The Weierstrass M Test

The Weierstrass M test.
The following is a standard result and in many cases the easiest and most
natural method to show a series is uniformly convergent.
Theorem 1 (Weierstrass M test.). Let f1 , f2 , f3 , . . . X R be a sequence of
functions form a set X to the real numbers. Assume that there are constants,
Mk , such that the two conditions
|fk (x)| Mk
holds for all x X and k 1
and
Mk <
k=1
hold. Then the series
fk (x)
k=1
converges absolutely and uniformly on X.

P
Problem 1. Prove this. Hint: First note that
k=1 fk (x) converges absolutely
on
X
just
by
comparison
with
the
series
of non-negative terms
P
M
.
Let
k
k=1
X
F (x) =
fk (x).
k=1
Pn
So we only need show the partial sums Sn (x) =
k=1 fk (x) converge
uniformly to F (x). Note

X

X
X

|F (x) Sn (x)| =
fk (x)
|fk (x)|
Mk

k=n+1
k=n+1
and you should be able to take it from here.
k=n+1
Here is an example of the use of the M -test. Define a function on R by

f (x) =
X
k=1
1
.
x2 + k 2
The the k-th term satisfies

x2
1
1
2 =: Mk .
2
+k
k
P
P
2
The series
k=1 Mk =
k=1 1/k converges, there by the M -test the series
for f (x) converges uniformly. Moreover we can say more. As each term of the
series is continuous and the convergence is uniform the sum function is also
continuous. (As the uniform limit of continuous functions is continuous.)
Problem 2. Define a function on the plane R2 by
X
1
p
f (x, y) =
.
3/2
k + x2 + y 4
k=1
Show this series converges uniformly and that f is continuous on R2 .
Trigonometric functions.
We can now give definitions of the trigonometric functions. It is enough
to define sin and cos as all the others can be defined in terms of these two.
Theorem 2. The two series
X
x2 x4 x6 x8 x10
(1)k x2k
=1
+
+
c(x) =
(2k)!
2
4!
6!
8!
10!
s(x) =
k=0
X
k=0
x3 x5 x7 x9 x11
(1)k x2k+1
=x
+
+
(2k + 1)!
3!
5!
7!
9!
11!
converge absolutely for all x R and therefore these series are absolutely
convergent and differentiable for all x R. The derivatives satisfy
c0 (x) = s(x),
s0 (x) = c(x).
The values at x = 0 are

c(0) = 1,
s(0) = 0.
Also
c00 (x) = c(x),
s00 (x) = s(x).
Problem 3. Prove this.
Proposition 3. These functions satisfy

c(x)2 + s(x)2 = 1.
Problem 4. Prove this. Hint: Show that c(x)2 +s(x)2 is constant by taking
its derivative. Note that showing it is constant does not finish the problem,
you still have to show the constant is 1.

Lemma 4. If g is two times differentiable on R and
g 00 = g,
g(0) = 0,
g 0 (0) = 0
then g(x) = 0 for all x.

Problem 5. Prove this. Hint: Let E = g 2 + (g 0 )2 and show E 0 = 0.
Theorem 5. If f is twice differentiable on R and
f 00 = f
then f is a linear combination of c and s. In particular
f (x) = f (0)c(x) + f 0 (0)s(x).
Problem 6. Prove this. Hint: Let g(x) = f (x) f (0)c(x) f 0 (0)s(x) and
use Lemma 4.

Theorem 6. The functions c and s satisfy
c(x + a) = c(a)c(x) s(a)s(x)
s(x + a) = s(a)c(x) + c(a)s(x).
Problem 7. Prove this. Hint: For the first one let f (x) = c(x + a). Then
f 00 (x) = f (x). Thus, by Theorem 5,
f (x) = f (0)c(x) + f 0 (0)s(x).
Lemma 7. If for 0 < x < 6 the inequality

x3
x2
s(x) > x
=x 1
6
6
holds. In particular s(x) > 0 for 0 < x < 3.

Proof. For any x we have
x3 x5 x7 x9 x11 x13 x15
+
+
3!
5! 7!
9! 11! 13!
15!

x2
x9
x2
x13
x2
x3 x5
+
1
+
1
+
1
+
=x
3!
5!
67
9!
10 11
13!
14 15
s(x) = x
If 0 < x < 6 then x2 < 6 7 < 10 11 < 14 15. Therefore all the terms

x5
x2
x9
x2
x13
x2
1
,
1
,
1
,...
5!
67
9!
10 11
13!
14 15
are positive and the result follows.
Lemma 8. If 0 < x < 7 the inequality

c(x) < 1
x2 x4
+
2
24
holds.
Theorem 9. The function c(x) has a unique zero in the interval [0, 2]. We
denote this zero by /2. This is our official definition of the number .
Proof. As 2 < 6 we have, by Lemma 7, that s(x) > 0 on the interval (0, 2).
Therefore when 0 < x < 2
c0 (x) = s(x) < 0.
This shows that c(x) is strictly decreasing on [0, 2]. Thus c(x) can have at
most one zero on [0, 2]. But
c(0) = 1 > 0
and by Lemma 8
22
24
1
+
=
<0
2
24
3
and therefore c(x) has at least one root in [0, 2] by the Intermediate Value
Theorem.

c(2) < 1
Proposition 10. The following hold

c(/2) = 0
s(/2) = 1
c() = 1
s() = 0
c(2) = 1
s(2) = 0
Problem 9. Prove this. Hint: That c(/2) = 0 is the definition of . Then

c(/2)2 + s(/2)2 = 1 implies s(/2) = 1. Use Lemma 7 to rule out
s(/2) = 1. The rest should now follow from Theorem 6.

Theorem 11. The following hold.
c(x + /2) = s(x)
s(x + /2) = c(x)
c(x + ) = c(x)
s(x + ) = s(x)
c(x + 2) = c(x)
s(x + 2) = s(x)
Definition 12. Our official definition of cos and sin is

cos(x) = c(x),
sin(x) = s(x)
where c and s are as in Theorem 2. Then tan(x) is defined by

tan(x) =
sin(x)
cos(x)
with the usual formulas for sec(x) etc.
Proposition 13. The tangent satisfies

d
tan(x) = 1 + tan2 (x).
dx
Also the restriction tan : (/2, /2) R is a bijection. Let arctan : R
(/2, /2) be the inverse of this restriction of tan. Then
tan(x + ) = tan(x),
d
1
arctan(x) =
.
dx
1 + x2
Remark 14. To compute we can use the series for the arctan:
arctan(x) = x
x3 x5 x7
+
+ .
3
5
7
For this to be efficient we wish to use values of x that are close to zero. In
1796 John Machin showed1
1
1
.
= 16 arctan 4 arctan
5
239
Using this and the series for arctan(x) gives
X
X
(1)k
(1)k
= 16
4
(2k + 1)52k+1
(2k + 1)(239)2k+1
k=0
k=0
allows one to compute to five or six decimals without much trouble. Just
using the first five terms in the series gives
3.14159268240440
while the correct value to 15 significant digits is
= 3.14159265358979 . . .
so we are already good to seven decimals. Using nine terms in the series
gives you 15 significant digits.
For a less off the wall identity note that if 1 = arctan(1/2) and 2 =
arctan(1/3), so that tan 1 = 1/2 and tan 2 , then using the addition for tan
we have
1/2 + 1/3
tan(1 ) + tan(2 )
=
=1
tan(1 + 2 ) =
1 tan 1 tan 2
1 (1/2)(1/3)
and therefore
1 + 2 = arctan(1) = .
4
This gives

X
1
(1)k
1
1
1
=4
+
.
= 4 arctan + arctan
2
3
2k + 1 22k+1 32k+1
k=0
Using ten terms in this series gives the approximation

3.14159257960635
(correct to 7 decimals) which is good enough for any piratical application.
To get 15 significant digits using 22 terms in this series is enough.
For a modern method there is the formula found in 1995 by Simon Plouffe:

X
1
4
2
1
1
=
.
16k 8k + 1 8k + 4 8k + 5 8k + 6
k=0
A thousand terms of this gives well over a thousand decimal places.
1If you wish to prove this, probably the easiest way is to notice that (5 + i)4 (239 i) =
114244(1 + i) and use the polar form of complex numbers to get the result.

The Weierstrass M Test

Uploaded by

Copyright:

Available Formats

The Weierstrass M Test

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Weierstrass M Test

Uploaded by

Copyright:

Available Formats

The Weierstrass M test.

holds for all x X and k 1

hold. Then the series

converges absolutely and uniformly on X.

and you should be able to take it from here.

Here is an example of the use of the M -test. Define a function on R by

The the k-th term satisfies

Problem 2. Define a function on the plane R2 by

The values at x = 0 are

s00 (x) = s(x).

Problem 3. Prove this.

Proposition 3. These functions satisfy

then g(x) = 0 for all x.

Lemma 7. If for 0 < x < 6 the inequality

holds. In particular s(x) > 0 for 0 < x < 3.

Lemma 8. If 0 < x < 7 the inequality

Proposition 10. The following hold

Problem 9. Prove this. Hint: That c(/2) = 0 is the definition of . Then

s(x + /2) = c(x)

Problem 10. Prove this.

Definition 12. Our official definition of cos and sin is

where c and s are as in Theorem 2. Then tan(x) is defined by

with the usual formulas for sec(x) etc.

Proposition 13. The tangent satisfies

Using ten terms in this series gives the approximation

A thousand terms of this gives well over a thousand decimal places.

You might also like