Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
FOURIER SERIES AND INTEGRALS
4.1 FOURIER SERIES FOR PERIODIC FUNCTIONS
This section explains three Fourier series: sines, cosines, and exponentials e
ikx
.
Square waves (1 or 0 or −1) are great examples, with delta functions in the derivative.
We look at a spike, a step function, and a ramp—and smoother functions too.
Start with sin x. It has period 2π since sin(x + 2π) = sin x. It is an odd function
since sin(−x) = −sin x, and it vanishes at x = 0 and x = π. Every function sin nx
has those three properties, and Fourier looked at inﬁnite combinations of the sines:
Fourier sine series S(x) = b
1
sin x + b
2
sin 2x + b
3
sin 3x +· · · =
∞
n=1
b
n
sin nx (1)
If the numbers b
1
, b
2
, . . . drop oﬀ quickly enough (we are foreshadowing the im
portance of the decay rate) then the sum S(x) will inherit all three properties:
Periodic S(x + 2π) = S(x) Odd S(−x) = −S(x) S(0) = S(π) = 0
200 years ago, Fourier startled the mathematicians in France by suggesting that any
function S(x) with those properties could be expressed as an inﬁnite series of sines.
This idea started an enormous development of Fourier series. Our ﬁrst step is to
compute from S(x) the number b
k
that multiplies sin kx.
Suppose S(x) =
b
n
sin nx. Multiply both sides by sin kx. Integrate from 0 to π:
_
π
0
S(x) sin kxdx =
_
π
0
b
1
sin xsin kx dx +· · · +
_
π
0
b
k
sin kx sin kxdx +· · · (2)
On the right side, all integrals are zero except the highlighted one with n = k.
This property of “orthogonality” will dominate the whole chapter. The sines make
90
◦
angles in function space, when their inner products are integrals from 0 to π:
Orthogonality
_
π
0
sin nx sin kxdx = 0 if n = k . (3)
317
318 Chapter 4 Fourier Series and Integrals
Zero comes quickly if we integrate
_
cos mxdx =
_
sinmx
m
¸
π
0
= 0 −0. So we use this:
Product of sines sin nx sin kx =
1
2
cos(n −k)x −
1
2
cos(n + k)x. (4)
Integrating cos mx with m = n −k and m = n +k proves orthogonality of the sines.
The exception is when n = k. Then we are integrating (sin kx)
2
=
1
2
−
1
2
cos 2kx:
_
π
0
sin kx sin kxdx =
_
π
0
1
2
dx −
_
π
0
1
2
cos 2kxdx =
π
2
. (5)
The highlighted term in equation (2) is b
k
π/2. Multiply both sides of (2) by 2/π:
Sine coeﬃcients
S(−x) = −S(x)
b
k
=
2
π
_
π
0
S(x) sin kxdx =
1
π
_
π
−π
S(x) sin kxdx. (6)
Notice that S(x) sin kx is even (equal integrals from −π to 0 and from 0 to π).
I will go immediately to the most important example of a Fourier sine series. S(x)
is an odd square wave with SW(x) = 1 for 0 < x < π. It is drawn in Figure 4.1 as
an odd function (with period 2π) that vanishes at x = 0 and x = π.
E
x
SW(x) = 1
−π 0
π
2π
Figure 4.1: The odd square wave with SW(x + 2π) = SW(x) = {1 or 0 or −1}.
Example 1 Find the Fourier sine coeﬃcients b
k
of the square wave SW(x).
Solution For k = 1, 2, . . . use the ﬁrst formula (6) with S(x) = 1 between 0 and π:
b
k
=
2
π
_
π
0
sin kxdx =
2
π
_
−cos kx
k
_
π
0
=
2
π
_
2
1
,
0
2
,
2
3
,
0
4
,
2
5
,
0
6
, . . .
_
(7)
The evennumbered coeﬃcients b
2k
are all zero because cos 2kπ = cos 0 = 1. The
oddnumbered coeﬃcients b
k
= 4/πk decrease at the rate 1/k. We will see that same
1/k decay rate for all functions formed from smooth pieces and jumps.
Put those coeﬃcients 4/πk and zero into the Fourier sine series for SW(x):
Square wave SW(x) =
4
π
_
sin x
1
+
sin 3x
3
+
sin 5x
5
+
sin 7x
7
+· · ·
_
(8)
Figure 4.2 graphs this sum after one term, then two terms, and then ﬁve terms. You
can see the allimportant Gibbs phenomenon appearing as these “partial sums”
4.1 Fourier Series for Periodic Functions 319
include more terms. Away from the jumps, we safely approach SW(x) = 1 or −1.
At x = π/2, the series gives a beautiful alternating formula for the number π:
1 =
4
π
_
1
1
−
1
3
+
1
5
−
1
7
+· · ·
_
so that π = 4
_
1
1
−
1
3
+
1
5
−
1
7
+· · ·
_
. (9)
The Gibbs phenomenon is the overshoot that moves closer and closer to the jumps.
Its height approaches 1.18 . . . and it does not decrease with more terms of the series!
Overshoot is the one greatest obstacle to calculation of all discontinuous functions
(like shock waves in ﬂuid ﬂow). We try hard to avoid Gibbs but sometimes we can’t.
x x
−π
π
Dashed
4
π
sin x
1
Solid curve
4
π
_
sin x
1
+
sin 3x
3
_
5 terms:
4
π
_
sin x
1
+· · · +
sin 9x
9
_
overshoot−→
SW = 1
π
2
Figure 4.2: Gibbs phenomenon: Partial sums
N
1
b
n
sin nx overshoot near jumps.
Fourier Coeﬃcients are Best
Let me look again at the ﬁrst term b
1
sin x = (4/π) sinx. This is the closest possible
approximation to the square wave SW, by any multiple of sin x (closest in the least
squares sense). To see this optimal property of the Fourier coeﬃcients, minimize the
error over all b
1
:
The error is
_
π
0
(SW−b
1
sin x)
2
dx The b
1
derivative is −2
_
π
0
(SW−b
1
sin x) sin xdx.
The integral of sin
2
x is π/2. So the derivative is zero when b
1
= (2/π)
_
π
0
S(x) sin xdx.
This is exactly equation (6) for the Fourier coeﬃcient.
Each b
k
sin kx is as close as possible to SW. We can ﬁnd the coeﬃcients b
k
one
at a time, because the sines are orthogonal. The square wave has b
2
= 0 because all
other multiples of sin 2x increase the error. Term by term, we are “projecting the
function onto each axis sin kx.”
Fourier Cosine Series
The cosine series applies to even functions with C(−x) = C(x):
Cosine series C(x) = a
0
+ a
1
cos x + a
2
cos 2x +· · · = a
0
+
∞
n=1
a
n
cos nx. (10)
320 Chapter 4 Fourier Series and Integrals
Every cosine has period 2π. Figure 4.3 shows two even functions, the repeating
ramp RR(x) and the updown train UD(x) of delta functions. That sawtooth
ramp RR is the integral of the square wave. The delta functions in UD give the
derivative of the square wave. (For sines, the integral and derivative are cosines.)
RR and UD will be valuable examples, one smoother than SW, one less smooth.
First we ﬁnd formulas for the cosine coeﬃcients a
0
and a
k
. The constant term a
0
is the average value of the function C(x):
a
0
= Average a
0
=
1
π
_
π
0
C(x) dx =
1
2π
_
π
−π
C(x) dx. (11)
I just integrated every term in the cosine series (10) from 0 to π. On the right side,
the integral of a
0
is a
0
π (divide both sides by π). All other integrals are zero:
_
π
0
cos nxdx =
_
sin nx
n
_
π
0
= 0 −0 = 0. (12)
In words, the constant function 1 is orthogonal to cos nx over the interval [0, π].
The other cosine coeﬃcients a
k
come from the orthogonality of cosines. As with
sines, we multiply both sides of (10) by cos kx and integrate from 0 to π:
_
π
0
C(x) cos kxdx =
_
π
0
a
0
cos kxdx+
_
π
0
a
1
cos xcos kxdx+··+
_
π
0
a
k
(cos kx)
2
dx+··
You know what is coming. On the right side, only the highlighted term can be
nonzero. Problem 4.1.1 proves this by an identity for cos nx cos kx—now (4) has a
plus sign. The bold nonzero term is a
k
π/2 and we multiply both sides by 2/π:
Cosine coeﬃcients
C(−x) = C(x)
a
k
=
2
π
_
π
0
C(x) cos kxdx =
1
π
_
π
−π
C(x) cos kxdx. (13)
Again the integral over a full period from −π to π (also 0 to 2π) is just doubled.
E
x
−π 0
π
2π
RR(x)=x
Repeating Ramp RR(x)
Integral of Square Wave
E
c
T
c
T
x
−π 0
π
2π
−2δ(x + π)
2δ(x)
−2δ(x −π)
2δ(x −2π)
Updown UD(x)
Figure 4.3: The repeating ramp RR and the updown UD (periodic spikes) are even.
The derivative of RR is the odd square wave SW. The derivative of SW is UD.
4.1 Fourier Series for Periodic Functions 321
Example 2 Find the cosine coeﬃcients of the ramp RR(x) and the updown UD(x).
Solution The simplest way is to start with the sine series for the square wave:
SW(x) =
4
π
_
sin x
1
+
sin 3x
3
+
sin 5x
5
+
sin 7x
7
+· · ·
_
.
Take the derivative of every term to produce cosines in the updown delta function:
Updown series UD(x) =
4
π
[cos x + cos 3x + cos 5x + cos 7x +· · · ] . (14)
Those coeﬃcients don’t decay at all. The terms in the series don’t approach zero, so
oﬃcially the series cannot converge. Nevertheless it is somehow correct and important.
Unoﬃcially this sum of cosines has all 1’s at x = 0 and all −1’s at x = π. Then +∞
and −∞ are consistent with 2δ(x) and −2δ(x − π). The true way to recognize δ(x) is
by the test
_
δ(x)f(x) dx = f(0) and Example 3 will do this.
For the repeating ramp, we integrate the square wave series for SW(x) and add the
average ramp height a
0
= π/2, halfway from 0 to π:
Ramp series RR(x) =
π
2
−
π
4
_
cos x
1
2
+
cos 3x
3
2
+
cos 5x
5
2
+
cos 7x
7
2
+· · ·
_
. (15)
The constant of integration is a
0
. Those coeﬃcients a
k
drop oﬀ like 1/k
2
. They could be
computed directly from formula (13) using
_
xcos kxdx, but this requires an integration
by parts (or a table of integrals or an appeal to Mathematica or Maple). It was much
easier to integrate every sine separately in SW(x), which makes clear the crucial point:
Each “degree of smoothness” in the function is reﬂected in a faster decay rate of its
Fourier coeﬃcients a
k
and b
k
.
No decay Delta functions (with spikes)
1/k decay Step functions (with jumps)
1/k
2
decay Ramp functions (with corners)
1/k
4
decay Spline functions (jumps in f
)
r
k
decay with r < 1 Analytic functions like 1/(2 −cos x)
Each integration divides the kth coeﬃcient by k. So the decay rate has an extra
1/k. The “RiemannLebesgue lemma” says that a
k
and b
k
approach zero for any
continuous function (in fact whenever
_
f(x)dx is ﬁnite). Analytic functions achieve
a new level of smoothness—they can be diﬀerentiated forever. Their Fourier series
and Taylor series in Chapter 5 converge exponentially fast.
The poles of 1/(2−cos x) will be complex solutions of cos x = 2. Its Fourier series
converges quickly because r
k
decays faster than any power 1/k
p
. Analytic functions
are ideal for computations—the Gibbs phenomenon will never appear.
Now we go back to δ(x) for what could be the most important example of all.
322 Chapter 4 Fourier Series and Integrals
Example 3 Find the (cosine) coeﬃcients of the delta function δ(x), made 2πperiodic.
Solution The spike occurs at the start of the interval [0, π] so safer to integrate from
−π to π. We ﬁnd a
0
= 1/2π and the other a
k
= 1/π (cosines because δ(x) is even):
Average a
0
=
1
2π
_
π
−π
δ(x) dx =
1
2π
Cosines a
k
=
1
π
_
π
−π
δ(x) cos kxdx =
1
π
Then the series for the delta function has all cosines in equal amounts:
Delta function δ(x) =
1
2π
+
1
π
[cos x + cos 2x + cos 3x +· · · ] . (16)
Again this series cannot truly converge (its terms don’t approach zero). But we can graph
the sum after cos 5x and after cos 10x. Figure 4.4 shows how these “partial sums” are
doing their best to approach δ(x). They oscillate faster and faster away from x = 0.
Actually there is a neat formula for the partial sum δ
N
(x) that stops at cos Nx. Start
by writing each term 2 cos θ as e
iθ
+ e
−iθ
:
δ
N
=
1
2π
[1 + 2 cos x +· · · + 2 cos Nx] =
1
2π
_
1 + e
ix
+ e
−ix
+· · · + e
iNx
+ e
−iNx
¸
.
This is a geometric progression that starts from e
−iNx
and ends at e
iNx
. We have powers
of the same factor e
ix
. The sum of a geometric series is known:
Partial sum
up to cos Nx
δ
N
(x) =
1
2π
e
i(N+
1
2
)x
−e
−i(N+
1
2
)x
e
ix/2
−e
−ix/2
=
1
2π
sin(N +
1
2
)x
sin
1
2
x
. (17)
This is the function graphed in Figure 4.4. We claim that for any N the area underneath
δ
N
(x) is 1. (Each cosine integrated from −π to π gives zero. The integral of 1/2π is
1.) The central “lobe” in the graph ends when sin(N +
1
2
)x comes down to zero, and
that happens when (N +
1
2
)x = ±π. I think the area under that lobe (marked by bullets)
approaches the same number 1.18 . . . that appears in the Gibbs phenomenon.
In what way does δ
N
(x) approach δ(x)? The terms cos nx in the series jump around
at each point x = 0, not approaching zero. At x = π we see
1
2π
[1 −2 +2 −2 +· · · ] and
the sum is 1/2π or −1/2π. The bumps in the partial sums don’t get smaller than 1/2π.
The right test for the delta function δ(x) is to multiply by a smooth f(x) =
a
k
cos kx
and integrate, because we only know δ(x) from its integrals
_
δ(x)f(x) dx = f(0):
Weak convergence
of δ
N
(x) to δ(x)
_
π
−π
δ
N
(x)f(x) dx = a
0
+· · · +a
N
→f(0) . (18)
In this integrated sense (weak sense) the sums δ
N
(x) do approach the delta function !
The convergence of a
0
+· · · + a
N
is the statement that at x = 0 the Fourier series of a
smooth f(x) =
a
k
cos kx converges to the number f(0).
4.1 Fourier Series for Periodic Functions 323
−π π 0
δ
5
(x)
δ
10
(x)
height 11/2π
height 21/2π
height −1/2π
height 1/2π
Figure 4.4: The sums δ
N
(x) = (1 +2 cos x +· · · +2 cos Nx)/2π try to approach δ(x).
Complete Series: Sines and Cosines
Over the halfperiod [0, π], the sines are not orthogonal to all the cosines. In fact the
integral of sin x times 1 is not zero. So for functions F(x) that are not odd or even,
we move to the complete series (sines plus cosines) on the full interval. Since our
functions are periodic, that “full interval” can be [−π, π] or [0, 2π]:
Complete Fourier series F(x) = a
0
+
∞
n=1
a
n
cos nx +
∞
n=1
b
n
sin nx. (19)
On every “2π interval” all sines and cosines are mutually orthogonal. We ﬁnd the
Fourier coeﬃcients a
k
and b
k
in the usual way: Multiply (19) by 1 and cos kx and
sin kx, and integrate both sides from −π to π:
a
0
=
1
2π
_
π
−π
F(x) dx a
k
=
1
π
_
π
−π
F(x) cos kxdx b
k
=
1
π
_
π
−π
F(x) sin xdx. (20)
Orthogonality kills oﬀ inﬁnitely many integrals and leaves only the one we want.
Another approach is to split F(x) = C(x) + S(x) into an even part and an odd
part. Then we can use the earlier cosine and sine formulas. The two parts are
C(x) = F
even
(x) =
F(x) + F(−x)
2
S(x) = F
odd
(x) =
F(x) −F(−x)
2
. (21)
The even part gives the a’s and the odd part gives the b’s. Test on a short square
pulse from x = 0 to x = h—this onesided function is not odd or even.
324 Chapter 4 Fourier Series and Integrals
Example 4 Find the a’s and b’s if F(x) = square pulse =
_
1 for 0 < x < h
0 for h < x < 2π
Solution The integrals for a
0
and a
k
and b
k
stop at x = h where F(x) drops to zero.
The coeﬃcients decay like 1/k because of the jump at x = 0 and the drop at x = h:
Coeﬃcients of square pulse a
0
=
1
2π
_
h
0
1 dx =
h
2π
= average
a
k
=
1
π
_
h
0
cos kxdx =
sinkh
πk
b
k
=
1
π
_
h
0
sin kxdx =
1 −cos kh
πk
. (22)
If we divide F(x) by h, its graph is a tall thin rectangle: height
1
h
, base h, and area = 1.
When h approaches zero, F(x)/h is squeezed into a very thin interval. The tall
rectangle approaches (weakly) the delta function δ(x). The average height is area/2π =
1/2π. Its other coeﬃcients a
k
/h and b
k
/h approach 1/π and 0, already known for δ(x):
F(x)
h
→δ(x)
a
k
h
=
1
π
sin kh
kh
→
1
π
and
b
k
h
=
1 −cos kh
πkh
→0 as h →0. (23)
When the function has a jump, its Fourier series picks the halfway point. This
example would converge to F(0) =
1
2
and F(h) =
1
2
, halfway up and halfway down.
The Fourier series converges to F(x) at each point where the function is smooth.
This is a highly developed theory, and Carleson won the 2006 Abel Prize by proving
convergence for every x except a set of measure zero. If the function has ﬁnite energy
_
F(x)
2
dx, he showed that the Fourier series converges “almost everywhere.”
Energy in Function = Energy in Coeﬃcients
There is an extremely important equation (the energy identity) that comes from
integrating (F(x))
2
. When we square the Fourier series of F(x), and integrate from
−π to π, all the “cross terms” drop out. The only nonzero integrals come from 1
2
and cos
2
kx and sin
2
kx, multiplied by a
2
0
and a
2
k
and b
2
k
:
Energy in F(x) =
_
π
−π
(a
0
+
a
k
cos kx +
b
k
sin kx)
2
dx
_
π
−π
(F(x))
2
dx = 2πa
2
0
+ π(a
2
1
+ b
2
1
+ a
2
2
+ b
2
2
+· · · ).
(24)
The energy in F(x) equals the energy in the coeﬃcients. The left side is like the
length squared of a vector, except the vector is a function. The right side comes from
an inﬁnitely long vector of a’s and b’s. The lengths are equal, which says that the
Fourier transform from function to vector is like an orthogonal matrix. Normalized
by constants
√
2π and
√
π, we have an orthonormal basis in function space.
What is this function space ? It is like ordinary 3dimensional space, except the
“vectors” are functions. Their length f comes from integrating instead of adding:
f
2
=
_
f(x)
2
dx. These functions ﬁll Hilbert space. The rules of geometry hold:
4.1 Fourier Series for Periodic Functions 325
Length f
2
= (f, f) comes from the inner product (f, g) =
_
f(x)g(x) dx
Orthogonal functions (f, g) = 0 produce a right triangle: f + g
2
= f
2
+g
2
I have tried to draw Hilbert space in Figure 4.5. It has inﬁnitely many axes. The
energy identity (24) is exactly the Pythagoras Law in inﬁnitedimensional space.
©
s
X
v
0
=
1
√
2π
v
1
=
cos x
√
π
v
2
=
sin x
√
π
f = A
0
v
0
+ A
1
v
1
+ B
1
v
2
+· · ·
function in Hilbert space
f
2
= A
2
0
+ A
2
1
+ B
2
1
+· · ·
v
2k−1
=
cos kx
√
π
v
2k
=
sin kx
√
π
90
◦
(v
i
, v
j
)=0
Figure 4.5: The Fourier series is a combination of orthonormal v’s (sines and cosines).
Complex Exponentials c
k
e
ikx
This is a small step and we have to take it. In place of separate formulas for a
0
and a
k
and b
k
, we will have one formula for all the complex coeﬃcients c
k
. And the function
F(x) might be complex (as in quantum mechanics). The Discrete Fourier Transform
will be much simpler when we use N complex exponentials for a vector. We practice
in advance with the complex inﬁnite series for a 2πperiodic function:
Complex Fourier series F(x) = c
0
+ c
1
e
ix
+ c
−1
e
−ix
+· · · =
∞
n=−∞
c
n
e
inx
(25)
If every c
n
= c
−n
, we can combine e
inx
with e
−inx
into 2 cos nx. Then (25) is the
cosine series for an even function. If every c
n
= −c
−n
, we use e
inx
−e
−inx
= 2i sin nx.
Then (25) is the sine series for an odd function and the c’s are pure imaginary.
To ﬁnd c
k
, multiply (25) by e
−ikx
(not e
ikx
) and integrate from −π to π:
_
π
−π
F(x)e
−ikx
dx =
_
π
−π
c
0
e
−ikx
dx+
_
π
−π
c
1
e
ix
e
−ikx
dx+· · · +
_
π
−π
c
k
e
ikx
e
−ikx
dx+· · ·
The complex exponentials are orthogonal. Every integral on the right side is zero,
except for the highlighted term (when n = k and e
ikx
e
−ikx
= 1). The integral of 1 is
2π. That surviving term gives the formula for c
k
:
Fourier coeﬃcients
_
π
−π
F(x)e
−ikx
dx = 2πc
k
for k = 0, ±1, . . . (26)
326 Chapter 4 Fourier Series and Integrals
Notice that c
0
= a
0
is still the average of F(x), because e
0
= 1. The orthogonality
of e
inx
and e
ikx
is checked by integrating, as always. But the complex inner product
(F, G) takes the complex conjugate G of G. Before integrating, change e
ikx
to e
−ikx
:
Complex inner product Orthogonality of e
inx
and e
ikx
(F, G) =
_
π
−π
F(x)G(x) dx
_
π
−π
e
i(n−k)x
dx =
_
e
i(n−k)x
i(n −k)
_
π
−π
= 0 .
(27)
Example 5 Add the complex series for 1/(2 −e
ix
) and 1/(2 −e
−ix
). These geometric
series have exponentially fast decay from 1/2
k
. The functions are analytic.
_
1
2
+
e
ix
4
+
e
2ix
8
+··
_
+
_
1
2
+
e
−ix
4
+
e
−2ix
8
+··
_
= 1 +
cos x
2
+
cos 2x
4
+
cos 3x
8
+··
When we add those functions, we get a real analytic function:
1
2 −e
ix
+
1
2 −e
−ix
=
(2 −e
−ix
) + (2 −e
ix
)
(2 −e
ix
)(2 −e
−ix
)
=
4 −2 cos x
5 −4 cos x
(28)
This ratio is the inﬁnitely smooth function whose cosine coeﬃcients are 1/2
k
.
Example 6 Find c
k
for the 2πperiodic shifted pulse F(x) =
_
1 for s ≤ x ≤ s + h
0 elsewhere in [−π, π]
Solution The integrals (26) from −π to π become integrals from s to s + h:
c
k
=
1
2π
_
s +h
s
1 · e
−ikx
dx =
1
2π
_
e
−ikx
−ik
_
s +h
s
= e
−iks
_
1 −e
−ikh
2πik
_
. (29)
Notice above all the simple eﬀect of the shift by s. It “modulates” each c
k
by e
−iks
. The
energy is unchanged, the integral of F
2
just shifts, and all e
−iks
 = 1:
Shift F(x) to F(x −s) ←→ Multiply c
k
by e
−iks
. (30)
Example 7 Centered pulse with shift s = −h/2. The square pulse becomes centered
around x = 0. This even function equals 1 on the interval from −h/2 to h/2:
Centered by s = −
h
2
c
k
= e
ikh/2
1 −e
−ikh
2πik
=
1
2π
sin(kh/2)
k/2
.
Divide by h for a tall pulse. The ratio of sin(kh/2) to kh/2 is the sinc function:
Tall pulse
F
centered
h
=
1
2π
∞
−∞
sinc
_
kh
2
_
e
ikx
=
_
1/h for −h/2 ≤ x ≤ h/2
0 elsewhere in [−π, π]
That division by h produces area = 1. Every coeﬃcient approaches
1
2π
as h →0.
The Fourier series for the tall thin pulse again approaches the Fourier series for δ(x).
4.1 Fourier Series for Periodic Functions 327
Hilbert space can contain vectors c = (c
0
, c
1
, c
−1
, c
2
, c
−2
, · · · ) instead of functions
F(x). The length of c is 2π
c
k

2
=
_
F
2
dx. The function space is often denoted
by L
2
and the vector space is
2
. The energy identity is trivial (but deep). Integrating
the Fourier series for F(x) times F(x), orthogonality kills every c
n
c
k
for n = k. This
leaves the c
k
c
k
= c
k

2
:
_
π
−π
F(x)
2
dx =
_
π
−π
(
c
n
e
inx
)(
c
k
e
−ikx
)dx = 2π(c
0

2
+c
1

2
+c
−1

2
+··) . (31)
This is Plancherel’s identity: The energy in xspace equals the energy in kspace.
Finally I want to emphasize the three big rules for operating on F(x) =
c
k
e
ikx
:
1. The derivative
dF
dx
has Fourier coeﬃcients ikc
k
(energy moves to high k).
2. The integral of F(x) has Fourier coeﬃcients
c
k
ik
, k = 0 (faster decay).
3. The shift to F(x−s) has Fourier coeﬃcients e
−iks
c
k
(no change in energy).
Application: Laplace’s Equation in a Circle
Our ﬁrst application is to Laplace’s equation. The idea is to construct u(x, y) as an
inﬁnite series, choosing its coeﬃcients to match u
0
(x, y) along the boundary. Every
thing depends on the shape of the boundary, and we take a circle of radius 1.
Begin with the simple solutions 1, r cos θ, r sin θ, r
2
cos 2θ, r
2
sin 2θ, ... to Laplace’s
equation. Combinations of these special solutions give all solutions in the circle:
u(r, θ) = a
0
+ a
1
r cos θ + b
1
r sin θ + a
2
r
2
cos 2θ + b
2
r
2
sin 2θ +· · · (32)
It remains to choose the constants a
k
and b
k
to make u = u
0
on the boundary.
For a circle u
0
(θ) is periodic, since θ and θ + 2π give the same point:
Set r = 1 u
0
(θ) = a
0
+a
1
cos θ +b
1
sin θ +a
2
cos 2θ +b
2
sin 2θ +· · · (33)
This is exactly the Fourier series for u
0
. The constants a
k
and b
k
must be the
Fourier coeﬃcients of u
0
(θ). Thus the problem is completely solved, if an inﬁnite
series (32) is acceptable as the solution.
Example 8 Point source u
0
= δ(θ) at θ = 0 The whole boundary is held at u
0
= 0,
except for the source at x = 1, y = 0. Find the temperature u(r, θ) inside.
Fourier series for δ u
0
(θ) =
1
2π
+
1
π
(cos θ + cos 2θ + cos 3θ +· · · ) =
1
2π
∞
−∞
e
inθ
328 Chapter 4 Fourier Series and Integrals
Inside the circle, each cos nθ is multiplied by r
n
:
Inﬁnite series for u u(r, θ) =
1
2π
+
1
π
(r cos θ + r
2
cos 2θ + r
3
cos 3θ +· · · ) (34)
Poisson managed to sum this inﬁnite series! It involves a series of powers of re
iθ
.
So we know the response at every (r, θ) to the point source at r = 1, θ = 0:
Temperature inside circle u(r, θ) =
1
2π
1 −r
2
1 + r
2
−2r cos θ
(35)
At the center r = 0, this produces the average of u
0
= δ(θ) which is a
0
= 1/2π. On the
boundary r = 1, this produces u = 0 except at the point source where cos 0 = 1:
On the ray θ = 0 u(r, θ) =
1
2π
1 −r
2
1 + r
2
−2r
=
1
2π
1 + r
1 −r
. (36)
As r approaches 1, the solution becomes inﬁnite as the point source requires.
Example 9 Solve for any boundary values u
0
(θ) by integrating over point sources.
When the point source swings around to angle ϕ, the solution (35) changes from θ to
θ −ϕ. Integrate this “Green’s function” to solve in the circle:
Poisson’s formula u(r, θ) =
1
2π
_
π
−π
u
0
(ϕ)
1 −r
2
1 + r
2
−2r cos(θ −ϕ)
dϕ (37)
Ar r = 0 the fraction disappears and u is the average
_
u
0
(ϕ)dϕ/2π. The steady
state temperature at the center is the average temperature around the circle.
Poisson’s formula illustrates a key idea. Think of any u
0
(θ) as a circle of point sources.
The source at angle ϕ = θ produces the solution inside the integral (37). Integrating
around the circle adds up the responses to all sources and gives the response to u
0
(θ).
Example 10 u
0
(θ) = 1 on the top half of the circle and u
0
= −1 on the bottom half.
Solution The boundary values are the square wave SW(θ). Its sine series is in (8):
Square wave for u
0
(θ) SW(θ) =
4
π
_
sin θ
1
+
sin 3θ
3
+
sin 5θ
5
+· · ·
_
(38)
Inside the circle, multiplying by r, r
2
, r
3
,... gives fast decay of high frequencies:
Rapid decay inside u(r, θ) =
4
π
_
r sin θ
1
+
r
3
sin 3θ
3
+
r
5
sin 5θ
5
+· · ·
_
(39)
Laplace’s equation has smooth solutions, even when u
0
(θ) is not smooth.
4.1 Fourier Series for Periodic Functions 329
WORKED EXAMPLE
A hot metal bar is moved into a freezer (zero temperature). The sides of the bar
are coated so that heat only escapes at the ends. What is the temperature u(x, t)
along the bar at time t? It will approach u = 0 as all the heat leaves the bar.
Solution The heat equation is u
t
= u
xx
. At t = 0 the whole bar is at a constant
temperature, say u=1. The ends of the bar are at zero temperature for all time t >0.
This is an initialboundary value problem:
Heat equation u
t
= u
xx
with u(x, 0) = 1 and u(0, t) = u(π, t) = 0. (40)
Those zero boundary conditions suggest a sine series. Its coeﬃcients depend on t:
Series solution of the heat equation u(x, t) =
∞
1
b
n
(t) sin nx. (41)
The form of the solution shows separation of variables. In a comment below, we
look for products A(x) B(t) that solve the heat equation and the boundary conditions.
What we reach is exactly A(x) = sin nx and the series solution (41).
Two steps remain. First, choose each b
n
(t) sin nx to satisfy the heat equation:
Substitute into u
t
= u
xx
b
n
(t) sin nx = −n
2
b
n
(t) sin nx b
n
(t) = e
−n
2
t
b
n
(0).
Notice b
n
= −n
2
b
n
. Now determine each b
n
(0) from the initial condition u(x, 0) = 1
on (0, π). Those numbers are the Fourier sine coeﬃcients of SW(x) in equation (38):
Box function/square wave
∞
1
b
n
(0) sin nx = 1 b
n
(0) =
4
πn
for odd n
This completes the series solution of the initialboundary value problem:
Bar temperature u(x, t) =
odd n
4
πn
e
−n
2
t
sin nx. (42)
For large n (high frequencies) the decay of e
−n
2
t
is very fast. The dominant term
(4/π)e
−t
sin x for large times will come from n = 1. This is typical of the heat
equation and all diﬀusion, that the solution (the temperature proﬁle) becomes very
smooth as t increases.
Numerical diﬃculty I regret any bad news in such a beautiful solution. To compute
u(x, t), we would probably truncate the series in (42) to N terms. When that ﬁnite
series is graphed on the website, serious bumps appear in u
N
(x, t). You ask if there
is a physical reason but there isn’t. The solution should have maximum temperature
at the midpoint x = π/2, and decay smoothly to zero at the ends of the bar.
330 Chapter 4 Fourier Series and Integrals
Those unphysical bumps are precisely the Gibbs phenomenon. The initial
u(x, 0) is 1 on (0, π) but its odd reﬂection is −1 on (−π, 0). That jump has produced
the slow 4/πn decay of the coeﬃcients, with Gibbs oscillations near x = 0 and x = π.
The sine series for u(x, t) is not a success numerically. Would ﬁnite diﬀerences help?
Separation of variables We found b
n
(t) as the coeﬃcient of an eigenfunction sin nx.
Another good approach is to put u = A(x) B(t) directly into u
t
= u
xx
:
Separation A(x) B
(t) = A
(x) B(t) requires
A
(x)
A(x)
=
B
(t)
B(t)
= constant. (43)
A
/A is constant in space, B
/B is constant in time, and they are equal:
A
A
= −λ gives A = sin
√
λx and cos
√
λ x
B
B
= −λ gives B = e
−λt
The products AB = e
−λt
sin
√
λ x and e
−λt
cos
√
λ x solve the heat equation for any
number λ. But the boundary condition u(0, t) = 0 eliminates the cosines. Then
u(π, t) = 0 requires λ = n
2
= 1, 4, 9, . . . to have sin
√
λπ = 0. Separation of variables
has recovered the functions in the series solution (42).
Finally u(x, 0) = 1 determines the numbers 4/πn for odd n. We ﬁnd zero for even
n because sin nx has n/2 positive loops and n/2 negative loops. For odd n, the extra
positive loop is a fraction 1/n of all loops, giving slow decay of the coeﬃcients.
Heat bath (the opposite problem) The solution on the website is 1 − u(x, t),
because it solves a diﬀerent problem. The bar is initially frozen at U(x, 0) =
0. It is placed into a heat bath at the ﬁxed temperature U = 1 (or U = T
0
).
The new unknown is U and its boundary conditions are no longer zero.
The heat equation and its boundary conditions are solved ﬁrst by U
B
(x, t). In
this example U
B
≡ 1 is constant. Then the diﬀerence V = U −U
B
has zero boundary
values, and its initial values are V = −1. Now the eigenfunction method (or sepa
ration of variables) solves for V . (The series in (42) is multiplied by −1 to account
for V (x, 0) = −1.) Adding back U
B
solves the heat bath problem: U = U
B
+ V =
1 −u(x, t).
Here U
B
≡ 1 is the steady state solution at t = ∞, and V is the transient solution.
The transient starts at V = −1 and decays quickly to V = 0.
Heat bath at one end The website problem is diﬀerent in another way too. The
Dirichlet condition u(π, t) = 1 is replaced by the Neumann condition u
(1, t) = 0.
Only the left end is in the heat bath. Heat ﬂows down the metal bar and out at the
far end, now located at x = 1. How does the solution change for ﬁxedfree?
Again U
B
= 1 is a steady state. The boundary conditions apply to V = 1 −U
B
:
Fixedfree
eigenfunctions
V (0) = 0 and V
(1) = 0 lead to A(x) = sin
_
n +
1
2
_
πx. (44)
4.1 Fourier Series for Periodic Functions 331
Those eigenfunctions give a new form for the sum of B
n
(t) A
n
(x):
Fixedfree solution V (x, t) =
odd n
B
n
(0) e
−(n+
1
2
)
2
π
2
t
sin
_
n +
1
2
_
πx. (45)
All frequencies shift by
1
2
and multiply by π, because A
= −λA has a free end
at x = 1. The crucial question is: Does orthogonality still hold for these new
eigenfunctions sin
_
n +
1
2
_
πx on [0, 1]? The answer is yes because this ﬁxedfree
“Sturm–Liouville problem” A
= −λA is still symmetric.
Summary The series solutions all succeed but the truncated series all fail. We can
see the overall behavior of u(x, t) and V (x, t). But their exact values close to the
jumps are not computed well until we improve on Gibbs.
We could have solved the ﬁxedfree problem on [0, 1] with the ﬁxedﬁxed solution
on [0, 2]. That solution will be symmetric around x = 1 so its slope there is zero.
Then rescaling x by 2π changes sin(n +
1
2
)πx into sin(2n + 1)x. I hope you like the
graphics created by Aslan Kasimov on the cse website.
Problem Set 4.1
1 Find the Fourier series on −π ≤ x ≤ π for
(a) f(x) = sin
3
x, an odd function
(b) f(x) =  sinx, an even function
(c) f(x) = x
(d) f(x) = e
x
, using the complex form of the series.
What are the even and odd parts of f(x) = e
x
and f(x) = e
ix
?
2 From Parseval’s formula the square wave sine coeﬃcients satisfy
π(b
2
1
+ b
2
2
+· · · ) =
_
π
−π
f(x)
2
dx =
_
π
−π
1 dx = 2π.
Dirive the remarkable sum π
2
= 8(1 +
1
9
+
1
25
+· · · ).
3 If a square pulse is centered at x = 0 to give
f(x) = 1 for x <
π
2
, f(x) = 0 for
π
2
< x < π,
draw its graph and ﬁnd its Fourier coeﬃcients a
k
and b
k
.
4 Suppose f has period T instead of 2x, so that f(x) = f(x +T). Its graph from
−T/2 to T/2 is repeated on each successive interval and its real and complex
Fourier series are
f(x) = a
0
+ a
1
cos
2πx
T
+ b
1
sin
2πx
T
+· · · =
∞
−∞
c
k
e
ik2πx/T
Multiplying by the right functions and integrating from −T/2 to T/2, ﬁnd a
k
,
b
k
, and c
k
.
332 Chapter 4 Fourier Series and Integrals
5 Plot the ﬁrst three partial sums and the function itself:
x(π −x) =
8
π
_
sin x
1
+
sin 3x
27
+
sin 5x
125
+· · ·
_
, 0 < x < π.
Why is 1/k
3
the decay rate for this function? What is the second derivative?
6 What constant function is closest in the least square sense to f = cos
2
x? What
multiple of cos x is closest to f = cos
3
x?
7 Sketch the 2πperiodic half wave with f(x) = sin x for 0 < x < π and f(x) = 0
for −π < x < 0. Find its Fourier series.
8 (a) Find the lengths of the vectors u = (1,
1
2
,
1
4
,
1
8
, . . .) and v = (1,
1
3
,
1
9
, . . .) in
Hilbert space and test the Schwarz inequality u
T
v
2
≤ (u
T
u)(v
T
v).
(b) For the functions f = 1 +
1
2
e
ix
+
1
4
e
2ix
+· · · and g = 1 +
1
3
e
ix
+
1
9
e
2ix
+· · ·
use part (a) to ﬁnd the numerical value of each term in

_
π
−π
f(x) g(x) dx
2
≤
_
π
−π
f(x)
2
dx
_
π
−π
g(x)
2
dx.
Substitute for f and g and use orthogonality (or Parseval).
9 Find the solution to Laplace’s equation with u
0
= θ on the boundary. Why is
this the imaginary part of 2(z −z
2
/2 + z
3
/3 · · · ) = 2 log(1 +z)? Conﬁrm that
on the unit circle z = e
iθ
, the imaginary part of 2 log(1 +z) agrees with θ.
10 If the boundary condition for Laplace’s equation is u
0
= 1 for 0 < θ < π and
u
0
= 0 for −π < θ < 0, ﬁnd the Fourier series solution u(r, θ) inside the unit
circle. What is u at the origin?
11 With boundary values u
0
(θ) = 1 +
1
2
e
iθ
+
1
4
e
2iθ
+· · · , what is the Fourier series
solution to Laplace’s equation in the circle? Sum the series.
12 (a) Verify that the fraction in Poisson’s formula satisﬁes Laplace’s equation.
(b) What is the response u(r, θ) to an impulse at the point (0, 1), at the angle
ϕ = π/2?
(c) If u
0
(ϕ) = 1 in the quartercircle 0 < ϕ < π/2 and u
0
= 0 elsewhere, show
that at points on the horizontal axis (and especially at the origin)
u(r, 0) =
1
2
+
1
2π
tan
−1
_
1 −r
2
−2r
_
by using
_
dϕ
b + c cos ϕ
=
1
√
b
2
−c
2
tan
−1
_√
b
2
−c
2
sin ϕ
c + b cos ϕ
_
.
4.1 Fourier Series for Periodic Functions 333
13 When the centered square pulse in Example 7 has width h = π, ﬁnd
(a) its energy
_
F(x)
2
dx by direct integration
(b) its Fourier coeﬃcients c
k
as speciﬁc numbers
(c) the sum in the energy identity (31) or (24)
If h = 2π, why is c
0
= 1 the only nonzero coeﬃcient ? What is F(x)?
14 In Example 5, F(x) = 1+(cos x)/2+· · · +(cos nx)/2
n
+· · · is inﬁnitely smooth:
(a) If you take 10 derivatives, what is the Fourier series of d
10
F/dx
10
?
(b) Does that series still converge quickly? Compare n
10
with 2
n
for n
1024
.
15 (A touch of complex analysis) The analytic function in Example 5 blows up
when 4 cos x = 5. This cannot happen for real x, but equation (28) shows
blowup if e
ix
= 2 or
1
2
. In that case we have poles at x = ±i log 2. Why are
there also poles at all the complex numbers x = ±i log 2 + 2πn?
16 (A second touch) Change 2’s to 3’s so that equation (28) has 1/(3 − e
ix
) +
1/(3 −e
−ix
). Complete that equation to ﬁnd the function that gives fast decay
at the rate 1/3
k
.
17 (For complex professors only) Change those 2’s and 3’s to 1’s:
1
1 −e
ix
+
1
1 −e
−ix
=
(1 −e
−ix
) + (1 −e
ix
)
(1 −e
ix
)(1 −e
−ix
)
=
2 −e
ix
−e
−ix
2 −e
ix
−e
−ix
= 1 .
A constant ! What happened to the pole at e
ix
= 1 ? Where is the dangerous
series (1 + e
ix
+· · · ) + (1 +e
−ix
+· · · ) = 2 + 2 cos x +· · · involving δ(x) ?
18 Following the Worked Example, solve the heat equation u
t
= u
xx
from a point
source u(x, 0) = δ(x) with free boundary conditions u
(π, t) = u
(−π, t) = 0.
Use the inﬁnite cosine series for δ(x) with time decay factors b
n
(t).
and then ﬁve terms. .1 as an odd function (with period 2π) that vanishes at x = 0 and x = π. Solution For k = 1. . use the ﬁrst formula (6) with S(x) = 1 between 0 and π: 2 π π π bk = sin kx dx = 0 2 − cos kx π k = 0 2 π 2 0 2 0 2 0 . . Example 1 Find the Fourier sine coeﬃcients bk of the square wave SW (x).. So we use this: 1 1 cos(n − k)x − cos(n + k)x . We will see that same 1/k decay rate for all functions formed from smooth pieces and jumps. . I will go immediately to the most important example of a Fourier sine series... You can see the allimportant Gibbs phenomenon appearing as these “partial sums” .1: The odd square wave with SW (x + 2π) = SW (x) = {1 or 0 or −1}. Then we are integrating (sin kx)2 = π π 1 2 − 1 cos 2kx: 2 (5) sin kx sin kx dx = 0 0 1 dx − 2 π 0 1 π cos 2kx dx = . Multiply both sides of (2) by 2/π: Sine coeﬃcients S(−x) = −S(x) bk = 2 π π S(x) sin kx dx = 0 1 π S(x) sin kx dx. . 2. . . The oddnumbered coeﬃcients bk = 4/πk decrease at the rate 1/k. π −π (6) Notice that S(x) sin kx is even (equal integrals from −π to 0 and from 0 to π). 2 2 The highlighted term in equation (2) is bkπ/2. It is drawn in Figure 4. .2 graphs this sum after one term. S(x) is an odd square wave with SW (x) = 1 for 0 < x < π.318 Chapter 4 Fourier Series and Integrals cos mx dx = sin mx π m 0 Zero comes quickly if we integrate Product of sines = 0 − 0. Put those coeﬃcients 4/πk and zero into the Fourier sine series for SW (x): Square wave SW (x) = 4 sin x sin 3x sin 5x sin 7x + + + +··· π 1 3 5 7 (8) Figure 4. SW (x) = 1 E −π 0 π 2π x Figure 4. sin nx sin kx = The exception is when n = k. (4) 2 2 Integrating cos mx with m = n − k and m = n + k proves orthogonality of the sines. 1 2 3 4 5 6 (7) The evennumbered coeﬃcients b2k are all zero because cos 2kπ = cos 0 = 1. then two terms.
This is the closest possible approximation to the square wave SW .1 Fourier Series for Periodic Functions 319 include more terms.4. Its height approaches 1. we safely approach SW (x) = 1 or −1. we are “projecting the function onto each axis sin kx. (10) . minimize the error over all b1 : π The error is 0 (SW −b1 sin x)2 dx The b1 derivative is −2 0 π (SW −b1 sin x) sin x dx. (9) The Gibbs phenomenon is the overshoot that moves closer and closer to the jumps. . by any multiple of sin x (closest in the least squares sense). Fourier Coeﬃcients are Best Let me look again at the ﬁrst term b1 sin x = (4/π) sin x. To see this optimal property of the Fourier coeﬃcients.2: Gibbs phenomenon: Partial sums bn sin nx overshoot near jumps. At x = π/2. The square wave has b2 = 0 because all other multiples of sin 2x increase the error. π 0 The integral of sin2 x is π/2.” Fourier Cosine Series The cosine series applies to even functions with C(−x) = C(x): ∞ Cosine series C(x) = a0 + a1 cos x + a2 cos 2x + · · · = a0 + n=1 an cos nx. We can ﬁnd the coeﬃcients bk one at a time. Each bk sin kx is as close as possible to SW . Term by term. 4 sin x sin 3x + π 1 3 4 sin x Dashed π 1 −π π Solid curve 5 terms: sin 9x 4 sin x +···+ π 1 9 overshoot−→ SW = 1 π 2 N 1 x x Figure 4. We try hard to avoid Gibbs but sometimes we can’t. and it does not decrease with more terms of the series! Overshoot is the one greatest obstacle to calculation of all discontinuous functions (like shock waves in ﬂuid ﬂow). So the derivative is zero when b1 = (2/π) This is exactly equation (6) for the Fourier coeﬃcient. . because the sines are orthogonal. the series gives a beautiful alternating formula for the number π: 1= 4 1 1 1 1 − + − +··· π 1 3 5 7 so that π=4 1 1 − 1 3 + 1 5 − 1 7 +··· .18 . S(x) sin x dx. Away from the jumps.
The derivative of RR is the odd square wave SW . First we ﬁnd formulas for the cosine coeﬃcients a0 and ak . Problem 4.320 Chapter 4 Fourier Series and Integrals Every cosine has period 2π. π −π (13) Again the integral over a full period from −π to π (also 0 to 2π) is just doubled. All other integrals are zero: π 0 sin nx cos nx dx = n π 0 = 0 − 0 = 0. The other cosine coeﬃcients ak come from the orthogonality of cosines.1 proves this by an identity for cos nx cos kx—now (4) has a plus sign. π]. the integral of a0 is a0 π (divide both sides by π). we multiply both sides of (10) by cos kx and integrate from 0 to π: π π π π C(x) cos kx dx = 0 0 a0 cos kx dx+ 0 a1 cos x cos kx dx+··+ 0 ak(cos kx)2 dx+·· You know what is coming. On the right side. The bold nonzero term is akπ/2 and we multiply both sides by 2/π: Cosine coeﬃcients C(−x) = C(x) ak = 2 π π C(x) cos kx dx = 0 1 π C(x) cos kx dx . one less smooth. only the highlighted term can be nonzero. one smoother than SW . The derivative of SW is U D. the constant function 1 is orthogonal to cos nx over the interval [0.) RR and UD will be valuable examples. RR(x) = x E 2δ(x) T x −π c 2δ(x − 2π) T π c E Updown U D(x) −π 0 π 2π 0 −2δ(x + π) 2π x Repeating Ramp RR(x) Integral of Square Wave −2δ(x − π) Figure 4. The constant term a0 is the average value of the function C(x): a0 = Average a0 = 1 π π C(x) dx = 0 π 1 C(x) dx.3: The repeating ramp RR and the updown UD (periodic spikes) are even. On the right side. (12) In words. . (For sines.1. 2π −π (11) I just integrated every term in the cosine series (10) from 0 to π. That sawtooth ramp RR is the integral of the square wave. As with sines. the repeating ramp RR(x) and the updown train UD(x) of delta functions. Figure 4. the integral and derivative are cosines.3 shows two even functions. The delta functions in UD give the derivative of the square wave.
Analytic functions achieve a new level of smoothness—they can be diﬀerentiated forever. Unoﬃcially this sum of cosines has all 1’s at x = 0 and all −1’s at x = π. 2 4 12 32 52 72 (15) The constant of integration is a0 .1 Fourier Series for Periodic Functions 321 Example 2 Find the cosine coeﬃcients of the ramp RR(x) and the updown UD(x). π (14) Those coeﬃcients don’t decay at all. The “RiemannLebesgue lemma” says that ak and bk approach zero for any continuous function (in fact whenever f (x)dx is ﬁnite). Those coeﬃcients ak drop oﬀ like 1/k 2 . which makes clear the crucial point: Each “degree of smoothness” in the function is reﬂected in a faster decay rate of its Fourier coeﬃcients ak and bk . Its Fourier series converges quickly because r k decays faster than any power 1/k p . For the repeating ramp. π 1 3 5 7 Take the derivative of every term to produce cosines in the updown delta function: Updown series UD(x) = 4 [cos x + cos 3x + cos 5x + cos 7x + · · · ] . Their Fourier series and Taylor series in Chapter 5 converge exponentially fast. The true way to recognize δ(x) is by the test δ(x)f (x) dx = f (0) and Example 3 will do this. Solution The simplest way is to start with the sine series for the square wave: SW (x) = 4 sin x sin 3x sin 5x sin 7x + + + +··· . Now we go back to δ(x) for what could be the most important example of all. Then +∞ and −∞ are consistent with 2δ(x) and −2δ(x − π). .4. The terms in the series don’t approach zero. we integrate the square wave series for SW (x) and add the average ramp height a0 = π/2. halfway from 0 to π: Ramp series RR(x) = π π cos x cos 3x cos 5x cos 7x − + + + +··· . No decay 1/k decay 1/k2 decay 1/k4 decay r k decay with r < 1 Delta functions (with spikes) Step functions (with jumps) Ramp functions (with corners) Spline functions (jumps in f ) Analytic functions like 1/(2 − cos x) Each integration divides the kth coeﬃcient by k. but this requires an integration by parts (or a table of integrals or an appeal to Mathematica or Maple). Analytic functions are ideal for computations—the Gibbs phenomenon will never appear. It was much easier to integrate every sine separately in SW (x). So the decay rate has an extra 1/k. They could be computed directly from formula (13) using x cos kx dx. so oﬃcially the series cannot converge. Nevertheless it is somehow correct and important. The poles of 1/(2 − cos x) will be complex solutions of cos x = 2.
2π 2π This is a geometric progression that starts from e−iN x and ends at eiN x . We have powers of the same factor eix . At x = π we see 2π [1 − 2 + 2 − 2 + · · · ] and the sum is 1/2π or −1/2π. Start by writing each term 2 cos θ as eiθ + e−iθ : δN = 1 1 [1 + 2 cos x + · · · + 2 cos Nx] = 1 + eix + e−ix + · · · + eiN x + e−iN x . .4. Actually there is a neat formula for the partial sum δN (x) that stops at cos Nx. that appears in the Gibbs phenomenon. In what way does δN (x) approach δ(x)? The terms cos nx in the series jump around 1 at each point x = 0.18 . We claim that for any N the area underneath δN (x) is 1. π] so safer to integrate from −π to π. = 2π eix/2 − e−ix/2 2π sin 1 x 2 1 1 (17) This is the function graphed in Figure 4. 2π π (16) Again this series cannot truly converge (its terms don’t approach zero). Figure 4. made 2πperiodic. Solution The spike occurs at the start of the interval [0. I think the area under that lobe (marked by bullets) 2 approaches the same number 1. (18) In this integrated sense (weak sense) the sums δN (x) do approach the delta function ! The convergence of a0 + · · · + aN is the statement that at x = 0 the Fourier series of a smooth f (x) = ak cos kx converges to the number f (0). The right test for the delta function δ(x) is to multiply by a smooth f (x) = ak cos kx and integrate. not approaching zero.322 Chapter 4 Fourier Series and Integrals Example 3 Find the (cosine) coeﬃcients of the delta function δ(x). . The bumps in the partial sums don’t get smaller than 1/2π.4 shows how these “partial sums” are doing their best to approach δ(x). and 2 that happens when (N + 1 )x = ±π. because we only know δ(x) from its integrals δ(x)f (x) dx = f (0): Weak convergence of δN (x) to δ(x) π −π δN (x)f (x) dx = a0 + · · · + aN → f (0) .) The central “lobe” in the graph ends when sin(N + 1 )x comes down to zero. (Each cosine integrated from −π to π gives zero. We ﬁnd a0 = 1/2π and the other ak = 1/π (cosines because δ(x) is even): Average a0 = π 1 1 δ(x) dx = 2π −π 2π Cosines ak = 1 1 π δ(x) cos kx dx = π −π π Then the series for the delta function has all cosines in equal amounts: Delta function δ(x) = 1 1 + [cos x + cos 2x + cos 3x + · · · ] . The integral of 1/2π is 1. But we can graph the sum after cos 5x and after cos 10x. The sum of a geometric series is known: Partial sum up to cos N x δN (x) = 1 ei(N + 2 )x − e−i(N + 2 )x 1 sin(N + 1 )x 2 . . They oscillate faster and faster away from x = 0.
We ﬁnd the Fourier coeﬃcients ak and bk in the usual way: Multiply (19) by 1 and cos kx and sin kx. The two parts are C(x) = Feven (x) = F (x) + F (−x) 2 S(x) = Fodd (x) = F (x) − F (−x) . Test on a short square pulse from x = 0 to x = h—this onesided function is not odd or even. So for functions F (x) that are not odd or even. π]. Since our functions are periodic. Another approach is to split F (x) = C(x) + S(x) into an even part and an odd part. . 2π]: ∞ ∞ Complete Fourier series F (x) = a0 + n=1 an cos nx + n=1 bn sin nx .4. (19) On every “2π interval” all sines and cosines are mutually orthogonal. Complete Series: Sines and Cosines Over the halfperiod [0. the sines are not orthogonal to all the cosines. 2 (21) The even part gives the a’s and the odd part gives the b’s.1 Fourier Series for Periodic Functions δ10 (x) height 21/2π 323 δ5 (x) height 11/2π height 1/2π height −1/2π −π 0 π Figure 4. that “full interval” can be [−π. Then we can use the earlier cosine and sine formulas. (20) 2π −π π −π π −π Orthogonality kills oﬀ inﬁnitely many integrals and leaves only the one we want.4: The sums δN (x) = (1 + 2 cos x + · · · + 2 cos Nx)/2π try to approach δ(x). π] or [0. In fact the integral of sin x times 1 is not zero. we move to the complete series (sines plus cosines) on the full interval. and integrate both sides from −π to π: a0 = π 1 1 π 1 π F (x) dx ak = F (x) cos kx dx bk = F (x) sin x dx.
its graph is a tall thin rectangle: height h . The coeﬃcients decay like 1/k because of the jump at x = 0 and the drop at x = h: Coeﬃcients of square pulse ak = 1 π h a0 = 1 2π h 1 dx = 0 h = average 2π 1 − cos kh πk . The tall rectangle approaches (weakly) the delta function δ(x). base h. and integrate from −π to π. except the “vectors” are functions. Its other coeﬃcients ak /h and bk /h approach 1/π and 0. The average height is area/2π = 1/2π. The right side comes from an inﬁnitely long vector of a’s and b’s. F (x)/h is squeezed into a very thin interval. The only nonzero integrals come from 12 and cos2 kx and sin2 kx. already known for δ(x): F (x) → δ(x) h ak 1 sin kh 1 = → h π kh π and bk 1 − cos kh = → 0 as h → 0. and Carleson won the 2006 Abel Prize by proving convergence for every x except a set of measure zero. and area = 1. (22) cos kx dx = 0 sin kh πk bk = 1 π h sin kx dx = 0 1 If we divide F (x) by h. all the “cross terms” drop out. The left side is like the length squared of a vector.324 Chapter 4 Fourier Series and Integrals 1 for 0 < x < h 0 for h < x < 2π Example 4 Find the a’s and b’s if F (x) = square pulse = Solution The integrals for a0 and ak and bk stop at x = h where F (x) drops to zero. its Fourier series picks the halfway point. multiplied by a2 and a2 and b2 : 0 k k Energy in F (x) = π (F (x))2dx −π π −π (a0 + ak cos kx + bk sin kx)2 dx = 2πa2 + π(a2 + b2 + a2 + b2 + · · · ). This example would converge to F (0) = 1 and F (h) = 1 . When h approaches zero. The rules of geometry hold: . which says that the Fourier transform from√ function to vector is like an orthogonal matrix. When we square the Fourier series of F (x).” Energy in Function = Energy in Coeﬃcients There is an extremely important equation (the energy identity) that comes from integrating (F (x))2 . If the function has ﬁnite energy F (x)2 dx. This is a highly developed theory. 2 2 The Fourier series converges to F (x) at each point where the function is smooth. Normalized √ by constants 2π and π. (23) h πkh When the function has a jump. The lengths are equal. except the vector is a function. Their length f comes from integrating instead of adding: f 2 = f (x)2 dx. we have an orthonormal basis in function space. halfway up and halfway down. These functions ﬁll Hilbert space. 0 1 1 2 2 (24) The energy in F (x) equals the energy in the coeﬃcients. he showed that the Fourier series converges “almost everywhere. What is this function space ? It is like ordinary 3dimensional space.
5. . Complex Exponentials ck eikx This is a small step and we have to take it. If every cn = −c−n .5: The Fourier series is a combination of orthonormal v’s (sines and cosines). Then (25) is the sine series for an odd function and the c’s are pure imaginary.1 Fourier Series for Periodic Functions Length f 2 325 = (f. . And the function F (x) might be complex (as in quantum mechanics). ±1. Every integral on the right side is zero. multiply (25) by e−ikx (not eikx ) and integrate from −π to π: π −π π −π π −π π −π F (x)e−ikx dx = c0 e−ikx dx+ c1 eix e−ikx dx+ · · ·+ ckeikxe−ikxdx + · · · The complex exponentials are orthogonal. vj ) = 0 ◦ cos x v1 = √ π Figure 4. g) = 0 produce a right triangle: f + g = f 2 + g 2 I have tried to draw Hilbert space in Figure 4. The energy identity (24) is exactly the Pythagoras Law in inﬁnitedimensional space.4. We practice in advance with the complex inﬁnite series for a 2πperiodic function: ∞ Complex Fourier series F (x) = c0 + c1 eix + c−1 e−ix + · · · = n=−∞ cn einx (25) If every cn = c−n . To ﬁnd ck. we can combine einx with e−inx into 2 cos nx. we will have one formula for all the complex coeﬃcients ck . The Discrete Fourier Transform will be much simpler when we use N complex exponentials for a vector. . cos kx v2k−1 = √ π s sin kx v2k = √ π sin x v2 = √ π X f = A0 v0 + A1 v1 + B1 v2 + · · · function in Hilbert space f 2 2 = A2 + A2 + B1 + · · · 0 1 1 © v0 = √ 2π 90 (vi. except for the highlighted term (when n = k and eikx e−ikx = 1). In place of separate formulas for a0 and ak and bk . (26) . we use einx − e−inx = 2i sin nx. Then (25) is the cosine series for an even function. f ) comes from the inner product (f. g) = f (x)g(x) dx 2 Orthogonal functions (f. It has inﬁnitely many axes. The integral of 1 is 2π. That surviving term gives the formula for ck : π Fourier coeﬃcients −π F (x)e−ikx dx = 2πck for k = 0.
1 eix e2ix + + + ·· + 2 4 8 1 e−ix e−2ix + + + ·· 2 4 8 =1+ cos x cos 2x cos 3x + + + ·· 2 4 8 When we add those functions. It “modulates” each c k by e−iks . The orthogonality of einx and eikx is checked by integrating. we get a real analytic function: 1 1 (2 − e−ix ) + (2 − eix ) 4 − 2 cos x + = = ix −ix ix )(2 − e−ix ) 2−e 2−e (2 − e 5 − 4 cos x This ratio is the inﬁnitely smooth function whose cosine coeﬃcients are 1/2 k . . The energy is unchanged. 1 for s ≤ x ≤ s + h 0 elsewhere in [−π. The functions are analytic. G) = −π F (x)G(x) dx Orthogonality of einx and eikx π π ei(n−k)x ei(n−k)x dx = = 0. Every coeﬃcient approaches 2π as h → 0. The square pulse becomes centered around x = 0. G) takes the complex conjugate G of G.326 Chapter 4 Fourier Series and Integrals Notice that c0 = a0 is still the average of F (x). These geometric series have exponentially fast decay from 1/2k . change eikx to e−ikx : Complex inner product π (F. the integral of F 2 just shifts. This even function equals 1 on the interval from −h/2 to h/2: Centered by s = − h 2 ck = eikh/2 1 sin(kh/2) 1 − e−ikh = . But the complex inner product (F. (30) Example 7 Centered pulse with shift s = −h/2. and all e−iks  = 1: Shift F (x) to F (x − s) ←→ Multiply ck by e−iks . i(n − k) −π −π (27) Example 5 Add the complex series for 1/(2 − eix ) and 1/(2 − e−ix ). because e0 = 1. The ratio of sin(kh/2) to kh/2 is the sinc function: Tall pulse Fcentered 1 = h 2π ∞ sinc −∞ kh 2 eikx = 1/h for − h/2 ≤ x ≤ h/2 0 elsewhere in [−π. (29) Notice above all the simple eﬀect of the shift by s. The Fourier series for the tall thin pulse again approaches the Fourier series for δ(x). Before integrating. as always. π] (28) Example 6 Find ck for the 2πperiodic shifted pulse F (x) = Solution The integrals (26) from −π to π become integrals from s to s + h: s+h s+h 1 1 e−ikx 1 · e−ikx dx = ck = 2π s 2π −ik s = e−iks 1 − e−ikh 2πik . π] 1 That division by h produces area = 1. 2πik 2π k/2 Divide by h for a tall pulse.
since θ and θ + 2π give the same point: Set r = 1 u0 (θ) = a0 + a1 cos θ + b1 sin θ + a2 cos 2θ + b2 sin 2θ + · · · (33) This is exactly the Fourier series for u0 . except for the source at x = 1. r 2 sin 2θ. y = 0. · · · ) instead of functions F (x). to Laplace’s equation. θ) = a0 + a1 r cos θ + b1 r sin θ + a2 r 2 cos 2θ + b2 r 2 sin 2θ + · · · (32) It remains to choose the constants ak and bk to make u = u0 on the boundary.. Thus the problem is completely solved. The derivative ck eikx : dF has Fourier coeﬃcients ikck (energy moves to high k). The length of c is 2π ck 2 = F 2dx. 2. y) as an inﬁnite series. c−2 . and we take a circle of radius 1.4. Finally I want to emphasize the three big rules for operating on F (x) = 1. c2 . Integrating the Fourier series for F (x) times F (x). r cos θ. k = 0 (faster decay). Everything depends on the shape of the boundary. . choosing its coeﬃcients to match u0 (x. θ) inside. The idea is to construct u(x. r sin θ.1 Fourier Series for Periodic Functions 327 Hilbert space can contain vectors c = (c0 . Find the temperature u(r. r 2 cos 2θ. The function space is often denoted by L2 and the vector space is 2 . orthogonality kills every cn ck for n = k. y) along the boundary. Combinations of these special solutions give all solutions in the circle: u(r. (31) This is Plancherel’s identity: The energy in xspace equals the energy in kspace. Example 8 Point source u0 = δ(θ) at θ = 0 The whole boundary is held at u0 = 0. ik The shift to F (x−s) has Fourier coeﬃcients e−iks ck (no change in energy). Fourier series for δ u0 (θ) = 1 1 1 + (cos θ + cos 2θ + cos 3θ + · · · ) = 2π π 2π ∞ einθ −∞ . For a circle u0 (θ) is periodic. 3. Application: Laplace’s Equation in a Circle Our ﬁrst application is to Laplace’s equation. c1 . The energy identity is trivial (but deep). dx ck The integral of F (x) has Fourier coeﬃcients . Begin with the simple solutions 1.. This leaves the ck ck = ck 2 : π −π F (x)2dx = π ( −π cn einx )( ck e−ikx )dx = 2π(c0 2 + c1 2 + c−1 2 + ··) . if an inﬁnite series (32) is acceptable as the solution. c−1 . The constants ak and bk must be the Fourier coeﬃcients of u0 (θ).
θ) = 1 2π π −π u0 (ϕ) 1 − r2 dϕ 1 + r 2 − 2r cos(θ − ϕ) (37) Ar r = 0 the fraction disappears and u is the average u0 (ϕ)dϕ/2π.. r 3 . gives fast decay of high frequencies: Rapid decay inside u(r. each cos nθ is multiplied by r n : Inﬁnite series for u u(r. . the solution (35) changes from θ to θ − ϕ. Integrate this “Green’s function” to solve in the circle: Poisson’s formula u(r. this produces u = 0 except at the point source where cos 0 = 1: On the ray θ = 0 u(r. Its sine series is in (8): SW (θ) = 4 sin θ sin 3θ sin 5θ + + +··· π 1 3 5 (38) Square wave for u0 (θ) Inside the circle. The source at angle ϕ = θ produces the solution inside the integral (37).. θ) = 4 r sin θ r 3 sin 3θ r 5 sin 5θ + + +··· π 1 3 5 (39) Laplace’s equation has smooth solutions. Solution The boundary values are the square wave SW (θ). The steady state temperature at the center is the average temperature around the circle. 2π 1 + r 2 − 2r 2π 1 − r (36) As r approaches 1. multiplying by r. Example 9 Solve for any boundary values u0 (θ) by integrating over point sources. θ) = 1 1 + (r cos θ + r 2 cos 2θ + r 3 cos 3θ + · · · ) (34) 2π π Poisson managed to sum this inﬁnite series! It involves a series of powers of re iθ .. θ) to the point source at r = 1. the solution becomes inﬁnite as the point source requires. even when u0 (θ) is not smooth.328 Chapter 4 Fourier Series and Integrals Inside the circle. Example 10 u0 (θ) = 1 on the top half of the circle and u0 = −1 on the bottom half. θ) = 1 − r2 1 1 1+r = . θ) = 1 − r2 1 2π 1 + r 2 − 2r cos θ (35) At the center r = 0. this produces the average of u0 = δ(θ) which is a0 = 1/2π. θ = 0: Temperature inside circle u(r. On the boundary r = 1. r 2 . When the point source swings around to angle ϕ. Poisson’s formula illustrates a key idea. Integrating around the circle adds up the responses to all sources and gives the response to u 0 (θ). So we know the response at every (r. Think of any u 0(θ) as a circle of point sources.
What is the temperature u(x. The solution should have maximum temperature at the midpoint x = π/2. The dominant term (4/π)e−t sin x for large times will come from n = 1. t) = odd 4 −n2 t sin nx. serious bumps appear in uN (x. choose each bn (t) sin nx to satisfy the heat equation: Substitute into ut = uxx bn (t) sin nx = −n2 bn (t) sin nx bn (t) = e−n t bn (0). 0) = 1 on (0. What we reach is exactly A(x) = sin nx and the series solution (41). First. we would probably truncate the series in (42) to N terms. This is an initialboundary value problem: Heat equation ut = uxx with u(x. (40) Those zero boundary conditions suggest a sine series. The sides of the bar are coated so that heat only escapes at the ends.1 Fourier Series for Periodic Functions 329 WORKED EXAMPLE A hot metal bar is moved into a freezer (zero temperature). 0) = 1 and u(0. Now determine each bn (0) from the initial condition u(x. Its coeﬃcients depend on t: ∞ Series solution of the heat equation u(x. 2 Notice bn = −n2 bn .4. t) = 0. that the solution (the temperature proﬁle) becomes very smooth as t increases. At t = 0 the whole bar is at a constant temperature. t) = u(π. When that ﬁnite series is graphed on the website. . The ends of the bar are at zero temperature for all time t > 0. t). π). Two steps remain. say u = 1. e πn n 2 (42) For large n (high frequencies) the decay of e−n t is very fast. t) = 1 bn (t) sin nx. and decay smoothly to zero at the ends of the bar. To compute u(x. Solution The heat equation is ut = uxx . t). (41) The form of the solution shows separation of variables. t) along the bar at time t? It will approach u = 0 as all the heat leaves the bar. In a comment below. This is typical of the heat equation and all diﬀusion. Those numbers are the Fourier sine coeﬃcients of SW (x) in equation (38): ∞ Box function/square wave 1 bn (0) sin nx = 1 bn (0) = 4 for odd n πn This completes the series solution of the initialboundary value problem: Bar temperature u(x. You ask if there is a physical reason but there isn’t. Numerical diﬃculty I regret any bad news in such a beautiful solution. we look for products A(x) B(t) that solve the heat equation and the boundary conditions.
and V is the transient solution. The sine series for u(x. . . now located at x = 1. (43) A(x) B(t) A /A is constant in space. Then the diﬀerence V = U − UB has zero boundary values. (44) . to have sin λ π = 0.330 Chapter 4 Fourier Series and Integrals Those unphysical bumps are precisely the Gibbs phenomenon.) Adding back UB solves the heat bath problem: U = UB + V = 1 − u(x. The new unknown is U and its boundary conditions are no longer zero. giving slow decay of the coeﬃcients. The initial u(x. 0). In this example UB ≡ 1 is constant. 0) = 1 determines the numbers 4/πn for odd n. The Dirichlet condition u(π. 0) is 1 on (0. Separation of variables has recovered the functions in the series solution (42). t). Heat bath at one end The website problem is diﬀerent in another way too. The heat equation and its boundary conditions are solved ﬁrst by UB (x. The boundary conditions apply to V = 1 − UB : Fixedfree eigenfunctions V (0) = 0 and V (1) = 0 lead to A(x) = sin n + 1 2 πx. Heat bath (the opposite problem) The solution on the website is 1 − u(x. The transient starts at V = −1 and decays quickly to V = 0. t). with Gibbs oscillations near x = 0 and x = π. t) =√0 eliminates the cosines. and they are equal: √ √ A = −λ gives A = sin λ x and cos λ x A B = −λ gives B = e−λt B √ √ The products AB = e−λt sin λ x and e−λt cos λ x solve the heat equation for any number λ. Would ﬁnite diﬀerences help? Separation of variables We found bn (t) as the coeﬃcient of an eigenfunction sin nx. Then u(π. Now the eigenfunction method (or separation of variables) solves for V . B /B is constant in time. (The series in (42) is multiplied by −1 to account for V (x. 4. That jump has produced the slow 4/πn decay of the coeﬃcients. For odd n. . Finally u(x. 0) = −1. t) = 0 requires λ = n2 = 1. the extra positive loop is a fraction 1/n of all loops. Heat ﬂows down the metal bar and out at the far end. It is placed into a heat bath at the ﬁxed temperature U = 1 (or U = T0 ). The bar is initially frozen at U(x. and its initial values are V = −1. But the boundary condition u(0. t) = 1 is replaced by the Neumann condition u (1. t). Only the left end is in the heat bath. π) but its odd reﬂection is −1 on (−π. because it solves a diﬀerent problem. We ﬁnd zero for even n because sin nx has n/2 positive loops and n/2 negative loops. t) is not a success numerically. Another good approach is to put u = A(x) B(t) directly into ut = uxx : Separation A(x) B (t) = A (x) B(t) requires B (t) A (x) = = constant. Here UB ≡ 1 is the steady state solution at t = ∞. 0) = 0. 9. t) = 0. How does the solution change for ﬁxedfree? Again UB = 1 is a steady state.
f (x) = 0 for < x < π. bk . 1] with the ﬁxedﬁxed solution on [0. an odd function f (x) =  sin x. What are the even and odd parts of f (x) = ex and f (x) = eix ? 2 From Parseval’s formula the square wave sine coeﬃcients satisfy π(b2 + b2 + · · · ) = 1 2 π −π π f (x)2 dx = 1 25 1 dx = 2π. 1]? The answer is yes because this ﬁxedfree 2 “Sturm–Liouville problem” A = −λA is still symmetric. using the complex form of the series.4. an even function f (x) = x f (x) = ex . We could have solved the ﬁxedfree problem on [0. Problem Set 4. Summary The series solutions all succeed but the truncated series all fail. Its graph from −T /2 to T /2 is repeated on each successive interval and its real and complex Fourier series are ∞ 2πx 2πx f (x) = a0 + a1 cos ck eik2πx/T + b1 sin +··· = T T −∞ Multiplying by the right functions and integrating from −T /2 to T /2. t) = odd n 331 Bn (0) e−(n+ 2 ) 1 2 2 π t sin n + 1 πx. But their exact values close to the jumps are not computed well until we improve on Gibbs. −π 1 Dirive the remarkable sum π 2 = 8(1 + 9 + + · · · ). because A = −λA has a free end 2 at x = 1. ﬁnd ak . . We can see the overall behavior of u(x. 3 If a square pulse is centered at x = 0 to give π π .1 1 Find the Fourier series on −π ≤ x ≤ π for (a) (b) (c) (d) f (x) = sin3 x. f (x) = 1 for x < 4 Suppose f has period T instead of 2x. 2 2 draw its graph and ﬁnd its Fourier coeﬃcients ak and bk . t) and V (x. so that f (x) = f (x + T ). I hope you like the 2 graphics created by Aslan Kasimov on the cse website.1 Fourier Series for Periodic Functions Those eigenfunctions give a new form for the sum of Bn (t) An (x): Fixedfree solution V (x. 2]. Then rescaling x by 2π changes sin(n + 1 )πx into sin(2n + 1)x. That solution will be symmetric around x = 1 so its slope there is zero. t). The crucial question is: Does orthogonality still hold for these new eigenfunctions sin n + 1 πx on [0. and ck . 2 (45) All frequencies shift by 1 and multiply by π.
1 . Why is this the imaginary part of 2(z − z 2 /2 + z 3 /3 · · · ) = 2 log(1 + z)? Conﬁrm that on the unit circle z = eiθ . (b) What is the response u(r. (a) Find the lengths of the vectors u = (1. Substitute for f and g and use orthogonality (or Parseval). 1 . Why is 1/k 3 the decay rate for this function? What is the second derivative? 6 7 8 What constant function is closest in the least square sense to f = cos2 x? What multiple of cos x is closest to f = cos3 x? Sketch the 2πperiodic half wave with f (x) = sin x for 0 < x < π and f (x) = 0 for −π < x < 0. 9 Find the solution to Laplace’s equation with u0 = θ on the boundary. 1 . . (b) For the functions f = 1 + 1 eix + 1 e2ix + · · · and g = 1 + 1 eix + 1 e2ix + · · · 2 4 3 9 use part (a) to ﬁnd the numerical value of each term in  π −π f (x) g(x) dx2 ≤ π −π f (x)2 dx π −π g(x)2 dx. If the boundary condition for Laplace’s equation is u0 = 1 for 0 < θ < π and u0 = 0 for −π < θ < 0. θ) to an impulse at the point (0. the imaginary part of 2 log(1 + z) agrees with θ. at the angle ϕ = π/2? (c) If u0 (ϕ) = 1 in the quartercircle 0 < ϕ < π/2 and u0 = 0 elsewhere. ﬁnd the Fourier series solution u(r. . 0 < x < π. . What is u at the origin? With boundary values u0 (θ) = 1 + 1 eiθ + 1 e2iθ + · · · . Find its Fourier series. . θ) inside the unit circle. . what is the Fourier series 2 4 solution to Laplace’s equation in the circle? Sum the series.) in 2 4 8 3 9 Hilbert space and test the Schwarz inequality uT v2 ≤ (uT u)(v T v). 1).) and v = (1. . (a) Verify that the fraction in Poisson’s formula satisﬁes Laplace’s equation. show that at points on the horizontal axis (and especially at the origin) u(r.332 5 Chapter 4 Fourier Series and Integrals Plot the ﬁrst three partial sums and the function itself: x(π − x) = 8 π sin x sin 3x sin 5x + + +··· 1 27 125 . 1 . 1 . 0) = 1 1 + tan−1 2 2π 1 − r2 −2r √ by using b2 − c2 sin ϕ c + b cos ϕ 10 11 12 1 dϕ =√ tan−1 2 − c2 b + c cos ϕ b . .
but equation (28) shows 1 blowup if eix = 2 or 2 . = 1 − eix 1 − e−ix (1 − eix )(1 − e−ix ) 2 − eix − e−ix A constant ! What happened to the pole at eix = 1 ? Where is the dangerous series (1 + eix + · · · ) + (1 + e−ix + · · · ) = 2 + 2 cos x + · · · involving δ(x) ? 18 Following the Worked Example. 16 17 . what is the Fourier series of d10 F/dx10 ? (b) Does that series still converge quickly? Compare n10 with 2n for n1024 . Complete that equation to ﬁnd the function that gives fast decay at the rate 1/3k . Use the inﬁnite cosine series for δ(x) with time decay factors bn (t). In that case we have poles at x = ±i log 2. Why are there also poles at all the complex numbers x = ±i log 2 + 2πn ? (A second touch) Change 2’s to 3’s so that equation (28) has 1/(3 − eix ) + 1/(3 − e−ix ). (For complex professors only) Change those 2’s and 3’s to 1’s: 1 1 (1 − e−ix ) + (1 − eix ) 2 − eix − e−ix + = = 1. 15 (A touch of complex analysis) The analytic function in Example 5 blows up when 4 cos x = 5. 0) = δ(x) with free boundary conditions u (π. This cannot happen for real x. why is c0 = 1 the only nonzero coeﬃcient ? What is F (x)? 14 In Example 5.4. t) = 0. F (x) = 1 + (cos x)/2 +· · ·+ (cos nx)/2n + · · · is inﬁnitely smooth: (a) If you take 10 derivatives. t) = u (−π. solve the heat equation ut = uxx from a point source u(x.1 Fourier Series for Periodic Functions 13 When the centered square pulse in Example 7 has width h = π. ﬁnd (a) its energy F (x)2 dx by direct integration 333 (b) its Fourier coeﬃcients ck as speciﬁc numbers (c) the sum in the energy identity (31) or (24) If h = 2π.
This action might not be possible to undo. Are you sure you want to continue?
Use one of your book credits to continue reading from where you left off, or restart the preview.