idea behind Integration
by Parts. Let us give the practical steps how to perform this technique:
1
Write the given integral

where you identify the two functions f(x) and g(x). Note that if you are given
only one function, then set the second one to be the constant
function g(x)=1.
2
Introduce the intermediary functions u(x) and v(x) as:

Then you need to make one derivative (of f(x)) and one integration (of g(x))
to get

Note that at this step, you have the choice whether to differentiate f(x)
or g(x). We will discuss this in little more details later.

3
Use the formula

4
Take care of the new integral

The first problem one faces when dealing with this technique is the choice that we
encountered in Step 2. There is no general rule to follow. It is truly a matter of
experience. But we do suggest not to waste time thinking about the best choice,
just go for any choice and do the calculations. In order to appreciate whether your
choice was the best one, go to Step 3: if the new integral (you will be handling) is
easier than the initial one, then your choice was a good one, otherwise go back to
Step 2 and make the switch. It is after many integrals that you will start to have a
feeling for the right choice.
In the above discussion, we only considered indefinite integrals. For the definite
integral

, we have two ways to go:

1
Evaluate the indefinite integral

which gives

Use the above steps describing Integration by Parts directly on

the given definite integral. This is how it goes:
(i)
Write down the given definite integral

where you identify the two functions f(x) and g(x).

(ii)
Introduce the intermediary functions u(x) and v(x) as:

Then you need to make one derivative (of f(x)) and one integration (of g(x))
to get

(iii)
Use the formula

(iv)
Take care of the new integral

References
http://www.sosmath.com/calculus/calculus.html
http://hyperphysics.phy-astr.gsu.edu/hbase/math/derint.html