Sei sulla pagina 1di 5

Riemann Sums Explanation

Z. A. Cameron
can you tell me why does dx always go away after integration?
Look here. This image shows four instances of a graph of f (x) = (x2 +
1) x.
3

R
What (x2 + 1)3 x dx is doing is finding the area between the curve
f (x) = (x2 + 1)3 x and the x-axis. You can approximate this area with
rectangles packed side by side that each have a width of some number x,
and that are stretched either up or down from the x-axis until their left corner
touches the graph.

Look at the top-left graph in the image. There are 10 rectangles which are
approximating the area. Each rectangle has a width of x = 0.2, and each
one is stretched from the x-axis until their left corner touches the graph.
Lets look at only the blue rectangles. Look at any one of those blue
rectangles. Notice that their left corner (the one touching the graph) is
positioned either above or below some point x along the x-axis. This means
that their height is (x2 + 1)3 x.
For example, look at the blue rectangle under which I wrote ` x a. This
rectangles left corner is positioned over the point x = 0.6 on the x-axis. So,
its height (rounding to three decimal places) is:
f (0.6) = (0.62 + 1)3 0.6 = 1.509
and so this rectangles area would be height times width:
f (0.6) x = 1.509 0.2 = 0.3018.
1

So the total area between the blue half of the curve and the x-axis is going
to be the sum of the area of all 5 rectangles:
4
X

f (ix)x = f (0)0.2+f (0.2)0.2+f (0.4)0.2+f (0.6)0.2+f (0.8)0.2

i=0

This approximation is equal to about 1.177, which is not too far off from the
actual area which is exactly 1.875.
Why do I say there are 5 blue rectangles when you can only see 4? Because
0
) 0.2 = 0.
the one you cant see has its left corner at x = 0, so its area is f ( 10
Without any area, theres no rectangle to draw!

Now look at the top-right graph. Now there are 20 rectangles approximating the area, where each rectangle has a width of x = 0.1.
Lets again add up the area of all 10 blue rectangles:
9
X

f (i x) x

i=0

This approximation is about 1.5, which is getting closer to the actual area of
1.875.

Now look at the bottom-left graph, which now has 40 rectangles with width
x = 0.05. The sum of the 20 blue rectangles is:
19
X

f (i x) x

i=0

This approx. is about 1.681, which is still getting closer to the actual area
of 1.875.
As youd expect, the bottom-right graph with 40 blue rectangles with
widths x = 0.025 is an even better approximation for the area under that
half of the curve.
2

Total area of those 40 blue rectangles: 1.777


Total complete area: 1.875

You can see a pattern emerging here. Since weve been looking at the blue
half of the graph of f (x) = (x2 + 1)3 x from x = 0 to x = 1, for any number
N of blue rectangles, x = N1 , and the approximate area under that half of
the curve is:
N
1
X
f (i x) x
i=0

However, for any value of *N*, this sum will always be an approximation
and not the exact answer. It doesnt matter if N =
2
22
2

(22 )2

((22 )2 )(2
2

2 )22

(((22 )2 )(2
2

2 )22

2 )22

((((22 )2 )(2

2 )22 )(22 )2

)((2

2 )22 )(22 )2

)((2

2 )22 )(22 )2

)(((2

2
2 )22 )(22 )2

)((2

etc . . .
No matter what value of N we choose to approximate the area under the
curve, there will always be larger values of N that will give a more accurate
estimate but never an exact value for the area under the curve.
So instead of picking particular values for N , we instead need to consider
what happens as N continues to get larger and larger; in other words, as N
tends to infinity. Its important to note that we cant just consider N =
because infinity is not a real number.
Now were really getting to your question.

As N tends to infinity (commonly written as N , meaning N tends


to infinity or is tending to infinity), x 0. In the case that x tends
to 0, x becomes known as an infinitesimal and we instead represent x as
dx, which is called a differential. Also, since N and x 0, instead
of writing the sum of the area of N rectangles under the curve as
N
1
X

f (i x) x

i=0

we write this sum in the integral notation you are familiar with:
Z
f (x) dx
R
where the integral symbol actually evolved from an elongated S for Sum.
So you can see that:
P
the
symbol is no longer adequate since were noR longer summing
over discrete intervals, so instead we replace it with
f (i x becomes f (x) because x is no longer restricted to being a
multiple of x but instead can be any real number
x becomes dx
So the dx which is not a number in the same sense that infinity is not
a number in an integral expression is a way of expressing the concept that
we are summing all the values of f (x) over each and every value of x on
the x-axis, much the same way that we were summing all the areas of N
rectangles with heights f (x) (where x is some multiple of x) and widths
x.

Now, why does the dx always go away after integration? Well, why does
x = 0.2 go away after solving:
4
X

f (i x) x = f (0) 0.2 + f (0.2) 0.2 + f (0.4) 0.2 + f (0.6) 0.2 + f (0.8) 0.2

i=0

and getting an answer of about 1.177? The answer doesnt really have anything to do with algebra or arithmetic. Its just a matter of notation. You
have to keep in mind that all the kinds of notations used in mathematics
are only parts of a language developed for the sake of expressing abstract
ideas which by themselves have no communicable representation. Of course,
this notation language has been developed over centuries to be intricately
connected with the abstract ideas it is meant to represent, so were able
to use algebraic notations to solve for x for example, but sometimes the
language doesnt have mechanical uses (like algebraic manipulations) and is
instead meant to be a convention that expresses a particular concept of a
larger abstract idea, such as dx in the scope of the idea of integration.

Edit: Why did I only consider the blue half of those graphs and ignored
the red half? Because also considering the red half would have added an
extra idea that wouldnt have contributed to the explanation leading to the
answer for your question. That idea is that any area which is *below* the
x-axis and *above* the curve is considered to be negative area. Remember
how integrating f (x) = (x2 + 1)3 x from 0 to 1 resulted in 1.875? Well,
integrating from -1 to 0 is -1.875. So if you integrate from -1 to 1, or from
k to k for any value k, the result will be 0. With this in mind, and without
doing any calculations, what would be the result of the integral of sin(x)
from 0 to 2?

Potrebbero piacerti anche