This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

**Rui Niu May 12, 2006
**

Abstract Fourier series are of great importance in both theoretical and ap plied mathematics. For orthonormal families of complexvalued functions {φn }, Fourier Series are sums of the φn that can approximate periodic, complexvalued functions with arbitrary precision. This paper will focus on the Fourier Series of the complex exponentials. Of the many possi ble methods of estimating complexvalued functions, Fourier series are especially attractive because uniform convergence of the Fourier series (as more terms are added) is guaranteed for continuous, bounded functions. Furthermore, the Fourier coeﬃcients are designed to minimize the square of the error from the actual function. Finally, complex exponentials are relatively simple to deal with and ubiquitous in physical phenomena. This paper ﬁrst deﬁnes generalized Fourier series, with an emphasis on the se ries with complex exponentials. Then, important properties of Fourier series are described and proved, and their relevance is explained. A com plete example is then given, and the paper concludes by brieﬂy mentioning some of the applications of Fourier series and the generalization of Fourier series, Fourier transforms.

1

Introduction and Background Information

In the mideighteenth century, physical problems such as the conduction pat terns of heat and the study of vibrations and oscillations led to the study of Fourier series. Of central interest was the problem of how arbitrary realvalued functions could be represented by sums of simpler functions. As we shall see later, a Fourier series is an inﬁnite sum of trigonometric functions that can be used to model realvalued, periodic functions. We shall begin by giving a brief description of the trigonometric polynomials, and especially of their relation to the complex exponentials. Let us deﬁne: cos(x) = sin(x) = 1 ix [e + e−ix ], 2 1 ix [e − e−ix ]. 2i (1) (2)

1

. both f1 (x) = einx and f2 (x) = ein �π inx both have periods of 2π. since f2 (x) = f2 (x + 2π). we know that −π e(inx) dx = ein = � p+2π f1 (x) = 0 if n = 1.}. −π ei(n−m)x dx = 0. and if f2 (x). (4) inx Now consider the function ein . Thus.Another way to express this is that cos(x) is the real part of eix and that sin(x) is the imaginary part. For complex numbers {a0 . where m is any integer. Clearly. In this case. for n = 1. b2 . and real x. 3 . 2. Then from the above identity. for all | π −|N | ≤ n ≤ |N |. a1 . b3 . � 2 . First. . deﬁne cn = 2 (an − ibn ). deﬁne c0 = a0 . Also. Now let us multiply equation (4) above by e−imx . (3) f (x) = inx cn einx . in this case. . c−n = 2 (an + ibn ). ei(x+2π) eix e2iπ 2 eix e(iπ) eix (−1)2 eix = = = = We now move to the deﬁnition of a trigonometric polynomial. 3 . We obtain the following: N � n=−N e−imx f (x) = e−imx N � n=−N cn einx = [(cn einx )(e−imx )] (cn ei(n−m)x ) = N � n=−N Consider two cases.} and {b1 . suppose that |m� > |N |. To do so. 2. Furthermore. Then. we deﬁne a trigonometric polynomial to be a ﬁnite sum of the form: f (x) = a0 + N � n=0 1 1 Now. Let us quickly show that both functions are periodic with period 2π. we can also write a trigonometric polynomial in equivalent form by: N � n=−N (an cos(nx) + bn sin(nx)). ... p f1 (x) = 2π. . a2 . we just need to show that ei(x+2π) = eix . .. . n − m = 0. and the periodicity of sin(x) and cos(x) follow by deﬁnition. p � p+2π n = 0.

there is exactly one n such that �π n ∈ Z.} be a series of complex functions. −N ≤ n ≤ N . In this case. �π � π −imx e f (x)dx = −π cn ei(m−n)x dx. φ3 . φ2 . 2 General Fourier Series Before focusing on Fourier series with trigonometric functions. for all integers m �= n. cm �π −π e−imx f (x)dx. We start with the notion of orthogonal systems of functions.. �b a ¯ φm (x)φm (x)dx = 1. Let {φ1 . Since −π einx dx = 2π if and only if n = 0. � a b ¯ φm (x)φn (x)dx = 0. we shall give a description of general Fourier functions.b] if. and 1 2π cn = Since m = n. we shall deﬁne the Fourier series of f(x) as the trigonometric series with coeﬃcients of the form given in equation (5). suppose that |m| ≤ |N |. for a given function f (x). �π −π e−imx f (x)dx = 2πcn . 3 . −π where m = n. (8) As a further note. 1 = 2π � π e−imx f (x)dx (5) −π We now deﬁne a trigonometric series to be of the form ∞ � n=−∞ cn einx .Now.. Thus. (6) where the N th partial sum is N � n=−N cn einx . (7) Furthermore. if for all integers m > 0. We say that {φn } is an orthogonal system of functions on [a. and that is n = m.

φ4 (x)} = {eix . x) = Then.. � π inx −imx e e = 0. e(−2ix) } ¯ and let φ(x) = einx .. x)| dx ≤ a 2 � b |f − tn (f. (9) ¯ . We have already seen that the functions einx . 3. π]. Suppose that we have two sets of complex numbers. (11) which is the N th partial sum of the Fourier series for f . 1. (12) |f − sn (f. deﬁne the Fourier series of f with respect to {φn (x)} to be ∞ � n=1 cn φn (x) (10) To see how our deﬁnition for the Fourier series with respect to trigonometric functions matches this pattern. 3 Some Properties of Fourier series We now present a few important properties of Fourier series from Walter Rudin’s Principles of Mathematical Analysis. cn and dn . as deﬁned in equation (9). e2ix . φ2 (x). 2. Theorem 8.. consider two series of functions. φ(x) = e(−inx) .. π]. In terms of generalized Fourier series. � a b N � n=1 dn {φn }. and tN (f. We now deﬁne the Fourier coeﬃcients with respect to −π {φn (x)} as follows: � cn = a b ¯ f (x)φn (x). n = 1. n = 0.11 in Rudin: Suppose that {φn } is an orthonormal system of functions on the interval [−π. since einx = e−inx . x)|2 dx. Now. let {φn (x)} = {φ1 (x). φ3 (x). and cn are the Fourier Coeﬃcients for {φn }.we say that {φn } is an orthonormal system of functions. form an or ¯ thogonal system of functions on [−π. e(−ix) . 3. 2. x) = N � n=1 cn {φn }. (13) 4 . where φn (x) is the complex conjugate of the complexvalued function φ(x). and for m �= n. sN (f.

which we may rewrite as: � a b |tn |2 = n � 1 � dm φm a n b� 1 ¯ ¯ dk φk (15) Since {φn } is an orthonormal system of functions on [a. Consider: �b a ¯ f tn = � b �n ¯ ¯ f 1 dn φn a by the deﬁnition of tn . ﬁrst rewrite it as: 5 . We Now. Since ¯ |tn |2 = tn tn and 1 dm φm �n ¯ ¯ ¯ tn = 1 dm φm . which we can again rewrite as: � a b |tn |2 = n � 1 |dm |2 �b a (16) |f − tn |2 . consider the total squared error between f and tn . we can once again rewrite the above expression as: � a b ¯ f tn = n � 1 ¯ cn dn (14) Now. according to equation (8). Proof: Let {φn (x)} be orthonormal on the interval [a. for some periodic function f and some or thonormal system of functions {φn }.This theorem indicates that. b]. we can rewrite the above as: �b a |tn |2 = �n 1 ¯ dm dm . We can also write the above as: � b �n �n � b ¯ ¯ ¯ ¯ f 1 dn φn = 1 a f dn φn a Since �b a ¯ f φn = cn . consider the integral from a to b of |tn |2 . the Fourier series provides the least total squarederror approximation. tn = �n we have: �b a |tn |2 = � b �n a 1 dm φm �n ¯ ¯ 1 dk φk . b].

. . �n n ries 1 (cm ) converges absolutely (in other words. . we know that ¯ ¯ (f − tn ) = f − tn .�b a |f − tn |2 = �b a (f − tn )(f − tn ) Furthermore.11. Theorem 8. we can see that the total error squared is mini mized when dm = cm . so that � a b |f − tn |2 � = = a b ¯ ¯ (f − tn )(f − tn ) ¯ ¯ ¯ ¯ (f f − f tn − f tn + tn tn ) a � b � = a b ¯ ¯ (f 2 − f tn − f tn + t2 ) n By equations (14) and (16) above. substitute cn for dn . for m = 1. 2. Proof: In equation (16) above. We obtain the following: � a b |sn |2 (x)dx = n � 1 |cm |2 (19) The above step will not be necessary. we can write: � a b |f − tn | = a 2 � b (f 2 ) − n � 1 ¯ (cm dm ) − n � 1 (¯m dm ) + c n � ¯ (dm dm ) 1 (17) Since |dm − cm |2 ¯ = (dm − cm )(dm − cm ) ¯ 2 2 ¯ = |dm | + |cm | − (cm dm ) − (¯m dm ) c We can rewrite the above equation as: � a b |f − tn | = a 2 � b (f ) − 2 n � 1 |cm | + 2 n � 1 |dm − cm |2 (18) From this above equation. 6 . 3 . Consider the sequence of terms {cn } = c�c2 . the series 1 |cm | converges).12 in Rudin: Assume all the notation used in the description of Theorem 8. c3 .. The se 1. . but it is interesting to point out that the integral of the absolute value of any nt h Fourier trigonometric polynomial will be less than the integral of the absolute value of the function f ..

At all points a with the property above. Thus. it is self adjoint — for every trigonometric polynomial f (x). there is a trigono ¯ metric polynomial f (x). then |f (x + t) − f (x)| ≤ M |t|. Since we know that that n � 1 �b a |f − tn |2 ≥ 0. x) evaluated at x converges to f (x) (evaluated at x) as n approaches inﬁnity. An interesting consequence of this result is that for some function f (x) that is uniformly continuous on some segment (a. b). the Fourier series will converge to the function f (x) in value for all x in that segment. the series of Fourier polynomials converges pointwise to f (a) at a.15 in Rudin: This is another theorem of convergence. there is a δ > 0 and some ﬁnite. which is closed and bounded in R2 . Then. It clearly separates all points. this also implies that limn→∞ cn = 0 This result is fairly important because it shows us that it is the ﬁrst terms of a Fourier series that are most important. we see that: ∞ � 1 |cn |2 ≤ � a b |f (x)|2 (21) From our study of convergent series. Then. periodic. there is some trigonometric polynomial such that |P (x) − f (x)| < � for all x. Another way of stating this is that given any continuous. A stronger result that describes the uniform convergence of the Fourier series fol lows. there is some sequence of trigonometric polynomials that converges uniformly to f . This is basically a direct application of the Stone Weierstrauss Theorem for complexvalued functions. Theorem 8. Theorem 8. This theorem talks about the “pointwise convergence” of a Fourier series.Now consider equation (18). Suppose that f (x) is continuous with period 2π. the value of the inﬁnite Fourier series sn (f. and therefore compact. and vanishes at no points. real M such that if −δ < t < δ. The family of trigonometric functions is an algebra of functions. The metric space K of the complex exponentials is the unit circle. 7 . In terms of simulations. complex valued function f . although it does not mention Fourier series explicitly. the family of trigonometric polynomials is dense in K. Furthermore. for any � > 0.14 in Rudin: For a periodic function f (x). since it is closed under addition. and scalar multipli cation. it follows |cm | ≤ a 2 � b |f (x)|2 dx (20) If we let n go to inﬁnity. and that the Fourier coeﬃcients become arbitrarily small. this implies that a few terms may provide a very good model of a function. suppose that for some x. multiplication.

π] is x. But we ﬁrst need to calculate c0 . which is a continuous function with period 2π. � π 1 xe0 2π −π 1 1 2 = (π − π 2 ) 2π 2 = 0 = c0 8 .4 An Example We now apply the Fourier series to a few basic examples. since the method of integrating e0 is diﬀerent. We know that the Fourier coeﬃcients cn are given by: �π cn = 21 −π f (x)e−inx dx π Since �π −π 1 xe−inx dx = ( −in xe−inx + 1 −inx π )|−π n2 e we know that 1 e−inx (inx + 1)|π −π 2π n2 −inπ 1 e einπ (inπ + 1) − 2 (−inπ + 1) 2 2π �n n � 1 (cos(nπ) − isin(nπ))(inπ + 1) − (cos(nπ) + isin(nπ))(−inπ + 1) 2π n2 � � 1 inπcos(nπ) − isin(nπ) + inπcos(nπ) − isin(nπ) 2π n2 1 2(ncos(nπ)π − sin(nπ))i 2π n2 cn = = = = = and c−n = = = 1 einx (−inx + 1)|π −π 2π n2 inπ 1 e e−inπ (−inπ + 1) − (inπ + 1) 2π n2 n2 1 2(−ncos(nπ)π + sin(nπ))i 2π n2 This almost gives us all the information we need to determine the Fourier series. So. Let us ﬁrst consider the Fourier series for f1 (x). and whose value on the interval [−π.

For example. . 9 . x) = 0 + 2sin(x) − sin(2x) + 3 sin(3x) − 1 sin(4x) 2 Graphing this function. c−4 = − 4 . let’s take the ﬁrst 9 terms (or n = 0. and the 4th partial sum of the Fourier series of f1 (x) is given by: i i i sN (f1 . c3 = − 3 . 1. i i i i We have: c0 = 0. 4). x) = 0 − ieix + ie−ix + ( e2ix ) − e−2ix − e3ix + . i i c4 = 4 . c−3 = 3 . c2 = 2 . c1 = −i. c−1 = i. 3.Now we can construct the Fourier series for f1 (x). and rewrite each complex exponential in terms of its sine and cosine terms. . we obtain the formula 2 sN (f1 . Doing so. we can indeed see the shape of f1 (x) beginning to appear. c−2 = − 2 . 2 2 3 We can now combine terms raised to the negative exponents of each other. 2.

−π Indeed. we ﬁnd that: �π (x − sN (f1 . we can replace it with some real constant. c2 . π]. 2. n = 1. since the term associated with c0 in the Fourier series is c0 e0 . First.. and the square error from −π to π decreases dramatically. Let’s call this constant a0 = c0 .Figure 1: Fourier Series with First 5 Terms 4 3 2 1 0 −1 −2 −3 −4 −10 −8 −6 −4 −2 0 2 4 6 8 10 Figure 2: Fourier Series with First 15 Terms 4 3 2 1 0 −1 −2 −3 −4 −10 −8 −6 −4 −2 0 2 4 6 8 10 Figure 3: Fourier Series with First 30 Terms 4 3 2 1 0 −1 −2 −3 −4 −10 −8 −6 −4 −2 0 2 4 6 8 10 Furthermore.. We will now use some properties of complex exponentials to create a nicer form for the Fourier series. Sometimes. 3.. recall the deﬁnitions that 10 . x))2 ≈ 2.. if we use the ﬁrst 41 terms of the Fourier series (the 20th partial sum).} is not very convenient. c1 ..781. if we calculate the total squared error on the interval [−π. leaving the coeﬃcients in the forms {c0 . For cn and c−n . the function appears much more similar to f1 (x).

This property makes Fourier series very useful in many applications. Speciﬁcally. Consider the very common diﬀerential equation given by: x�� (t) + ax� (t) + b = f (t) (23) This equation describes the motion of a damped harmonic oscillator that is driven by some function f (t). then the solution is also a sinusoid which is not very diﬃcult to ﬁnd. realvalued func tions. we can say that an is twice the real part of cn (or c−n ). and inductor. or a string vibrated at some frequency. 5 Applications of Fourier series To recapitulate. Fourier series simplify the analysis of periodic. These relations let us put the Fourier series into an alternate form. There are two parts to the solution of equation (25). We now give a few. resistor. 1 (an − ibn ) 2 1 = (an + ibn ) 2 an = cn + c−n bn = i(cn − c−n ) Since we know that c−n is the complex conjugate of cn . The problem is that the driver is generally not a simple sinusoid. If f (t) is a sinusoid. In electronics. for example. suppose the driving force 11 . but some other periodic function. a common driving voltage function is the square wave s(t). it can break up a periodic function into an inﬁnite series of sine and cosine waves. an analog circuit with a capacitor.cn = c−n Then. This is what we will concern ourselves with. and bn is negative two times the real coeﬃcient of the imaginary part of cn (or c−n ). a periodic function (whose period we shall say is 2π) such that s(t) = 0 for −π ≤ t < 0 and s(t) = 1 for 0 ≤ t < π. what remains is the steadystate solution. such as a driven mass on a spring. It can be used to model an extensive variety of physical phenomena. given by: f (x) = a0 + N � n=0 (an cos(nx) + bn sin(nx)) (22) where an and bn are given as above. The ﬁrst part is a transient that fades away (generally) fairly quickly. When the transient is gone. The physical property of oscillating systems that makes Fourier Analysis useful is the property of superposition in other words.

Principles of Mathematical Analysis. Fourier series and Transforms have become an integral part of the toolboxes of mathematicians and scientists. Because this type of analysis is very computationintensive. “Fourier Series. and it is a simple matter to ﬁnd the steady state solution to a sinusoidally driven oscillator. optics. The basic equations of the Fourier series led to the development of the Fourier transform. signal processing in communications and astronomy.f1 (t).wolfram. along with some initial conditions. The result would be an inﬁnite sum of sin and cos terms of the form in equation (2). and cryptography. Then the driving force f3 (t) = f1 (t) + f2 (t) produces the steadystate response x3 (t) = x1 (t) + x2 (t). Then. With these new techniques.com/FourierSeries. diﬀerent FastFourier Transform algorithms have been devised. Today. We could then decompose the square wave into sinusoidal components as follows: � π 1 s(x)e−inx 2π −π � π 1 e−inx 2π 0 i (einπ − 1) 2nπ −i −inπ (e − 1) 2nπ cn = = = c−n = and then just combine the cn and c−n terms as before. [2] Eric W.” From Mathworld–a Wolfram Web Re source. where f (t) is the square wave function. we can ﬁnd the response to the arbitrary driving function � f (x) = a0 + (an cos(nx) + bn sin(nx)). So suppose we had our square wave equation. The steadystate response of the system to the square wave would then just be the sums of the steadystate responses to the sinusoidal components of the square wave. Third Edition McGraw Hill International Editions (1976). produces some steady state solution x1 (t). and that another driving force. since we can represent any period driving function as a Fourier series. f2 (t) produces the steady state solution x2 (t). http://mathworld.html 12 . which lower the order of growth of the number of operations from order(N 2 ) to order(n log(n)). Weisstein. it is used for applications as diverse as ﬁle compression (such as the JPEG image format). which can decompose a nonperiodic function much like the Fourier series decomposes a periodic function. References [1] Walter Rudin. acoustics.

Weisstein.com/FourierTransforms.[3] Eric W.” From Mathworld–a Wolfram Web Resource.wolfram. http://mathworld.html 13 . “Fourier Transforms.

- BUKU-PEDOMAN-AKADEMIK-FMIPA-Fisika.pdf
- digital20181486-S29191-Amsori(1).pdf
- 111-313-1-PB.pdf
- 23-1.pdf
- 13._radioaktivitas.pdf
- 5-interaksi-radiasi-dengan-materi(1).pdf
- HALAMAN DALAM SKRIPSI
- TYPO PJJ
- Kavling Batu
- polarisasi
- Mitsubishi Master_No Sub (INA)
- Deteksi Kanker Servik
- Cobalt60
- KANKER-PAYUDARA
- Cobalt 60
- Tokow theme
- Core Mobile Qm67 Hm65 Platform Brief
- good Cloud_ Radioaktivitas _Sinar Gamma_ dan Spektroskopi Sinar Gamma.pdf
- good Cloud_ Peluruhan Beta.pdf
- good Cloud_ Peluruhan alfa, Sistematika Peluruhan Alfa, dan Emisi Partikel Alfa.pdf
- good Cloud_ anekdot para fisikawan.pdf
- Sejarah Sinar Alpha .docx
- CDM 2 08 28 Rele
- CDM 2 08 28 Release Info.rtf
- Pedoman Penyusunan Proposal Usaha1

- Ilc
- Lecture8, Eigenvalues and Eigenvectors
- Ch6QM09
- Glossary of Notations
- Curvilinear Coordinates
- Fourier Series and Transforms
- Laplace Transform
- 3295231 Fourier Series Introduction
- IAS-Mains Mathematics Syllabus
- Vector,Grad,Div,Curl
- Syllabus
- chap8.pdf
- Mathematics Syllabus for Main Examination
- Quaternion Kinematic and Dynamic Differential Equations-Is7
- Sph205 Notes
- Mathematics Syllabus
- Chapter2-Mathematics of Quantum Mechanics
- J A G Vickers- Generalised cosmic strings
- Transformasi Laplace
- term
- Pier Francesco Roggero, Michele Nardelli, Francesco Di Noto - PROOF THAT THE FERMAT PRIME NUMBERS ARE ONLY ‘THE FIRST FIVE’ AND ALL THE OTHER NUMBERS ARE COMPOSITE.
- [(B) Chung] Spectral Graph Theory
- Math for Eng and Scientists 3
- 3. Lag Rang Ian and Hamilton Ian Formulations
- Vector Spaces
- FEM Primer
- Complex Numbers and Phasors
- 1009.2198v3
- (7)Partial Differential Equations

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd