Real Analysis II

12 Fourier Series
12.0 Power Series Revisited
We say f is analytic on an open interval (a, b) if for any x
0
(a, b), there is a power series centered at
x
0
that represents f in some open interval containing x
0
. That is, there exists a c < x
0
< d b such that
for all x (c, d)
f(x) =

k=0
a
k
(x x
0
)
k
Remark 1) If f(x) =

k=0
a
k
(x x
0
)
k
, then
a
k
=
f
(k)
(x
0
)
k!
and hence f is innitely dierentiable. We write f C

(a, b)
2) The Taylor Series for f at x
0
is given by

k=0
f
(k)
(x
0
)
k!
(x x
0
)
k
.
3) The nth Taylor Polynomial for f at x
0
is given by
P
n
(x) =
n

k=0
f
(k)
(x
0
)
k!
(x x
0
)
k
4) The remainder term of order n is dened by R
n
(x) = f(x) P
n
(x).
5) (Taylors Formula) If f is analytic in (a, b) and x
0
(a, b), then there exists c
x
between x and c such
that
R
n
(x) =
f
(n+1)
(c
x
)
(n + 1)!
(x x
0
)
n+1
6) (Lagrange) If f C

(a, b), then for any x, x

0
(a, b),
R
n
(x) =
1
n!
_
x
x
0
(x t)
n
f
(n)
(t) dt
7) If f C

(a, b) and if
f
(n)
(x) 0 for all x (a, b),
then f is analytic on (a, b).
8) Suppose f and g are analytic on (a, b) and x
0
(a, b). If f(x) = g(x) for all x (a, x
0
), then there
exists a > 0 such that
f(x) = g(x) for all x (x
0
, x
0
+).
9) (Analytic Continuation) Suppose that I and J are open interval, that f is analytic on I and g
is analytic on J. If a < b are points in I J and f(x) = g(x) for all x (a, b), then f(x) = g(x) for all x I J.
Examples a) Polynomials are analytic.
b) All convergent power series are analytic.
c) The exponential function e
x
, the trig functions sinx and cos x are analytic on (, ). The rational
function 1/(1 x) is analytic on (1, 1). Furthermore, on the intervals of analyticity, we have
i) e
x
=

j=0
x
j
j!
ii) sinx =

j=0
(1)
j
x
2j+1
(2j+1)!
iii) cos x =

j=0
(1)
j
x
2j
(2j)!
iv)
1
1x
=

j=0
x
j
d) (Cauchy) The function
f(x) =
_

_
e
2/x
2
, x = 0
0, x = 0
belongs to C

(, ) but is not analytic on any open interval that contains x = 0. Find its Taylor series.
12.1 Denition of Fourier Series
Theorem 1 (Orthogonality Theorem)
(a)
_

cos(kx) cos(nx) dx = 0 (n = k; n, k = 0, 1, 2, 3, )
(b)
_

cos
2
(nx) dx =
_
, n 1
2, n = 0
(c)
_

sin(kx) sin(nx) dx = 0, (k = n; n, k = 1, 2, 3, )
(d)
_

sin
2
(nx) dx = , (n = 1, 2, 3, )
(e)
_

cos(kx) sin(nx) dx = 0, (n, k = 1, 2, 3, )

Proof Use the trig identities
cos(kx +nx) = cos(kx) cos(nx) sin(kx) sin(nx)
cos(kx nx) = cos(kx) cos(nx) + sin(kx) sin(nx)
_

cos(ax) =
1
a
sin(ax)

= 0, (a = 1, 2, 3, )
_

sin(ax) =
1
a
cos(ax)

= 0, (a = 1, 2, 3, )
Remark 1. Let f be a function dened on [, ]. Assume that, for each x [, ], f(x) can be
expressed as
f(x) =
a
0
2
+

k=1
_
a
k
cos(kx) +b
k
sin(kx)
_
Then using Theorem 1, we have the following
_

f(x) dx = a
0
For n 1, we have
_

f(x) cos(nx) dx =
a
0
2
_

cos(nx) dx +

k=1
_
a
k
_

cos(nx) cos(kx) dx + b
k
_

cos(nx) sin(kx) dx
_
= a
n
_

cos
2
(nx) dx = a
n
Therefore,
a
n
=
1

f(x) cos(nx) dx.

Similarly,
b
n
=
1

f(x) sin(nx) dx.

Denition If f is Riemann integrable over [, ], then the Fourier Series of f is the series
a
0
2
+

k=1
_
a
k
cos(kx) +b
k
sin(kx)
_
,
where
a
k
=
1

f(x) cos(kx) dx (k = 0, 1, 2, 3, )
and
b
k
=
1

f(x) sin(kx) dx, (k = 1, 2, 3, .)

The a
k
and b
k
are called the Fourier Coecients of f. For the purpose of clarity, we will write a
k
(f)
instead of a
k
and b
k
(f) instead of b
k
.
It is also common to write
f
a
0
2
+

k=1
_
a
k
cos(kx) +b
k
sin(kx)
_
.
We shall also write S
n
f for the nth partial sum of the Fourier series of f. Thus,
S
0
f(x) =
a
0
2
and for n 1,
(S
n
f)(x) =
a
0
2
+
n

k=1
_
a
k
cos(kx) + b
k
sin(kx)
_
Example Show that the Fourier series for f(x) = x is
2

k=1
(1)
k+1
k
sin(kx)
Since xcos(kx) is odd and xsin(kx) is even, we see that
a
k
=
1

f(x) cos(kx) dx = 0, k = 0, 1, 2, 3,
b
k
=
1

f(x) sin(kx) dx =
2

_

0
f(x) sin(kx) dx k = 1, 2, 3, .
Integration by parts yields
b
k
=
2

xcos(kx)
k

0
+
1
k
_

0
cos(kx) dx
_
=
2(1)
k+1
k
.
Example Show that the Fourier series for f(x) = |x| is

2
+
4

k=1
cos((2k 1)x)
(2k 1)
2
Since |x| cos(kx) is even and |x| sin(kx) is odd, we have
b
k
=
1

f(x) sin(kx) dx = 0 k = 1, 2, 3, .
a
k
=
1

f(x) cos(kx) dx =
2

_

0
xcos(kx) dx k = 0, 1, 2, 3, .
If k = 0, then we have
a
0
=
2

_

0
xdx =
2

2
2
_
=
and if k 1, integration by parts yields
a
k
=
2
k
2
(cos(k) 1) =
_
0 if k is even

4
k
2
if k is odd.
12.2 Formulation of Convergence and Summability Problems
Convergence Question. Given a function f periodic on R and integrable on [, ], does the Fourier
series of f converge to f?
Uniqueness Question. If a trigonometric series converges to f, is the series the Fourier series of f?
Theorem 2. If the trigonometric series
a
0
2
+

k=1
_
a
k
cos(kx) + b
k
sin(kx)
_
converges to f uniformly, then it is the Fourier series of f. That is,
a
k
=
1

f(x) cos(kx) dx, k = 0, 1, 2, 3,

b
k
=
1

f(x) sin(kx) dx, k = 1, 2, 3, .

Proof This follows from Remark 1 and the fact that the series converges uniformly.
Denition. 1) A Dirichlet kernel of order n is the function dened by
D
0
(x) =
1
2
, D
n
(x) =
1
2
+
n

k=1
cos(kx).
2) The Fejer kernel of order n is dened by
K
0
(x) =
1
2
, K
n
(x) =
1
2
+
n

k=1
_
1
k
n + 1
_
cos(kx).
Lemma 1.
K
n
(x) =
D
0
(x) +D
1
(x) + +D
n
(x)
n + 1
.
Proof. The formula is trivially true if n = 0. Suppose n 1. Then
K
n
(x) =
1
n + 1
_
n + 1
2
+
n

k=1
(n k + 1) cos(kx)
_
=
1
n + 1
_
_
1
2
+
n
2
+
n

k=1
n

j=k
1 cos(kx)
_
_
=
1
n + 1
_
_
1
2
+
n

j=1
_
_
1
2
+
j

k=1
cos(kx)
_
_
_
_
=
D
0
(x) +D
1
(x) +D
2
(x) + +D
n
(x)
n + 1
Lemma 2. If x R, x = 2k for k I, then for each n = 0, 1, 2, ,
D
n
(x) =
sin
_
n +
1
2
_
x
2 sin
_
x
2
_
and K
n
(x) =
2
n + 1
_
_
sin
_
n+1
2
_
x
2 sin
_
x
2
_
_
_
2
Proof. For n = 0, the lemma is trivial.
Fix n 1 and apply the sum-angle and telescoping to get
_
D
n
(x)
1
2
_
sin
_
x
2
_
=
n

k=1
cos(kx) sin
_
x
2
_
=
1
2
n

k=1
_
sin
_
k +
1
2
_
x sin
_
k
1
2
_
x
_
=
1
2
_
sin
_
n +
1
2
_
x sin
_
x
2
__
and hence
D
n
(x) =
sin
_
n +
1
2
_
x
2 sin
_
x
2
_
To prove the second formula we use the formula just proved and sum-angle formula we have
D
k
(x) sin
2
_
x
2
_
=
1
2
sin
_
x
2
_
sin
_
k +
1
2
_
x =
1
4
[cos(kx) cos(k + 1)x]
By Lemma 1 and telescoping, we get
(n + 1)K
n
(x) sin
2
_
x
2
_
=
n

k=0
D
k
(x) sin
2
_
x
2
_
=
1
4
n

k=0
[cos(kx) cos(k + 1)x]
=
1
4
[1 cos(n + 1)x] =
1
2
sin
2
_
x
2
_
and the second formula of the lemma follows by dividing.
Denition. A series

k=0
a
k
with partial sums s
n
=

n
k=0
a
k
is said to be Cesaro summable to a nite
number L if and only if

n
=
s
0
+s
1
+s
2
+ +s
n
n + 1
converges to L. Cesaro summable is also called (C, 1) summable and we write

k=0
a
k
= L (C, 1).
(Sections 2.11, and 3.9 of the text have more on this.)
Example The series

k=0
(1)
k
= 1 1 + 1 1 + 1 1 +
is divergent, since
s
n
=
_
0 if n is odd
1 if n is even
However,

n
=
_
n+2
2(n+1)
if n is even
1
2
if n is odd.
Hence
lim
n

n
=
1
2
, and so

k=0
(1)
k
=
1
2
(C, 1).
Denition. The Cesaro means of a Fourier series of f is denoted by
n
f and is given by
(
n
f)(x) =
(S
0
f)(x) + (S
1
f)(x) + + (S
n
f)(x)
n + 1
,
where S
k
f is the k partial sum of the Fourier series of f.
Lemma 3. If f is periodic on R and integrable on [, ], then for all x R and n = 0, 1, 2, 3 , we
have
(
n
f)(x) =
1

f(x t)K
n
(t) dt.
Proof For simplicity, let us write a
k
for a
k
(f) and b
k
for b
k
(f). For each j, we have
a
j
cos(jx) +b
j
sin(jx) =
1

f(u) cos(ju) cos(jx) du +

1

f(u) sin(ju) sin(jx) du

=
1

f(u)
_
cos(ju) cos(jx) + sin(ju) sin(jx)
_
du
=
1

f(u) cos(j(u x)) du.

Summing over j = 1, 2, , k and adding a
0
/2, we have
(S
k
f)(x) =
a
0
2
+
n

j=1
a
j
cos(jx) +b
j
sin(jx) =
1

f(u)
_
_
1
2
+
k

j=1
cos(j(x u))
_
_
=
1

f(u)D
k
(x u) du
We now use the fact that f and D
k
are periodic and make change of variables t = x u to obtain
S
k
f(f) =
1

f(x t)D
k
(t) dt.
Using Lemma 2 we have
(
n
f)(x) =
1
n + 1
n

k=0
(S
k
f)(x) =
1
n + 1
n

k=0
1

f(x t)D
k
(t) dt
=
1

f(x t)K
n
(t) dt.
Lemma 4. For n = 0, 1, 2, 3, , we have
(i) K
n
(t) 0, for all t R,
(ii)
1

K
n
(x) dx = 1,
(iii) lim
n
_

|K
n
(x)| dx = 0 for any 0 < < .
Proof. (i) follows from
K
n
(x) =
2
n + 1
_
_
sin
_
n+1
2
_
x
2 sin
_
x
2
_
_
_
2
To prove (ii), note that
_

K
(
x) dx =
_

_
1
2
+
n

k=1
_
1
k
n + 1
_
cos(kx)
_
dx = .
To prove (iii), note that if 0 < < t < , then sin(/2) < sin(t/2) and using Lemma 2 we get
_

|K
n
(x) dx
2
n + 1
_

_
_
sin
_
n+1
2
_
x
2 sin
_

2
_
_
_
2
dt

2(n + 1)
1
sin
2
_

2
_
and then take limit as n .
Theorem 3. (Fejer) Suppose f is periodic on R and integrable on [, ].
1) If
L = lim
h0
f(x
0
+h) +f(x
0
h)
2
exists for some x
0
R, then lim
n
(
n
f)(x
0
) = L.
2) If f is continuous on some closed interval [a, b], then
n
f f uniformly on [a, b]
Proof. Since f is periodic, we may assume that x
0
[, ]. Fix n 1. By Lemmas 2 and 3 and change
of variables, we have
(
n
f)(x
0
) L =
1

K
n
(t)[f(x
0
t) L]dt =
2

_

0
K
n
(t)
_
f(x
0
+t) +f(x
0
t)
2
L
_
dt
=
2

_

0
K
n
(t)F(x
0
, t)dt
where
F(x
0
, t) =
f(x
0
+t) +f(x
0
t)
2
L.
Let > 0. By denition of L, we can choose > 0 with < such that if |t| < , then |F(x
0
, t)| < /3.
Using Lemma 3, we get

_

0
K
n
(t)F(x
0
, t)dt

2
3
_

0
|K
n
(t)| dt <
2
3
.
Let M = sup
x
|f(x)|. Then |F(x
0
, t)| M. Using the third equation of Lemma 3, we can choose N
1
such that for all n N
1
,

K
n
(t) dt

<

3M
.
Thus, we have

K
n
(t)F(x
0
, t)dt

M
_

0
|K
n
(t)| dt <

3
.
Therefore for n N
1
, we have
|(
n
f)(x
0
) L|
2

_

0
|K
n
(t)F(x
0
, t)| +
2

|K
n
(t)F(x
0
, t)|
2
3
+

3
= .
We proved the very denition of (i).
To prove (ii), we note that if f is continuous on [, ], then it is uniformly continuous on [, ]. The
above inequalities are valid if we replace x
0
by any x [, ]. (You should cary out the details.)
Corollary 1. If f is continuous and periodic on R, then

n
f f uniformly on R
Proof. Since f is periodic, we may assume that f is continuous on , ] and apply Fejers Theorem.
Corollary 2. (Completeness) If f is continuous and periodic on R, and if a
k
(f) = 0 and b
k
(f) = 0 for
all k = 0, 1, 2, 3, , then f(x) = 0 for all x R
Proof. From the assumption we have
n
f(x) = 0 for all x. By Corollary 1, we have f(x) = lim
n
(
n
f)(x) =
0.
Corollary 3. If f is continuous and periodic on R, then there is a sequence of trigonometric polynomials
T
1
, T
2
, such that
T
n
f uniformly on R
Proof S
n
f is a trig polynomial implies
n
f is a trig polynomial. Take T
n
to be
n
f and apply Fejers
Theorem.
Theorem 4.(Weierstrass Approximation Theorem) Let f be continuous on a closed and bounded
interval [a, b]. Given > 0, there exists a polynomial
P(x) =
n

k=0
p
k
x
k
,
where p
k
R such that for all x [a, b],
|f(x) P(x)| < .
12.3 Growth of Fourier Coecients
Lemma 5. If f is integrable on [, ], then for n = 0, 1, 2, 3, ,
1

f(x)(S
n
f)(x) dx =
|a
0
(f)|
2
2
+
n

k=1
_
|a
k
(f)|
2
+ |b
k
(f)|
2
_
=
1

|(S
n
f)(x)|
2
dx
Theorem 5. (Bessels Inequality) If f is Riemann integrable on [, ], then

k=1
|a
k
(f)|
2
and

k=1
|b
k
(f)|
2
are both convergent. Moreover,
|a
0
(f)|
2
2
+
n

k=1
_
|a
k
(f)|
2
+ |b
k
(f)|
2
_

1

|f(x)|
2
dx
Corollary (Riemann - Lebesgue Lemma) If f is Riemann integrable on [, ], then
lim
k
a
k
(f) = lim
k
b
k
(f) = 0.
Lemma 6. If f is Riemann integrable on [, ] and
T
n
=
c
0
2
+
n

k=1
_
c
k
cos(kx) + d
k
sin(kx)
_
is any trigonometric polynomial of degree n, then
_

|f(x) (S
n
f)(x)|
2
dx
_

|f(x) T
n
(x)|
2
dx
Theorem 6. (Parsevals Identity) If f is periodic and continuous on R, then
|a
0
(f)|
2
2
+
n

k=1
_
|a
k
(f)|
2
+ |b
k
(f)|
2
_
=
1

|f(x)|
2
dx
Theorem 7. (Riemann - Lebesgue Lemma) If f
(j)
exists and is Riemann integrable on [, ] and if
f
(l)
is periodic for 1 l < j, then
lim
k
k
j
a
k
(f) = lim
k
k
j
b
k
(f) = 0.
12.4 A Digression:
Functions of Bounded Variation
Denition Let : [a, b] R be a function and let P = {x
0
, x
1
, x
n
} be a partition of [a, b]. Dene
V (, P) =
n

k=1
|(x
k
) (x
k1
)| .
The total variation of on [a, b] is dened by
V ar() = sup{V (, P)|P is a partition of [a, b]}
A function is said to be of bounded variation if V () < .
Lemma 7. If C
1
[a, b], then is of bounded variation on [a, b].
Proof: Let P = {x
0
, x
1
, x
n
} be a partition of [a, b]. Since

is continuous on [a, b], by Extreme Value

Thereom, there exits M such that
|

(x)| M for all x [a, b].

On the other hand, by Mean Value Theorem, there exists c
k
[x
k1
, x
k
] such that
(x
k
) (x
k1
) =

(c
k
)(x
k
x
k1
).
Adding these, using the previous inequality, and telescoping, we see that
V (, P) =
n

k=1
|(x
k
) (x
k1
)| M(b a).
Taking the sup over all partitions P we see that
V ar() M(b a).
Example Let (x) = x
2
sin(1/x). Show that
a) is of bounded variation on [0, 1].
b) C
1
[0, 1]
Solution. a) Consider a partition P = {x
0
, x
1
, , x
n
} of [0, 1].
Choose n a large positive integer so that the values of x
k
that are close to zero are contained in the partition
Q = {0/n, 1/n, 1/n 1, , 1}
V ar(, Q) =
n

k=1
x
2
k
sin(1/x
k
) x
2
k1
sin(1/x
k1
)
n

k=1
_
x
2
k
+x
2
k1
_
2
n

j=1
1
k
2
2 + 2
n1

k=1
_
1
k

1
k + 1
_
= 4
2
n
4.
Thus V (, P) V (, Q) < 2 and taking the sup we see that V ar() < .
b) But note that for x = 0,

(x) = 2xsin(1/x) cos(1/x)

and hence
lim
x0

(x)
does not exist while

(0) = lim
h
h
2
sin(1/h)
h
= lim
h
hsin(1/h) = 0.
Therefore C
1
[0, 1].
Example Let (x) = x
2
sin(1/x
2
). Show that is not of bounded variation on [0, 1].
Lemma 8 If is monotone on [a, b], then is of bounded variation on [a, b].
Proof: Suppose is increasing and let P = {x
0
, x
1
, , x
n
} be a partition of [a, b]. Then
n

k=1
|(x
k
) (x
k1
)| =
n

k=1
((x
k
) (x
k1
)) = (x
n
) (x
0
) = (b) (a)
Since M = (b) (a) is nite, we see that the sup over all partitions P is also nite. Hence is of bounded
variation.
Lemma 9. If is of bounded variation on [a, b], then is bounded on [a, b].
Proof: For any x [a, b], we have
|(x) (a)| |(x) (a)| +|(b) (x)| V ar().
Thus
|(x) |(x) (a)| +|(a)| V ar() +|(a)|
and hence is bounded.
Example The function
(x) =
_
sin(1/x), x = 0
0, x = 0
is bounded (by 1).
But it is not of bounded variation.For if
x
j
=
_
0, x = 0
2
(nj)
, 0 < j < n 1.
then, as n ,
n

j=1
|(x
j
) (x
j1
)| = 2n .
Thus is not of bounded variation on [0, 2/].
Theorem 8. If and are functions of bounded variation on [a, b], then so are , and . If
there exists
0
> 0 such that (x)
0
, then / is also of bounded variation.
Proof: Let P = {x
0
, x
1
, x
2
, , x
n
} be a partition of [a, b]. Then
n

k=1
|((x
j
) (x
j
)) ((x
j1
) (x
j1
))|
n

k=1
|(x
j
) (x
j1
)| +
n

k=1
|(x
j
) (x
j1
)|
V ar() +V ar()
Therefore,
V ar( ) V ar() +V ar().
By Lemma 8, there are constants M
1
and M
2
such that
|(x)| M
1
and |(x)| M
2
for all x [a, b].
But then
n

j=1
|(x
j
)(x
j
) (x
j1
)(x
j1
)|
=
n

k=1
|(x
j
)(x
j
)) (x
j1
)(x
j
) +(x
j1)
(x
j
) (x
j1
)(x
j1
)|
M
2
n

k=1
|(x
j
) (x
j1
)| +M
1
n

k=1
|(x
j
) (x
j1
)|
M
2
V ar() +M
1
V ar()
Therefore,
V ar() M
2
V ar() +M
1
V ar().
To prove the / is also of bounded variation, we write
n

j=1

(x
j
)
(x
j

(x
j1
)
(x
j1
)

=
n

j=1

(x
j
)(x
j
) (x
j1
)(x
j1
)
(x
j
)(x
j1
)

2
_
M
2
n

k=1
|(x
j
) (x
j1
)| +M
1
n

k=1
|(x
j
) (x
j1
)|
_
Therefore,
V ar
_

M
2

2
0
V ar() +
M
1

2
0
V ar().
Denition. Let be of bounded variation on [a, b]. The total variation of is the function dened by
(x) = sup{
n

k=1
|(x
j
) (x
j1
)|}
where the sup is over all partitions P = {x
0
, x
1
, , x
n
} of [a, x].
Theorem 9. Let be of bounded variation and be its total variation. Then
(i) |(y) (x)| (y) (x) for all a x y b
(ii) and are increasing on [a, b]
(iii) V ar() V ar().
Proof: (i) Let x < y and let P = {x
0
, x
1
, , x
n
} be a partition of [a, x]. Then Q = {x
0
, x
1
, , x
n
, y}is
a partition of [a, y]. By denition of we have
n

j=1
|(x
j
) (x
j1
)|

j=1
|(x
j
) (x
j1
)| + |(y) (x)
(y)
Taking the sup over all such partitions P of [a, x] we see that
(x) (x) |(y) (x) (y)
and (i) follows.
(ii) Since is dened as the sup, it is clearly increasing.
By part (i), we have
(y) (x) |(y) (x)| (y) (x)
and hence (x) (x) (y) (y). Therefore, is also increasing.
(iii) Let P = {x
0
, x
1
, , x
n
} be a partition of [a, b]. By part (i) and the denition of , we have
n

k=1
|(x
k
) (x
k1
)|

)k = 1
n
|(x
k
) (x
k1
)|
V ar()
Taking the sup over all such P we get (iii).
Corollary is of bounded variation on [a, b] if and only if there exists increasing functions f and g on
[a, b] such that
(x) = f(x) g(x), for all x [a, b].
Proof: If is of bounded variation, let be its total variation. Then by Theorem 9, the functions f =
and g = are increasing and = f g.
Conversely, if f and g are monotone, then both are of bounded variation by Lemma 9. But then by Thereom
8 = f g is of bounded variation.
Remarks. 1) If f is monotone on [a, b], then the set points x in [a, b] at which f is discontinuous is at
most countable.
Thus if is of bounded variation on [a, b], then it has at most a countable set of discontinuity on [a, b].
2) If f is monotone, then for any x
0
(a, b], the limit lim
xx

0
f(x) exists. This limit is denoted by f(x
0
).
Similarly, for any x
0
[a, b), the limit lim
xx
+
0
f(x) exists. This limit is denoted by f(x
0
+).
Thus if is of bounded variation on [a, b], then the limits lim
xx
+
0
(x) and lim
xx
+
0
(x) both exist for all
x
0
(a, b).
3) Monotone functions are Riemann integrable over [a, b].
Thus, if is of bounded variation on [a, b], then is Riemann integrable. .
12.5 Convergence of Fourier Series
Lemma 10. If

k=0
a
k
converges to L, then it is Cesaro summable to L.
Proof: Let > 0. Choose N
1
such that if k N
1
then |s
k
L| <

2
. Use the Archimedean Property to
choose N
2
> N
1
such that

N
1
k=0
|s
k
L| <
N
2
2
. If n > N
2
, then
|
n
L|
1
n + 1
N
1

k=0
|s
k
L| +
1
n + 1
n

k=N
1
+1
|s
k
L|

N
2
2(n + 1)
+

2
_
n N
2
n + 1
_
<

2
+

2
= .
Theorem 10. (Tauberian Theorem) Let a
k
0 and let L R. If

k=0
a
k
= L (C, 1), then ;

k=0
a
k
= L.
In other words, if a series of nonnegative terms is Cesaro summable to L, then it converges to L.
Proof: If we show the series converges, then by Lemma 8, we know that it must converge to L. Thus we
need only show that

k=0
a
k
< . Suppose to the contrary that

k=0
a
k
= . Then given M > 0, there
exists n
0
> 1 such that if n n
0
, then s
n
=

n
k=0
a
k
M. Let n n
0
. Then

n
=
s
0
+s
1
+s
2
+ +s
n
n + 1
=
s
0
+s
1
+ +s
n
0
n + 1
+
s
n
0
+1
+s
n
0
+2
+ +s
n
n + 1
0 +
n n
0
n + 1
M
If we take the limit as n , we see that L M for all M > 0. This is a contradiction. ( Take M = L+10)
Corollary Let f be periodic on R and Riemann integrable on [, ]. If a
k
(f) = 0 and b
k
(f) 0 for all
k 1, then

k=1
b
k
(f)
k
< .
Proof: Assume a
0
(f) = 0. Otherwise take g(x) = f(x) a
0
(f). Let
F(x) =
_
x
0
f(t)dt.
Then F is continuous and periodic (note that a
0
(f) = 0) on R. Hence by Fejers Theorem (
n
F)(0) F(0) =
0 as n . Integrating by parts we get
a
k
(F) =
b
k
(f)
k
0 and b
k
(F) =
a
k
(f)
k
= 0.
Hence

k=1
b
k
(f)
k
Cesaro summable. Since the terms are nonnegative, the corollary follows from Taubers Theorem.
Theorem 11. (Hardy) Let E R and suppose the {f
k
} is a sequence of functions on E that satises
|kf
k
(x)| M
for all x E and all k N, and some M > 0. If

k=0
f
k
is uniformly Cesaro summable to a function f on E,
then

k=0
f
k
converges uniformly to f on E.
Proof: Let x E and assume, without loss of generality, that M 1. For each n = 0, 1, 2, , set
s
n
(x) =
n

k=0
f
k
(x)
and

n
(x) =
s
0
(x) +s
1
(x) +s
2
(x) + +s
n
(x)
n + 1
.
Consider (the delayed average) dened for n, k 0 by

n,k
(x) =
s
n
(x) +s
n+1
(x) + +s
n+k
(x)
k + 1
.
Let 0 < < 1. For each n choose k = k(n) such that
k + 1
n
2M
< k + 2.
But then
n 1
k + 1
<
n
k + 1
<
2M

< .
Note also that

n,k
(x) s
n
(x) =
(s
n
(x) s
n
(x)) + (s
n+1
(x) s
n
(x)) + + (s
n+k
(x) s
n
(x))
k + 1
=
k+n

j=n
_
1
j n
k + 1
_
f
j
(x).
By assumption k|f
k
(x)| M and by choice of k = k(n), we have
|
n,k
(x) s
n
(x)|
n+k

j=n+1
|f
j
(x)|
M
n+k

j=n+1
1
j
<
M(k + 1)
n + 1
<

2
Since
n
f uniformly on E, we choose N so that for all n N and for all x E,
|
n
(x) f(x)| <

2
12M
.
Since

n,k
=
_
1 +
n 1
k + 1
_

n+k

_
n 1
k + 1
_

n1
,
it follows that
|s
n
(x) f(x)| |s
n
(x)
n,k
(x)| +|
n,k
(x) f(x)|

2
+
_
1 +
n 1
k + 1
_
|
n+k
(x) f(x)| +
_
n 1
k + 1
_
|
n1
(x) f(x)|

2
+
_
1 +
n 1
k + 1
_
_

2
12M
_
+
2M

_

2
12M
_
=

2
+

2
12M
+

3
<

2
+

12
+

3
< .
Therefore,

k=0
f
k
f uniformly on E.
Theorem 12. (Dirichlet - Jordan) If f is periodic on R and continuous on some closed interval [a, b],
then
S
n
f f uniformly on ; [a, b]
Remark For the uniqueness question posed earlier we have the following theorems, whose proofs can
be found on pages 536 of WilliamR. Wades An Introduction to Analysis 3rd edition, published by Prentice Hall.
Theorem 13. (Cantor - Lebesgue Lemma) If
S =
a
0
2
+

k=1
_
a
k
cos(kx) + b
k
sin(kx)
_
is a trigonometric series that converges pointwise on some interval [a, b], then
lim
k
a
k
= 0 and lim
k
b
k
= 0.
Theorem 14. ( Cantor) Suppose
S =
a
0
2
+

k=1
_
a
k
cos(kx) + b
k
sin(kx)
_
is a trigonometric series that converges pointwise on [, ] to a periodic continuous function f. Then S is
the Fourier series of f, that is
a
k
= a
k
(f) =
1

f(x) cos(kx) dx, k = 0, 1, 2, 3,

b
k
= b
k
(f) =
1

f(x) sin(kx) dx, k = 1, 2, 3, .