0% found this document useful (0 votes)
11 views23 pages

Understanding Fourier Series and Trigonometric Functions

The document discusses Fourier series and trigonometric series, which are essential in the study of periodic functions in various fields such as physics and mathematics. It outlines the definitions, properties, and theorems related to Fourier coefficients and their convergence, as well as preliminary theorems on periodic functions and piecewise monotonic functions. The document also addresses key questions regarding the convergence of Fourier series and their representation of functions.

Uploaded by

sanjhayprabu
Copyright
© © All Rights Reserved
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
0% found this document useful (0 votes)
11 views23 pages

Understanding Fourier Series and Trigonometric Functions

The document discusses Fourier series and trigonometric series, which are essential in the study of periodic functions in various fields such as physics and mathematics. It outlines the definitions, properties, and theorems related to Fourier coefficients and their convergence, as well as preliminary theorems on periodic functions and piecewise monotonic functions. The document also addresses key questions regarding the convergence of Fourier series and their representation of functions.

Uploaded by

sanjhayprabu
Copyright
© © All Rights Reserved
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

Fourier Series

1. TRIGONOMETRICAL SERIES
One of the most important and also one of the most interesting fields to which we may apply the subject
developed so far, is provided by the theory ofFourier series and more generally by that oftrigonometrical
series, which we now propose to study. Such series were first obtained in theoretical physics, in the
course of investigations on periodic motion, chiefly in acoustics, optics, electrodynamics and the theory
of heat. The credit to initiate a thorough study of certain trigonometrical series goes to Fourier, although
he did not go far enough to discover the fundamental results of the theory.
Periodic processes(motions)are described mathematically by means of periodic functions. A function
/'of one independent variable x is said to be periodic if there exists (called its period) such that
f[x + X)= f(.v), for all values ofx. The trigonometric functions, sinx, cos x are the simplest periodic
functions with period 2k.
A series of the form

(«i cosx + /?! sin x)+(oj cos2x + bi sin 2^ + ...

or more compactly

—Qq + r (a„ cos nx + sin ;?x)

is called a trigonometric series, and the constants Oq, a„, b„{n= 1,2, 3 ...) are called its coefficients. If
the series converges for all values of x (in fact, in an interval c < x < c + 2.;r), then in view of the
periodicity of the trigonometric functions, its sum is a function/defined for all values ofx and periodic
with period 2;r. To save tedious repetition, itmay be mentioned once for all that the/w7c//o;75com'We;W
in this chapter are all assumed to be periodic.
. We now proceed to obtain a relationship between the function /and the coefficients a„, b„ which
was conjectured by Euler.
The following results of integral calculus form the basis of all subsequent work. Throughout, we
use m, n to denote positive integers or zero.
0, for m ^ n
cos mx cos nx dx =■ ■ 7t. for m = n > 0
●'-K
In, for m = n = 0
Fourier Series 453

0,for n and in = n = 0
sin mx sin nx dx = ■
n,{oT m = n> 0

fK
COS im sin nx dx = 0
J-K

1.1 Fourier Series


If the numbers a^, a l> ● ● ●■> ^/,5 6,, b ... are derived from a function /by means of Euler-
Fourierformulas:

I f'f
a, f [x)cosnx d.x, « = 0, I, 2, ...
7T
...(I)
1
b..= - /(.r) sin^LV dx, n = Q, 1, 2, ...
K
then the series

1
—a. Z (a„ cosn.T + sin n.r) ...(2)
2
is called the Fourier series offor the Fourier series generated byf and the coefficients a,„ b„ defined by
(1) as the Fourier coefficients* of/
It is to be noted that the Fourier coefficients have been obtained purely on the assumption that the
function/is bounded and integrable on [-;r, 7c\
There is nothing to suggest that the Fourier series (2) is convergent. In fact the series may not
converge at all, and even if it converges, the sum may not be/ though it often is and there is some
justification for the hope that the series may converge and have/ for its sum. In case the Fourier series
of/converges uniformly, the definitions of Fourier constants suggest that its sum will be/ and that/is
capable of a unique Fourier series expansion.
A host of questions arise but there are four major problems which prick the mind and which we ^
shall try to answer in the subsequent pages.
QC
1, Is the Fourier series of a given (integrable) function/convergent for some or all values ofjc in UJ
-7t < x<n:l H.
Q.
2. If it converges, does the Fourier series of/converge to/? <
- 3. When and how can a function be expressed as a Fourier series? X
● 1. When the trigonometric series be the Fourier series?
o

* Justification for the definitions can be seen from the following:

Let the series jOq + Z(a„ cos nx + b„ sin nx) converge uniformly on [-;r, n'l to the sum function/
Multiplying by cos nx (since uniformity of convergence of the series is not thus destroyed, term-by-term integration is
justified) and integrating, we get

a, / cos/i.r dx
K ●'-'T
Other terms are vanishing in view of the results of integral calculus mentioned earlier.
I
Similarly, a0 ” fdx, b„ = - /sin«.v dx
454 Mathematical Analysis

Note: One of the answer to the last question is straightforward, since we know that, if the trigonometric series

-Oq + I {a„ cos nx + b„ sin tvc)


2 '1=1

is uniformly convergent on {-n, 7t] and if/(;r) is its sum. then it is the Fourier series offix).
Further, since

I a,I cos/u;+ b„ sinnjcj <| costw|+| sin tu| |


< a„
| +1£>„ |.

Therefore, by Weierstrass’s M-test, it follows that, ifthe series I f a. + ^ converges, then


the trigonometric series converges absolutely and uniformly on every closed interval [a, b] and that it is
the Fourier series ofa continuous 2rc-periodicfunction.

2. SOME PRELIMINARY THEOREMS


Before dealing directly with the problems raised above, let us do some preliminary theorems which will
help in the subsequent development of the subject.

2.1 Periodic Functions

We know that a periodic function of period A is such that f[x± = f(x), for all x.

Theorem I. For a periodicfunction ofperiod 2tt, prove that


f/5 r^+2n
(/) f dx = f dx,
Ja Jcr+2;r

K [●a+27:
Ui) fdx = fdx.

'T:

iiii) f(x)dx = f{y + x)dx.


●>-K

(X, P,y being any numbers whatsoever.

(/■) For a periodic function of period 2;r, we know

Hence, putting t = x+2n, for all a and f, we get


‘P+2t: ● P+2ff ‘P+2k
^ f(x)dx = Jc+2;r f(t-2n)dt =
●'ct Ja+2;r
f{x)d^
pa+2n- ●-K fK (●«+2ff
Hi) fdx = fdx + fdx + f dx
Ja

/●-;r rn ra
f dx + / ^Ix + f dx [using (/)]
●>a J-K -K

FK

fdx
J-K
Fourier Series 455

(///■) Let y + x = t

fK

J-n
f{r + x)dx =

1‘y+ff
felt + fell + fdt
Jy-n J-/r in

e~K rn y-yn
fdt + fdt + fdt [using (/)]
●^y+;r ^ -K

(●T r/T

J-;r
fdt = fdx.
J-;r

These results, in fact, mean that the integral of a periodic function over any interval whose length
is equal to its period always has the same value.

2.2 Some Definitions


Piecewise monotonic functions. A function/is called piecewise monotonic on an interval [a, h\, if
there exists a partition of [a, b\ such that the function is monotonic (that is either monotone increasing
or monotone decreasing) on each of the sub-intervals. Piecewise continuous functions are defined in the
same way.
From the definition, it follows that if the function / is piecewise monotonic and bounded on the
interval [a, i], then it can have discontinuities of the first kind {orfinite discontinuous) only. Indeed, if
x = cis a point of discontinuity of the function, then by virtue of the monotonicity off there exist the
limits

lim/(a') = /(c-), .v-»c+


lim/(a) = /(c+)

i.e., the point c is afinite discontinuity or a discontinuity of the first kind. QC


LU
A point c is a point of infinite discontinuity of a function/if
Q.
lim / (a) = ± oo
x-^c <
X
Dirichlet’s Integral. Integrals of the following two forms are called Dirichlefs integrals O
sin /?A' *0
Sin nx
J()
f- dx.
Jo
f dx.
sin A A

2.3 Some Theorems


Theorem 2. Iff is bounded and integrable on \-7T, K] and ifa„. b„ are its Fourier coefficients, then

(«« +A?) converges.


456 Mathematical Analysis

The integral
K m \2

f - 11=1
Z («„ COSW.V + sin^w) dx>0
●> -It

as its integrand is never negative, now


-,T
m

J-!t
/ - n=\
Z (fl„ cos/i.v + sin /u) dx

-i2
●K ●n m

n=)
Z {a„ cos nx + £»„ sin dx

'K
- 2Z a f cos nxdx - 2Z i»„ sin/IX dx
J-;r J-n

f~dx + + K'Id>l - lit'Lai ~ 2;rZ/j„"

rn m . X

j^dx-K
J-n n=l ' '

Since this is non-negative, we have


●n
f-dx,
It

taking limit as /» ^ «>, we obtain

1
2: {al+bl) < —
H=l ' ;tJ-K '
dx (Bessel’s inequality)

Hence, the series £ f^l + bl) is convergent, since/ and so is inlegrable on {-k, 7t\
/i=i ^ '

Corolliiry. Since Z (a; + bl^ converges, its general term (^al + 0, and so, a„ -> 0, b„ 0.
Thus the Fourier coejficienls a„ and b„ ofan integrable {and bounded) function form a null sequence,
i.e., when n —>
1 K 1
a fcostixdx 0, and b„ = — f sin/Lxr/.t —> 0.
7tJ-!t Tt -J-K

Note: If the function is piecewise continuous on the interval \~tt, 7t\ then also its Fourier coefTicients approach zero
as H ^
For. then a partition of \^K, 7t\ exists in each sub-interval of which/and so/^ is continuous and integrable, and
therefore the integral over [-K, 7t] can be expressed as a sum ofdefmite integrals of continuous functions over the sub-
intervals.

Thcorem 3. Rienumn-Lehesf’ue. If a function (f) is bounded and integrable on the interval [a, /i],
then as /i —> ,
Fourier Series 457

fb (■h
A. = Jo ^cos ;;.v dx -» 0, and =
Ja
$'f,mnxdx 0.

If a and h both belong to one and the same interval of the form ■mrc < A' < nm (w, any positive
integer), we define /(a) = ^(a) in a<x<b, and/(A) = 0 at the remaining points of /»;r]: for
other real a,/is defined so as to be periodic with period 2;r. Then
ri> en
A„ = (pco^[Link] = f cos nx dx = ;ra„,
Jo J-Jt

and. similarly =7rb„, where c3„ and denote the Fourier coefficients of/
Using Theorem 2, A„, tend to zero.
Again, if a and b do not lie in the same interval of the type [-nm, ;n/r], we can split up the interval
[ti, d] into a finite number of sub-intervals, each of which lies in an interval of the form [—nm, nm\ A..n
and can be expressed then as the sum of a (fixed) finite number of terms each of which tends to 0.
as n Thus^,,and5„tendto0.

Deduction. For a bounded integrable function (p, limits of the Dirichlet’s integrals are equal, i.e.,
*0
sin ivc sin;?A'
li m (p— dx = lim (f) dx, 0< a < K
●'() sin A JO A

1 1 \
Assigning the value 0 to — — at the origin, the function becomes continuous and bounded in
sin A A

[0. a], a < 7T. Thus the function (p — — is bounded and integrable on [0, a].
sin A

Therefore by the above theorem


*0

lim (p — sinnxdx = 0 QC
l^sinA Xj UJ
H
or 0.
ra
<
sm»A fa sin/u'
lim
●●o
0—
smA
dx = lim
JO
<p
A
dx
o
Note: Compare with Example 16.

Theorem 4. If a function f is bounded and integrable in [0, a], a > 0, and monotone w]0, S[,
Q < S <a, then

sin/JA ■~sinA
lim / dx
Jo X clx = f{0+)
Jo X

Let us first consider the case when/(0+) = 0. We may suppose, as it does not affect the result that
./■(0) = o.
458 Mathematical Analysis

Let ^ be a positive number less than a.


By the second mean value theorem there exists g [O, such that
^ sin;iA ^1 sin nx f^sinjjA-
rf.v=/(0)JO dx
Jo A .V I .Y

dx
oj X

sinr fnS ctsm A


=/(«)" Ji5 dt = f(8)JnS. dx ...(1)
1 t X

Sin A
Since dx is convergent,there exists ^:>0 such that for all a>0.
Jo X

sinA
dx < k
Jo A

sinA sinA sinA


dx = dx - dx <2k ...(2)
J n5,I A ●0 A Jo A

Also f((5)-> 0, as 5^ 0 + 0. therefore for any f > 0, there exists y>0 such that
f{S) < e/4k, when 0 < S <y

Hence from equations(1),(2)and (3), we get


rS
^^'^^dx < f(5) ● 2A: < —e, when 0 < 5 < 7
Jo A 2
Now,writing
●a smnx ^ ,sin nA , f" ^ sin nx dx
dx = f dx+ f
0 A Jo A J5 A

and observing that(by Theorem 3)the second integral on the right —> 0, as /i —>«, we see that there
exists a positive integer m such that for all n> m,
*a smnA: 1 1
dx <-e + —e = E
0 A 2 2
●a sin/u
lim / dx = 0
n_*oo JO A

SO that the theorem is proved for the case when/(0+) = 0.


in the general case, when /(0+) ^ 0, since

lim [/(-^)- /(0+)l = 0


j->0+

replacing/by [/-/(0+)], we get


Fourier Series 459
*a Sin/2A*
lim /-/(0+) (lx = 0
/l— Jo A'

*0
sinnA ●a Sin/a*
lim / dx= it-i
Iim/(0+) clx
Jo A' Jo A

- /(0+)lim rff, {/ = ;jA')


;i— J() f

‘*'smA
= /(o+) Jo dx
A

Hence, the theorem.


Deductions
smA K
i Since dx = —
Jo A 2
*a ‘^sin A
lim JO /^^A=/(0+) Jo A
dx = —
2 /(0+)

2. Using deduction Theorem 3, we get


●a
lim , sin/u , \ ■”sm.v
f|->oc Jo
—^/A=/(0+) Jo A dx = 7/(0+)

Theorems. Iff is bounded and iniegrable in [-;r,;r] and monotonic in f-<5,0[ and JO.^], {not
necessarily in the same sense), where 0 < S < /r. then

/(0-) + /(0+) f-sin.r dx


—a,y = I. a II
2 11=] n Jo A
OC
where - 0, 1, 2, denote the Fourier's coefficients of f UJ
Now. I-
Q.
1 m 1 pt: m PK <
-o,, + «=)
I a„ = fdx^-Y. f cosnxdx X
2 27T n ;i=l J-K
o
m
1 f'r
f — + L cos nx dx
n: J-?r 2 n=l

sm m + - A
1 PK
2J dx
K J-/T
2sin|A

1
sm m + - X sin m + — A
1 2J dx + 1 PK
2J dx
2n 1 . 2k Jo
sin—A sm—A
2 2
460 Mathematical Analysis

nil
1 sin{2m + l)^ dx + 1 sin(2/« + l)jr dx
2 f(-2x)
In Jo ^ ’ sin^ In Jo ^ ' sinjc

Hence letting m

1 T r~sinjc
dx
2 u=I n^ "JO X

■“sinjc n
Corolliu'N. Using dx = we get
Jo X 2

1
—a„ + Z a
/(o-)-^/(o+)
2-0 ;i=l 2

(●“sinx
Deduction. The theorem may be used to deduce the value ofthe integral dx.
Jo X

Taking / (x) = I, V x, we get

1 n
—a \ 'dx=\
2 ^ 2ni~7i
1
and a. cosnxiic = 0
n J-^
so that

2 r~sinx
/(0-)+/(0+) r-sin;c dx = -
] = dx
n Jo ^ n Jo X

●~sinx 1
dx = —n
Jo ^ 2

3. THE MAIN THEOREM


We are now in a position to take up the main problem. Our approach will be to write down the partial sum
sequence of the Fourier series of the function / and lay down a set of sufficient conditions so that
the sequence (and consequently the Fourier series) converges to f. That will answer all the first three
questions raised earlier in § 1.1, that a function /satisfying these conditions will generate a Fourier
series which converges and has a sum/
A large number of sets of sufficient conditions for the purpose have been found, of which we shall
use only one, Dirichlet’s criterion, the great generality of which renders it sufficient for most purposes.
It was given by Dirichlet and was the first set of exact conditions of convergence in the theory of Fourier
series.

Theorem 6. If a function f is bounded periodic with period 2n and integrable on [-;r, n\ and

piecewise monotonic on [-;r,;r], then


Fourier Series 461

1 -[/ -)+f{^ +)]- for -;r < ^ < 7T,


I(a„ cos + b„ sin n^)=
2 ” H=1

^[/(^-)+/ +)]- for ^ = ± 7T

where a,,. b„ are Fourier coefficients of f.

1
Let —Gq + {^,1 cos nx + b„ sin ;ij;) be the Fourier series off, and ^, a point of \-n, n\

The wth partial sum at the point


m
1
— + H=1
L' cosn^ + sin/j(^)

1 rn r>i j
cos ALT cos»(^ + sin/ixsinn^]£/;r
iTT J-nfdx+Z-j
n=i 7T /

f/l
1 1*^

2;r J-^r / 1 + 2 Z cos?i(at -(^) dx


W
1
-;r
f(x+^) 1 + 2 n=l
IcosnA' dx (§2.1)
2;r J
1
sin m +- A
I /■Z
2J dx
2nJ -7T
sm-x
2
I
sin m + - X sm w + - A
1 f'^ 2J dx +
1 i‘)t
2J dx
2k -7C 2k Jo
Sin—A sm —A
2 2 GC
I 1 UJ
sin{2m + l)t dt + — sin (2m + l)? dt H
/(-2t + ^) f(2t + ^) CL
K Jo Sint K^o smt
<
f{x) X
o

/(^)

● /(^)

A’

Discontinuity at ^

Fig.1
462 Mathematical Analysis

Proceeding to limits when m 'x>,

1 1 1
I(«„cos«^ + ^„ sin «(;) =-
2 " '1=1 n 1

^ € [-;r, K
2
Thus, the Fourier series of a (periodic) function / which is bounded, integrable and piecewise
monotonic on [-;r,;r], converges to /(^+) at a point and (using
1r
periodicity of/) to ~ + attheends, ±;r.

CorollaiT 1. At a point ^ of continuity off, the sum of the Fourier series of/

1
— Qn +
2-° I {a„ cos < +
<1=1 b,^ sin nC)= ^[/
2 (C“)+/(C +)

=//(?)+/(C)]= /K)
Thus for a function/satisfying the conditions of the theorem, the sum of the Fourier series is
actuallyf(x) a! all points x wherefis continuous', while at points of(finite) discontinuity the sum of the

series is + /(-^+)]-

Corollary 2. Since a function of bounded variation can be expressed as a difference oftwo monotone
increasing function (Jordan theorem), it may be easily shown that a periodic function of bounded
variation can be expressed as a Fourier series.
Example I, Find the Fourier series ofthe periodic function/with period 2;r, defined as follows:

0, for -7t < a: < 0,


/W = - X,for 0 < jc < ;r
What is the sum of the series at x = 0, ±7r, 4tr,

m The function is bounded, integrable and piecewise monotonic on [-;r, k\ Let us determine the
Fourier coefficients

1 0 n
1 f'' (K
flo =- fdx - — 0 ■ dx + xdx
7t ●’-'T K Jo 2

K
1 K 1 .rsin?u 1 r''
an xcos,nxdx = — s'mnxdx
K n 0 n ●'0

n 0, for n even
I cos/u
Tin n 0 ^, for « odd
Kn
Fourier Series 463

1
n —,for n even
1 COS nTt n
bn xsinuxdx - —
7t Jo n 1
—, for n odd
n
In \-7T, the points ±k are the only points of discontinuity of/ Therefore at all points of the
interval other than ±k, we have
2 I cos a: cos3jc cos5.v sin a: sin2.r sin 3a:
/w=f 7t 1^ 3^ 5-
+... +
I 2 3
At a: = 0 where/is continuous, we get
1 1 I
/(0)=--- — + —r- + — +...
'
^’ 4 ;r 11^ 3^ 5‘

1 1 7t^
1+ + +...=
3- 5^
Because of periodicity, value of the Fourier series at a: = 4;r is same as at a: = 0.
At a: = ±7C, points of discontinuity of/ the sum of the series

, 71 _K 2 1 1 1
2~4 ll‘ 3^ 5^

or 1 + —7 + —7 +... =
3^ 5^

Again because of periodicity, value at -tn is same as at a:= ±;r.

Notf; Figure 2 gives the shape ofthe function/


CC
LU
Q.
y
<
X
o
► i
-5it -An -3ji -2ji -n 0 471 5ti

Fig. 2

Example 2. Expand in a series of sines and cosines of multiple angles of a:, the periodic function /
with period 2;r is defined as

/w= |-1, for -7t < a: <0


1, forO<A-<;r
464 Mathematical Analysis

Also calculate the sum of the series at ^ = 0, —n,± n.


2

■ The function (Fig. 3)is piecewise monotonic, bounded and integrable on \~7t, 7t\ Let us compute
its Fourier coefficients

1 f'T 1 fO rn
ao=-f
71 ●'-/r n J-7C
(-l)cf.r + Jo \ -dx =0

1 rO fK
an -COS fixdx + cosnxdx =0
7t J~ji Jo

0, for n even
1 fO fn 2
bU -sinnxdx + sin nxdx = —(] - cosn;r) = 4
71 ●'-TT Jo 7tn , for n odd
;r^?

The function is continuous at all points of [-JT, ;r] except±;r.

, 4 fsin.x sin 3.x sin5x


+...
1 3 5

which holds at all points with the exception of the discontinuities, ±7t.

-x< f
71 '271 371
X
-3711 -2n I -711 0

-y

Fig. 3

At X = ±;r, the sum of the series

Ir 1

1
At j: = —7t, a point of continuity,

1 = 1.- 1 i_
7T 3 5

1 1 7t
1 1
3 5 4
Fourier Series 465

Note; Figure 3 is the graph of the given function.

Figure 4 illustrates how the partial sums S„, of the series represent more and more accurately the
function/as m —> OO ,

0
¥
-K n
4 ,
-1 5ii = —sin.t
n

yf
1
N

0 //
X

-1
A . sin 3a-
S2 = SIRA +
3

yf

f'y.

-X K
I
7 s
\
r\ o| %
f \
j_
\ ;r T
X.
V
\\ X

T -1 V

A( . sin 3a sin 5a'


5, = sin A +
3 5

Fig.4 (T
lU
3.1 Fourier Series for Even and Odd Functions a.
<
Even Functions. If/is an even function,/,e.,/(-x) =/(x), V :r, then/cos nx is an even and/sin X
is an odd function and therefore
o
1 1 fO fn
aft f cos nxdx = —
2r
fcosnxdx+ fcosnxdx f cos nxdx
7t J-K n J-n JO TtJO

1 f'f
bn fsxnnxdx = 0.
K

So, the Fourier series of an even function consists of terms of cosines only and the coefficients an
may be computed from (1).
Also, since for an even function,
/(0+)=/(0-)=/(0)
466 Mathematical Analysis

and

/(-;T4.0)= /(;r-0)

the sum of the series is/(0) at 0 or ±(even multiple of k ), and is /{k -) at ±n or ±(odd multiple
of 7t).

Odd Functions, If/is an odd function, = -/^(a:), x, then/cos/jx is an odd and/sin wa:
an even function and therefore

1 fK
an f cosnxdx = 0
K

1 {‘Tt 2
b. /sin nxdx = — /sin nxdx ...(2)
n J-/r K ♦'0
So the Fourier series of an odd function consists of sine terms only and the coefficients may be
calculated from (2). Also for an odd function
/(0-) = 0=/(0+)
and

/(-;r + 0) = -/(;r-0)

The sum of the series is 0 at x = 0 (or any multiple of k ).

Fxninple 3. Find the Fourier series generated by the periodic function 1 x | of period 2;r, Also
compute the value of series at 0, 2;r, - 3;r.

a The function is monotone and continuous on [-;r, ;r], Moreover it is an even function and therefore
the Fourier series will consist of cosine terms only.
The function may be restated as

/w= -X, for -;r < X < 0


X, for0<x<;r

1 ●K
2 f'T
a0 “
-xdx + xdx xdx = n
K ●l-K Jo ;r Jo

2
an xcosnxdx = ^(cosn;r - l)
K Jo Kn

0, for n even

for /I odd
Kn^
We, thus, obtain the series

./ s K 4 cosx cos3x cos5x


/ (x) = ^ ^ ^ +...

The series converges at all points and its sum is equal to the given function.
Fourier Series 467

Sum of the series at Ik is the same as at 0.


At 0.
;r 4 I 1 I
()=-

I
1H -\ -
3“ 5-
At -3;r (same as at tt).
7T 4 , I 1
TT — — 1 H 7- H 7 +...
2 71 3- 5'

I 7T'
1 + —+ , same as above.
3- ^+...=

3.2 Half Range Series


With the help of the Main theorem and those of even and odd functions, we now consider the expansion
of a function over the interval [O,;r] in terms of(/) sine terms only. (//) cosine terms only.
(/) The sine series. If a fimcikmfis bounded, inlegrahle and piece^\’ise inonolonic in [O. 7r\
then the sum ofthe sine series
2
l.b„ sin nx. where =— _/sin/?.vr/.v
n Jo

is equal to — f(.v -)+ /{.v +) at every point x between 0 and K. and is equal to 0, when

X =0. K.

To obtain a series consisting of only sine terms we define an odd function F in [-/r, ;r]. JE
UJ
identical with/in [O, 7r\ H
D.
Let/^=/in and F(x)=- F(-v)=-/(-v) in [-;r,0]. <
I
Evidently. F is bounded, integrabie and piecewise monotone in [-tt, tt], {i.e. satisfies the O
conditions of the main theorem). Again, Fbeing an odd function, its Fourier series consists of
sine terms only with sum
2 r F^mnxdx
sin nx, where b„ =
K Jo

2 f'f
/sin nxdx
TT Jo

Ir
equal to ~ F[x-)+ F{x+) = 2‘-'f{x-)+ f{x+) at every point-V between 0 and tt, and

equal to 0 at x = 0, Tt.
468 Mathematical Analysis

(//) The cosine series. If f is hounded, inlegrahle and piecewise monotonic in [0,;r]. then the
sum ofthe cosine series
2 rK
—a,, + 2^/„cos/j.v, where a„ = — I y cos/[Link] r/.v
^ V/ ’I ’ It K ●'0

is ecjiial to — f (,v -) + / (x +) at every point x between 0 and k , andf{^ +) at x = 0, and

/(;r-) at x-k.
To prove the result, define an even function Fon [-;r. identical with/on [O. tt]. so
that

F = / on [0, 7t] and F(x) = F(-x)= /(-x) on [~;r, O],

Proceeding as above, we get the required result.

Example 4. Find the Fourier series consisting of (/) sine terms only, (//) cosine terms only, which
represents the periodic function/(x) = x in 0 < x < k.
m (/) The sine series. The function may be extended as an odd function./(x) = x in -;r < x < ;r,
periodic with period Itt.
d = 0, for « = 0, 1, 2, 3, ...
and

2 r xsin nx dx
b.
K -^1)

2cosn.;r -2!n, for n even


n 2Jn, for n odd

Hence for all points between 0 and 7t,

sinx sin2x sin3x


x = 2
1 2 3

According to § 3.2, the sum of the series must be 0 at x = 0, k and this fact can be verified
directly as well. The representation holds atx = 0, but not at x = ;r.
(ii) The cosine series. The function may be extended as an even periodic function/(x) = | x | in
[-;r, tr] with period Itt.

b„ = 0, for « = 1,2, 3, ...


and
2
a0 xdx = K

2 f« 0 for n even
a. X cos nxdx = ■
-Alitrf. for n odd
Fourier Series 469

Hence for ali points between 0 and k.

4 1cos.v cos 3a- cos5.v


x = —K —+ +...
2 K 3- 5'

The function being continuous, the relation holds for all x.


According to § 3.2, the sum of the series must be/(0+)= 0 at a = 0, and /(;r-)= ;r at
A = ;r which can be found directly from the above relation.
AtA =0 or TC. we gel

1 7T
1+ —+ — +...=
3- 5-

4. INTERVALS OTHER THAN[~7t,


So far we have considered the interval \-7t, ;r] only. It was just a matter of convenience, otherwise any
finite interval could have been used. We now show that by effecting certain transformations, any finite
interval can be made to correspond to the interval [-;r, 7t\

4.1 The Interval fO, 2ti]

Iffis bounded, integrable and piecewise monotonic in [O, 2;r], then the sum ofthe series

1
—«,) + L (a„cos;jA +/;„sinnA'),
2 "=i

1 f2/r 1
where a, fcosnxdx, b„ = — fsinnxdxis — f{x-)+ f[x+) at every point x between 0
71 Jo kJo

and 2k , and is f{2K-)+f[0+) at x = 0,2ti and is periodic with period 2k. CC


UJ

On substituting x - y k , we find that y varies from -k to ;r as a varies from 0 to 2;r, and Q.


<
f{x)= f{y+7t:)=F{y), say X
o
Since/is bounded, integrable and piecewise monotonic in [0, 2.;r], then so is F"in [-;r, k\ Hence

by Dirichlet’s criterion (Main theorem) the sum of the series—«o+ 2 (a'cos/iy-f Z?'sin/iy),
, 1 rK 1● f'f ^ ""
where -— F cos nydy, b'„ = — Fs\nnydy,
K '●-a 7T

IS — F'(y-) + at every pointy between -k and k.

Ir
and is — F(;t-) + F(-;t4-) at y = in', and is periodic with period 2;r.
470 Mathematical Analysis

On changing the variabie, we see that


1 2k
(-1)'
o. F{x - ;r)cos»(.v - K)dx = /{.v)cos/uy/.v
K Jo
/I

b'..n =-
2k
F[x - 7t) sin/i(A'- 7r)dx -
(-0 2n
f{.x)n\nn.\dx
;r J() K Jo
Also

^[^{y-)+/"(.v+)]= ^[/(-^-)+ /(-'-+)]


and

^[F{n~)+ F{-k +)]= \lf(2K -)+ /(0+)


On making these changes, we get the required result.

4.2 Interval {-I, /], / is a Real Number


Iffis hounded, inlegrable and piecewise monotonic in [-/. /]. then the sum ofthe series

I imx nnx
-Oq + I a„ cos + b..n sin
i i
I
1 f hkx nKx 1r
where a,
/ J-i
/cos
I
dx. b„ = I J-i
/.sin /
r/.v. is— /(.v-)+/(.v+) for every X between-}

and I. and is ~ /(/-)+/[~l +) for x = ±l, and is periodic with period 21.

On making the substitution y = tcx/I , we see that y varies from -;r to ;r, as .y varies from -/ to 1.
and

/w=/(yi] = F(y). say

The function F satisfies the conditions of the Main Theorem in [~7t, and therefore proceeding
as in the previous section (§ 4.1) we may prove the required result.

Examples. The function a" is periodic with period 2/on the interval [-/. /]. Find its Fourier series.
■ The substitution y = 7txH transforms the function into a periodic function with period lit on
\~K, n\ Moreover it is an even function.
=0,(( = 1,2, 3,...

2f 2/-
«o =T x-dx =
I Jo 3

2 cl . nnx 4/-
an X'cos dx = 2 2 cos nn
l Jo / K n
Fourier Series 471

4/-
—TT' for» even
TTn'


4r- for It odd
TTir

Also, the function is continuous on [-/, /]. Therefore for all .v.

. I- 4r- cos n:x/l cos Itt.x/I cos 3;ta7/


.V- = — + — +...
3 K' 2- 3-

(-1) n;r.v
T 2: :;^COS
3 TV «=i n I I
4.3 Any Interval {a, b]

If f is bounded, inlegrahle and piecewise monotonic in [t;. h\ then the series

n;r(2A' -- a -b) nK[2.\ - a - /;)


-r/o+I a„cos + h„ sin
2 ;i=l b-a h “a

2 fi’ nK{2x - a - b)
where a, f cos dx.
b-a ^ b-a

2 h
n7i(2x — a - b)
b. /sin —^ dx
b-a J b-a

represents /(.v -)+/(x +) in < x < /?, and

QC
= a,b LU
I-
Q.
and is periodic with period {b - a).
<
;r(2.v - a-b) X
On making the substitution }’ = we see that v varies from -k to k. as .r varies O
b-a
from a to b.

Considering the function F, where

y{b - a)+ 7t{a + b)'


= F(y)
In

and proceeding as above, we get the required result.

Note: The transformation is obtained by determining the two constants/and in v=/A'+ /«such that y = -;r when
x = a. and y = 7t when x = b.
472 Mathematical Analysis

4.4 The Interval [0,/]


If a function/satisfies the conditions of the main theorem (Dirichlet’s criterion) on [0. /]. the substitution
V = kxH determines a function F. where

/(-v)= /(y//;r)= F(y)


which satisfies the conditions of the main theorem on [O, n\ The function F may now be extended [as

in § 3.2] as an odd or an even function on [-;r, t:\ so as to give series consisting of terms of sines or
of cosines only.
We may, if we desire, reverse the above process—that first extend it as an odd or as an even
function on [-/. /] and then use the substitution y = kxH to transform it to a function F on [-;r, 7t\

Alternatively, like § 4.3 transformation y = k[2x -1)11 may be used to get a function Fsatisfying
the conditions of the main theorem on [-;r, tt] where

/w=/ ly + Ttl
2k
= f0’)

Example 6. Expand the periodic function x',0 < x < I, of period /. in a series of (/) sines only.
{ii) cosines only, {Hi) sines and cosines, of multiples of x.
Also find the sum of the series at a: = 0. /.
(/) The function may be extended as an odd function in [-/, /], by redefining it as

-x^, for -/ < jc < 0


/W = .V". for 0 < A- < /.

Substitution of y = kxH transforms it into an odd periodic function on 7t], so that the
Fourier coefficients are

a„= 0 for« = 0, 1,2, 3,..

1 0
b=-
t\ x~ s\n[rmx/l)dx + Jo x~ sin{n7tx/l)dx
I J~i

— A" sin(nn:A7/)(iv

I
2 / 1 imx rl IlKX
- A" cos + 2a cos dx
I UK I 0 Jo I

21- 4/^
COS UK +
UK 3_3 (cosUK ~\)
U'K
Fourier Series 473

21-
. for n even
lITt
11- 8/-
for n odd
nn u-Jt

The function is continuous at all points of[-/. /] except ±1. Therefore, the Fourier series
for all points of [0. /] except /. is

>mx
X- + I sin
tm 3_? /
H=l n K

At X = 0. a point of continuity of the function, the sum of the series is zero, a fact which
may be verified directly from the series.

At X = /. the sum of the series =

which is true on actual verification of the sum of the series.


Hence the relation (1) holds for all points of[0, /].
(//) Let us now extend the function as an even function on [-/, /], by redefining it as

/(.v)=.v^ I <x< I

Substitution of y = Ttx/l transforms/as an even function on [-/r, so that the Fourier


coefficients are
b„ = QJovn= 1,2, 3. ...

1 {I 2 r' 21-
a0 “ x-dx =- x^ dx = QC
/ J-t [ Jo 3 UJ
I-
Q.
2 c< . imx 4/^ <
o. X- cos dx = -cos UK =
I Jo I K~n-
7 0
I
o
The function is continuous at all points of[-/, /]. Therefore for all x in [0, /], we have
n
n^x
cos
3 n~ }

Atjc = 0,

0= —
3 n- 2} ^ 3-

1 n-
2^ 3^ 4- ■■■ 12
474 Mathematical Analysis

A[x = l.

— cos;i;r
3 7T "=> It

I- 4/'
= — + —tI-
3 7T~ n

IT
1+ + T+--
2- 3-^ 6

{iii\ Let US apply § 4.3.

If we make the substitution y = k{2x - /)//, theny varies from -n to k asx varies from
0 lo /. The Fourier coefficients are

2 2
a0 “ .v-r/.v = -/-
/ Jo 3

2 r<
an ’cos nn[2.x - /)dx =
-f'.x~ . 7/;r(2.v - /)d.x
-.vsin
/Jo I im Jo I

I
I
- .vcos
im{2.x -1) r- (-1)'
K~n~ I n K~
0

2 r' 7/;t(2a--/)
b.. =- -v"sin— d.x
I Jo I

n-l
/-
cosim +
2 C‘
-VCOS /J7T(2.V - /)d.x = r- (-')
nn im Jo I n n

Hence for all points between 0 and 1.


/i-i

A
77;t(2a-/) /- “
— ^- + — I (-■) sin nTT (2.x - /)
3 7T ;i=l n / ;r «=i 11 /

Atx = 0. /. the sum of the series = —il~ .


2^

— 2: —
2 3 w

\r + —rI + —+
I —1 + ...= ;r-
2-2 3-
.2 42 5

You might also like