Fourier Series: Philippe B. Laval Kennesaw State University March 24, 2008

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

Fourier Series

Philippe B. Laval Kennesaw State University March 24, 2008


Abstract These notes introduce Fourier series and discuss some applications.

Introduction

Joseph Fourier (1768-1830) who gave his name to Fourier series, was not the rst to use Fourier series neither did he answer all the questions about them. These series had already been studied by Euler, dAlembert, Bernoulli and others before him. Fourier also thought wrongly that any function could be represented by Fourier series. However, these series bear his name because he studied them extensively. The rst concise study of these series appeared in Fouriers publications in 1807, 1811 and 1822 in his Thorie analytique de la chaleur. He applied the technique of Fourier series to solve the heat equation. He had the insight to see the power of this new method. His work set the path for techniques that continue to be developed even today. Fourier Series, like Taylor series, are special types of expansion of functions. With Taylor series, we are interested in expanding a function in terms of the special set of functions 1, x, x2 , x3 , ... or more generally in terms of 1, (x a), 2 3 (x a) , (x a) , .... You will remember from calculus that if a function f has a power series representation at a then f (x) = f (n) (a) (x a)n n! n=0

(1)

With Fourier series, we are interested in expanding a function f in terms of the special set of functions 1, cos x, cos 2x, cos 3x, ..., sin x, sin 2x, sin 3x, ... Thus, a Fourier series expansion of a function is an expression of the form

f (x) = a0 +
n=1

(an cos nx + bn sin nx)

After reviewing periodic functions, we will focus on learning how to represent a function by its Fourier series. We will only partially answer the question regarding which functions have a Fourier series representation. We will nish these notes by discussing some applications. 1

Even, Odd and Periodic Functions

In this section, we review some results about even, odd and periodic functions. These results will be needed for the remaining sections. Denition 1 (Even and Odd) Let f be a function dened on an interval I (nite or innite) centered at x = 0. 1. f is said to be even if f (x) = f (x) for every x in I. 2. f is said to be odd if f (x) = f (x) for every x in I. The graph of an even function is symmetric with respect to the y-axis. The graph of an odd function is symmetric with respect to the origin. For example, 5, x2 , xn where n is even, cos x are even functions while x, x3 , xn where n is odd, sin x are odd. You will recall from calculus the following important theorem about integrating even and odd functions over an interval of the form [a, a] where a > 0. Theorem 2 Let f be a function which domain includes [a, a] where a > 0. 1. If f is even, then 2. If f is odd, then
a a a a

f (x) dx = 2

a 0

f (x) dx

f (x) dx = 0

There are several useful algebraic properties of even and odd functions as shown in the theorem below. Theorem 3 When adding or multiplying even and odd functions, the following is true: even + even = even even even = even odd + odd = odd odd odd = even even odd = odd Denition 4 (Periodic) Let T > 0. 1. A function f is called T -periodic or simply periodic if f (x + T ) = f (x) for all x. 2. The number T is called a period of f . 2 (2)

3. If f is non-constant, then the smallest positive number T with the above property is called the fundamental period or simply the period of f . Let us rst remark that if T is a period for f , then nT is also a period for any integer n > 0. This is easy to see using equation 2 repeatedly: f (x) = f (x + T ) = f ((x + T ) + T ) = f (x + 2T ) = f ((x + 2T ) + T ) = f (x + 3T ) . . . = f ((x + (n 1) T ) + T ) = f (x + nT ) Classical examples of periodic functions are sin x, cos x and other trigonometric functions. sin x and cos x have period 2. tan x has period . We will see more examples below. Because the values of a periodic function of period T repeat every T units, it is enough to know such a function on any interval of length T . Its graph is obtained by repeating the portion over any interval of length T . Consequently, to dene a T -periodic function, it is enough to dene it over any interval of length T . Since dierent intervals may be chosen, the same function may be dened dierent ways. Example 5 Describe the 2-periodic function shown in gure 1 in two dierent ways: 1. By considering its values on the interval 0 x < 2; 2. By considering its values on the interval 1 x < 1. Solution 1. On the interval 0 x < 2, the function is a portion of the line y = x + 1 thus f (x) = x + 1 if 0 x < 2. The relation f (x + 2) = f (x) describes f for all other values of x. 2. On the interval 1 x < 1, the function consists of two lines. So we have x 1 if 1 x < 0 f (x) = x + 1 if 0 x, 1 The relation f (x + 2) = f (x) describes f for all other values of x. Although we have dierent formulas, they describe the same function. Of course, in practice, we use common sense to select the most appropriate formula. Next, we look at an important theorem concerning integration of periodic functions over one period. Theorem 6 (Integration Over One Period) Suppose that f is T -periodic. Then for any real number a, we have
T 0 a+T

f (x) dx = 3

f (x) dx

(3)

Figure 1: A function of period 2 Proof. Dene F (a) = a f (x) dx. By the fundamental theorem of calculus, F (a) = f (a + T ) f (a) = 0 since f is T -periodic. Hence, F (a) is a constant for all a. In particular, F (0) = F (a) which implies the theorem. We illustrate this theorem with an example. Example 7 Let f be the 2-periodic function shown in gure 1. Compute the integrals below: 1. 2.
1 1 a+T

[f (x)]2 dx [f (x)]2 dx where N is any positive integer.

N N

Solution 8 1. We described this function earlier and noticed that its simplest expression was not over the interval [1, 1] but over the interval [0, 2]. 2 We should also note that if f is 2-periodic, so is [f (x)] (why?). Using theorem 6, we have
1 1

[f (x)] dx = = = =

2 0 2 0

[f (x)] dx (x + 1)2 dx
2 0

1 3 (x + 1) 3 2 3

2. We break length 2.
N N

N N

[f (x)]2 dx into the sum of N integrals over intervals of


N +2 N N +4 N+2 N N2

[f (x)]2 dx =

[f (x)]2 dx+

[f (x)]2 dx+...+

[f (x)]2 dx

By theorem 6, each integral is 2 . Thus 3


N N

[f (x)]2 dx =

2N 3

The following result about combining periodic functions is important. Theorem 9 When combining periodic functions, the following is true: 1. If f1 , f2 , ..., fn are T -periodic, then a1 f1 + a2 f2 + ... + an fn is also T periodic. 2. If f and g are two T -periodic functions so is f (x) g (x). 3. If f and g are two T -periodic functions so is 4. If f has period T and a > 0 then f period T . a
x a f (x) g(x)

where g (x) = 0.

has period aT and f (ax) has

5. If f has period T and g is any function (not necessarily periodic) then the composition g f has period T . Proof. See problems. The functions in the 2-periodic trigonometric system 1, cos x, cos 2x, ..., cos mx, ..., sin x, sin 2x, ..., sin nx, ... are among the most important periodic functions. The reader will verify that they are indeed 2-periodic. They share another important property. Denition 10 (Orthogonal Functions) Two functions f and g are said to be orthogonal over the interval [a, b] if
b a

f (x) g (x) dx = 0

(4)

The notion of orthogonality is very important in many areas of mathematics.

Theorem 11 The functions in the trigonometric system 1, cos x, cos 2x, ..., cos mx, ..., sin x, sin 2x, ..., sin nx, ... are orthogonal over the interval [, ] in other words, if m and n are two nonnegative integers, then

cos mx cos nxdx = 0 if m = n cos mx sin nxdx = 0 m, n sin mx sin nxdx = 0 if m = n

(5)

Proof. There are dierent ways to prove this theorem. One way involves using the identities sin cos cos sin sin sin cos cos = = = = 1 [sin ( + ) + sin ( )] 2 1 [sin ( + ) sin ( )] 2 1 [cos ( + ) cos ( )] 2 1 [cos ( + ) + cos ( )] 2

We illustrate the technique by proving cos mx cos nxdx = 0 if m = n. We see that cos mx cos nx = 1 [cos (m + n) x + cos (m n) x]. Therefore 2

cos mx cos nxdx = =

1 2

[cos (m + n) x + cos (m n)] dx


1 1 1 sin (m + n) x + sin (m n) x 2 m+n mn = 0

Remark 12 We also have the useful identities


cos2 mxdx =

sin2 mxdx = for all m = 0

(6)

We nish this section by looking at another example of a periodic function, which does not involve trigonometric function but rather the greatest integer function, also known as the oor function, denoted x . x represents the greatest integer not larger than x. For example, 5.2 = 5, 5 = 5, 5.2 = 6, 5 = 5. Its graph is shown in gure 2. Example 13 Let f (x) = x x . This gives the fractional part of x. For

4 3 2 1

-5

-4

-3

-2

-1 -1 -2 -3 -4 -5

Figure 2: Graph of x 0 x < 1, x = 0, so f (x) = x. Also, since x + 1 = 1 + x , we get f (x + 1) = = = = x+1 x+1 x+11 x x x f (x)

So, f is periodic with period 1. Its graph is obtained by repeating the portion of its graph over the interval 0 x < 1. Its graph is shown in gure The practice problem will explore further properties of periodic functions.

2.1

Practice Problems

1. Prove theorem 2. 2. Prove theorem 3. 3. Sums of periodic functions. Show that if f1 , f2 , ..., fn are T -periodic, then a1 f1 + a2 f2 + ... + an fn is also T -periodic. 4. Sums of periodic functions. Let f (x) = cos x + cos x. (a) Show that the equation f (x) = 2 has a unique solution. 7

-5

-4

-3

-2

-1 -1

Figure 3: Graph of x x (b) Conclude from part a that f is not periodic. Does this contradict the previous problem? 5. Finish proving theorem 11. 6. Operations on periodic functions. (a) Show that if f and g are two T -periodic functions so is f (x) g (x). (b) Show that if f and g are two T -periodic functions so is g (x) = 0. (c) Show that if f has period T and a > 0 then f f (ax) has period T . a
x a f (x) g(x)

where

has period aT and

(d) Show that if f has period T and g is any function (not necessarily periodic) then the composition g f has period T . 7. Using the previous problem, nd the period of the functions below. (a) sin 2x (b) cos 1 x + 3 sin 2x 2 (c) (d) e
1 2+sin x cos x

8. Antiderivative of periodic functions. Suppose that f is 2-periodic and let a be a xed real number. Dene
x

F (x) =

f (t) dt for all x


2 0

Show that F is 2-periodic if and only if theorem 6)

f (t) dt = 0. (hint: use

Fourier Series of 2-Periodic Functions

As noted earlier, Fourier Series are special expansions of functions of the form f (x) = a0 +
n=1

(an cos nx + bn sin nx)

(7)

To be able to use Fourier Series, we need to know: 1. Which functions have Fourier series expansions? 2. If a function has a Fourier series expansion, how do we compute the coefcients a0 , a1 , ..., b0 , b1 , ...? A general answer to the rst question is beyond the scope of these notes. In this section, we will answer the second question. In these notes, we will give conditions which are sucient for functions to have a Fourier Series Expansion.

3.1

Euler Formulas for the Coecients

The coecients which appear in the Fourier series were known to Euler before Fourier, hence they bear his name. We will derive them the same way Fourier did. This technique is worth remembering. Computation of a0 . Starting with equation 7, we integrate each side over the interval [, ] and assuming term by term integration is legitimate, we obtain

f (x) dx = a0

dx +
n=1

an

cos nxdx + bn

sin nxdx

Since

cos nxdx =

sin nxdx = 0 for n = 1, 2, 3, ..., we have

f (x) dx = a0

dx

= 2a0 1 2 9

Thus a0 =

f (x) dx

Computation of an . Again, starting with equation 7, we multiply each side by cos mx for a xed integer m 1 and integrate each side as before. We obtain

f (x) cos mxdx = a0

cos mxdx+
n=1

an

cos nx cos mxdx + bn

sin nx cos mxdx

Now, from equation 5, we have: 0 and


cos nx cos mxdx =


cos mxdx = 0, sin nx cos mxdx = if m = n . So, we are left with if m = n

f (x) cos mxdx = am


Thus am =

f (x) cos mxdx for m 1

Computation of bn . We proceed in a similar way. Starting with equation 7, we multiply each side by sin mx for a xed integer m 1 and integrate each side as before. We leave it to the reader to verify that bm = 1

f (x) sin mxdx for m 1

We summarize our ndings in the following proposition. Proposition 14 Suppose that the 2-periodic function f has the Fourier series representation

f (x) = a0 +
n=1

(an cos nx + bn sin nx)

Then the coecients a0 , an , bn for n = 1, 2, ... are called the Fourier coecients of f and are given by the Eulers formulas a0 = an = bn = 1 1

1 2

f (x) dx

(8) (9) (10)

f (x) cos nxdx for n = 1, 2, ... f (x) sin nxdx for n = 1, 2, ...

Denition 15 For a positive integer N , we denote the N th partial sum of the Fourier series of f by SN (x). So, we have
N

SN (x) = a0 +
n=1

(an cos nx + bn sin nx)

10

We now illustrate what we did with some examples. Example 16 Find the Fourier series of f (x) = sin x. Using the formulas above along with equation 7, we nd that a0 = an = bn = 1 1

1 2

sin xdx = 0

sin x cos nxdx = 0 for all n

sin x sin nxdx = 0 except when n = 1

When n = 1,we have a1 = 1. Thus, a Fourier series of sin x is sin x. Of course, this was to be expected. Example 17 Find the Fourier series of f (x) = sin x . 2 Clearly, this function is 2-periodic. Its graph is shown in gure 4.

1.4 1.2 1.0 0.8 0.6 0.4 0.2

-10

-8

-6

-4

-2 -0.2 -0.4

10

Figure 4: Graph of |sin x| 1. Computation of a0 . Using the formulas above along with equation 7,

11

we nd that a0 = = = 1 2 1 2
0

sin

x dx 2

x x x sin dx since sin is even and sin 0 on [0, ] 2 2 2

2. Computation of an . an = = = = = = 1 x sin cos nxdx 2 x 2 sin cos nxdx 0 2 2n + 1 n1 1 sin x sin x dx if n 1 0 2 2 1 2 2n + 1 2 2n 1 cos x+ cos x 2n + 1 2 2n 1 2 1 2 2 2n + 1 2n 1 4 (4n2 1)

3. Computation of bn . bn = 1

|sin x| sin nxdx


= 0 since 4. In conclusion sin

|sin x| sin nx is odd

2 4 x = + cos nx 2 n=1 (4n2 1)

To see how this series compares to the function, we will plot some of the partial sums. Let SN (x) = 2 4 + cos nx n=1 (4n2 1)
N

12

1.4 1.2 1.0 0.8 0.6 0.4 0.2

-10

-8

-6

-4

-2 -0.2 -0.4

10

Figure 5: Graph of sin x and S2 (x) 2

1.4 1.2 1.0 0.8 0.6 0.4 0.2

-10

-8

-6

-4

-2 -0.2 -0.4

10

Figure 6: Graph of sin x and S4 (x) 2

13

1.4 1.2 1.0 0.8 0.6 0.4 0.2

-10

-8

-6

-4

-2 -0.2 -0.4

10

Figure 7: Graph of sin x and S10 (x) 2 Example 18 We now look at a 2-periodic function with discontinuities and derive its Fourier series using the formulas of this section (assuming it is legitimate). This function is called the sawtooth function. It is dened by g (x) = ( x) if 0 < x 2 g (x + 2) otherwise
1 2

Find the Fourier series for this function. Plot this function as well as S1 (x) , S7 (x) , S20 (x) where SN (x) is the N th partial sum of its Fourier series. Since f is described between 0 and 2, we can use theorem 6 to compute the Fourier coecients integrating between 0 and 2. 1. Computation of a0 . a0 = = 1 2
2 0

f (x) dx ( x) dx

1 4 = 0

14

2. Computation of an . an = = = 1 1 2
2 0

f (x) cos nxdx for n = 1, 2, ... ( x) cos nxdx


2 0

2 0

1 2

cos nxdx

2 0

x cos nxdx

The rst integral is 0. The second can be evaluated by parts.


2 0

x cos nxdx = =

x sin nx n

2 0

1 n

2 0

sin nxdx

1 cos nx n2 = 0 an = 0

2 0

so 3. Computation of bn . bn = = = 1 2 f (x) sin nxdx for n = 1, 2, ... 0 2 1 ( x) sin nxdx 2 0 2 2 1 sin nxdx x sin nxdx 2 0 0

The rst integral is 0. The second can be done by parts.


2 0

x sin nxdx = = =

x cos nx n

2 0

1 n

2 0 2 0

cos nxdx

2 1 + 2 sin nx n n 2 n

Therefore bn = = 1 2 0 2 n 1 n

15

4. Conclusion. The Fourier series of the sawtooth function is g (x) = sin nx n n=1

Below, we show the graphs of S1 (x), S7 (x), S20 (x).

4 3 2 1

-6

-4

-2 -1 -2 -3 -4

10

12

Graph of the sawtooth function (black) and S1 (x) (red)

4 3 2 1

-6

-4

-2 -1 -2 -3 -4

10

12

Graph of the sawtooth function (black) and S7 (x) (red) 16

4 3 2 1

-6

-4

-2 -1 -2 -3 -4

10

12

Graph of the sawtooth function (black) and S20 (x) (red)

Remark 19 Several important facts are worth noticing here. 1. The Fourier series seems to agree with the function, except at the points of discontinuity. 2. At the points of discontinuity, the series converges to 0, which is the average value of the function from the left and from the right. 3. Near the points of discontinuity, the Fourier series overshoots its limiting values. This is a well known phenomenon, known as Gibbs phenomenon. To see a simulation of this phenomenon, visit the ??

3.2

Piecewise Continuous and Piecewise Smooth Functions

After dening some useful concepts, we give a sucient condition for a function to have a Fourier series representation. Notation 20 We will denote f (c) = lim f (x) and f (c+) = lim f (x) Remembering that a function f is continuous at c if lim f (x) = f (c), we xc see that a function f is continuous at c if and only if f (c) = f (c+) = f (c) 17

xc

xc

Denition 21 (Piecewise Continuous) A function f is said to be piecewise continuous on the interval [a, b] if the following are satised: 1. f (a+) and f (b) exist. 2. f is dened and continuous on (a, b) except at a nite number of points in (a, b) where the left and right limit at these points exist. Denition 22 (Piecewise Smooth) A function f , dened on [a, b] is said to be piecewise smooth if f and f are piecewise continuous on [a, b] that is if the following are satised: 1. f is piecewise continuous on [a, b] 2. f exists on (a, b) except possibly at nitely many points in (a, b) where theone sided limits of f at these points exists. 3. lim f (x) and lim f (x) exists. The sawtooth function is piecewise smooth. A simple example of a function which is not piecewise smooth is x for 1 x 1. Its derivative does not exist at 0, neither do the one sided limits of its derivative at 0. Denition 23 The average of f at c is dened to be f (c) + f (c+) 2 Clearly, if f is continuous at c, then its average at c is f (c). We are now ready to state a fundamental result in the theory of Fourier series. Theorem 24 Suppose that f is a 2-periodic piecewise smooth function. Then, for all x, we have f (x) + f (x+) (an cos nx + bn sin nx) = a0 + 2 n=1

3 1

where the coecients are given by equations 8, 9, and 10. In particular, if f is piecewise smooth and continuous at x, then

f (x) = a0 +
n=1

Thus, at points where f is continuous, the Fourier series converges to the function. At points of discontinuity, the series converges to the average of the function at these points. This was the case in the example with the sawtooth function. We do one more example. 18

xa

xb

(11)

(an cos nx + bn sin nx)

(12)

Example 25 (Triangular Wave) The 2-periodic triangular wave is given on the interval [, ] by h (x) = 1. Find its Fourier series. 2. Plot h (x) as well as some partial sums of its Fourier series. 3. Show how this series could be used to approximate actually 2 . Solution 26 1. We begin by plotting h (x)We see the function is piecewise + x if x 0 x if 0 x

-8

-7

-6

-5

-4

-3

-2

-1 -1

Figure 8: Plot of the triangular wave smooth and continuous for all x. Computation of a0 . a0 = = =
1 f (x) dx 2 1 2 2 2

19

Computation of an . an = = = = = = = = 1 1 2 2
0

f (x) cos nxdx

( + x) cos nxdx +

( x) cos nxdx

( x) cos nxdx replacing x by x in the rst integral


0

x sin nx n

1 n

sin nxdx

2 1 cos nx n2 0 2 1 cos n + n2 n2 2 1 (1)n n2 n2 0 if n even 4 if n odd n


2

Computation of bn . bn = 1 f (x) sin nxdx = 0 since the integrand is odd


n=0

Conclusion. 4 h (x) = + 2
N

cos (2n + 1) x (2n + 1)2

n=0

n=0

2 2 8

= = =

4 + 2 4

n=0

(2n + 1)2 1 1 1 = 1 + 2 + 2 + 2 + ... 3 5 7


n=0

3. From h (x) =

cos(2n+1)x , (2n+1)

2. Let SN (x) =

cos(2n+1)x . (2n+1)

We plot S1 (x), S5 (x) if we let x = 0, we get

2 n=0 (2n + 1)

(2n + 1)2 1

20

-8

-7

-6

-5

-4

-3

-2

-1 -1

Figure 9: Plot of the triangular wave and S1 (x)

-8

-7

-6

-5

-4

-3

-2

-1 -1

Figure 10: Plot of the triangular wave and S5 (x)

21

This allows us to approximate 2 .

3.3

Practice Problems
1 2
2 0

1. Show that another way to compute the Fourier coecients is a0 = an = bn = 1 1


2 0 2 0

f (x) dx

f (x) cos nxdx for n = 1, 2, ... f (x) sin nxdx for n = 1, 2, ...

2. In the problems below, you are given a 2-periodic function and its Fourier series. For each function, (a) Derive the Fourier series, (b) sketch the graph of the function and some of the partial sums of its Fourier series on the interval [2, 2]. (a) f (x) = |x| if x < , Fourier series: (b) f (x) = x 0 if 0x< 4 2
n=0

cos (2n + 1) x (2n + 1)2


n=1

if < x 0

, Fourier series:

1 1 4 2

(1)n 1 n2 n2

cos nx +

(1)n sin nx n

2 (1)n (c) f (x) = x2 if x , Fourier series: cos nx +4 3 n2 n=1 (d) f (x) = x if < x < , Fourier series: 2 3. Use the Fourier series of 2c) to obtain 1 1 1 2 = 1 + 2 + 2 + 2 + ... 6 2 3 4 4. Use the Fourier series of 2d) to obtain 1 1 1 = 1 + + ... 4 3 5 7 (1) n n=1
n+1

sin nx

Fourier Series of Functions with Arbitrary Periods

So far, weve worked with 2-periodic functions for convenience. Similar results exist for functions having any period. These results can be obtained in a similar 22

manner. However, there is an easier way, one calculus II students are familiar with: substitution. One way to obtain a new series representation is to perform a substitution in a known series. Suppose that f is a function with period T = 2p > 0 for which we want to nd a Fourier series. In other words f (x + 2p) = x and we dene f (x). If we let t = p g (t) = f (x) = f pt

2p Then, g has period p that is 2. So, g has a Fourier series representation

g (t) = a0 +
n=1

[an cos nt + bn sin nt]


where a0 = an = bn = Using the substitution t = x p 1 1

1 2

g (t) dt

g (t) cos ntdt g (t) sin tdt

x , we obtain p

g that is

= a0 +
n=1

an cos

nx nx + bn sin p p

f (x) = a0 +
n=1

an cos

nx nx + bn sin p p

Using the same substitution, we can express the Fourier coecients in terms of x then dt = dx so x and f . We do it for a0 . If t = p p a0 = = = So, we have the following: 23
1 g (t) dt 2 x 1 p g 2 p p p p 1 f (x) dx 2p p

dx

Theorem 27 Suppose that f is a 2p-periodic piecewise smooth function. The Fourier series of is given by

f (x) = a0 +
n=1

an cos
p p

nx nx + bn sin p p

(13)

where a0 = an = and bn = 1 p 1 p

1 2p
p p p p

f (x) dx nx dx p nx dx p

(14) (15)

f (x) cos

f (x) sin

(16) f (x) + f (x+) 2

The Fourier series converges to f (x) if f is continuous at x and to otherwise.

We nish this section by noticing that in the special cases that f is either nx p even or odd, the series simplies greatly. If f is even, then p f (x) sin is p odd so that bn = 0 and the series is simply a cosine series. Similarly, if f is odd, nx p is odd and an = 0 and the series is simply a sine series. then p f (x) cos p We summarize this in a theorem. Theorem 28 Suppose that f is 2p-periodic and has the Fourier series representation nx nx an cos + bn sin f (x) = a0 + p p n=1 Then: 1. f is even if and only if bn = 0 for all n and in this case

f (x) = a0 +
n=1

an cos

nx p

2. f is odd if and only if an = 0 for all n and in this case

f (x) =
n=1

bn sin

nx p

24

Some Applications

In the examples, we saw how we could use Fourier series to approximate . One of the main uses of Fourier series is in solving some of the dierential equations from mathematical physics such as the wave equation or the heat equation. Fourier developed his theory by working on the heat equation. Fourier series also have applications in music synthesis and image processing. All these will be presented in another talk. We will mention the relationship between sound (music) and Fourier series. nt nt an cos + bn sin , When we represent a signal f (t) by its Fourier series f (t) = a0 + p p n=1 n we are nding the contribution of each frequency to the signal. The value p of the corresponding coecients give us that contribution. The nth term of the nt nt partial sum of the Fourier series, an cos + bn sin , is called the nth harp p monic of f . Its amplitude is given by a2 + b2 . Conversely, we can create a n n nt nt an cos signal by using the Fourier series a0 + + bn sin for a given p p n=1 value of p and playing with the value of the coecients. Audio signals describe air pressure variations captured by our ears and perceived as sounds. We will focus here on periodic audio signals also known as tones. Such signals can be represented by Fourier series. A pure tone can be written as x (t) = a cos (t + ) where a > 0 is the amplitude , > 0 is the frequency in radians/seconds and is the phase angle. An alternative way to represent the frequency is in Hertz. The frequency f in . Hertz is given by f = 2 The pitch of a pure tone is logarithmically related to the frequency. An octave is a frequency range between f and 2f for a given frequency f in Hertz. Tones separated by an octave are perceived by our ears to be very similar. In western music, an octave is divided into 12 notes equally spaced on the logarithmic scale. The ordering of notes in the octave beginning at the frequency 220 Hz are shown below Note Frequency (Hz) A 220 A# 233 B 247 C 262 C# 277 D 294 D# 311 E 330 F 349 F# 370 G 392 G# 414 A 440

A more complicated tone can be represented by a Fourier series of the form x (t) = a1 cos (t + 1 ) + a2 cos (t + 2 ) + ...

References
[NA] N. H. Asmar, Partial Dierential Equations and Boundary Value Problems, Prentice Hall, Upper Saddle River, NJ, 2000. 25

[MG] M. D. Greenberg, Advanced Engineering Mathematics, Prentice Hall, Upper Saddle River, NJ, 2nd ed., 1998. [DM] D. A. McQuarrie, Mathematical Methods for Scientists and Engineers, University Science Books, Sausalito, CA, 2003. [WS] W. Strauss, Partial Dierential Equations: An Introduction, Wiley, 2008.

26

You might also like