Appendex c

November 4, 2004 15:00 k34-appc Sheet number 1 Page number 22 cyan magenta yellow black
A22
a p p e n d i x c
SELECTED PROOFS
PROOFS OF BASIC LIMIT THEOREMS
An extensive excursion into proofs of limit theorems would be too time consuming to
undertake, so we have selected a few proofs of results from Section 2.2 that illustrate some
of the basic ideas.
C.1 theorem. Let a be any real number, let k be a constant, and suppose that
lim
x →a
f(x) = L1 and that lim
x →a
g(x) = L2. Then
(a) lim
x →a
k = k
(b) lim
x →a
[f(x) + g(x)] = lim
x →a
f(x) + lim
x →a
g(x) = L1 + L2
(c) lim
x →a
[f(x)g(x)] = lim
x →a
f(x) lim
x →a
g(x) = L1L2
proof (a). We will apply Definition 2.4.1 with f(x) = k and L = k. Thus, given > 0,
we must find a number δ > 0 such that
|k − k| < if 0 < |x − a| < δ
or, equivalently,
0 < if 0 < |x − a| < δ
But the condition on the left side of this statement is always true, no matter how δ is chosen.
Thus, any positive value for δ will suffice.
proof (b). We must show that given > 0 we can find a number δ > 0 such that
|(f(x) + g(x)) − (L1 + L2)| < if 0 < |x − a| < δ (1)
However, from the limits of f and g in the hypothesis of the theorem we can find numbers
δ1 and δ2 such that
|f(x) − L1| < /2 if 0 < |x − a| < δ1
|g(x) − L2| < /2 if 0 < |x − a| < δ2
Moreover, the inequalities on the left sides of these statements both hold if we replace δ1
and δ2 by any positive number δ that is less than both δ1 and δ2. Thus, for any such δ it
follows that
|f(x) − L1| + |g(x) − L2| < if 0 < |x − a| < δ (2)
However, it follows from the triangle inequality [Theorem E.5 of Appendix E] that
|(f(x) + g(x)) − (L1 + L2)| = |(f(x) − L1) + (g(x) − L2)|
≤ |f(x) − L1| + |g(x) − L2|
so that (1) follows from (2).

Appendix C: Selected Proofs A23
proof (c). We must show that given > 0 we can find a number δ > 0 such that
|f(x)g(x) − L1L2| < if 0 < |x − a| < δ (3)
To find δ it will be helpful to express (3) in a different form. If we rewrite f(x) and g(x) as
f(x) = L1 + (f(x) − L1) and g(x) = L2 + (g(x) − L2)
then the inequality on the left side of (3) can be expressed as (verify)
|L1(g(x) − L2) + L2(f(x) − L1) + (f(x) − L1)(g(x) − L2)| < (4)
Since
lim
x →a
f(x) = L1 and lim
x →a
g(x) = L2
we can find positive numbers δ1, δ2, δ3, and δ4 such that
|f(x) − L1| <
√
/3 if 0 < |x − a| < δ1
|f(x) − L1| <
3(1 + |L2|)
if 0 < |x − a| < δ2
|g(x) − L2| <
√
/3 if 0 < |x − a| < δ3
|g(x) − L2| <
3(1 + |L1|)
if 0 < |x − a| < δ4
(5)
Moreover, the inequalities on the left sides of these four statements all hold if we replace
δ1, δ2, δ3, and δ4 by any positive number δ that is smaller than δ1, δ2, δ3, and δ4. Thus, for
any such δ it follows with the help of the triangle inequality that
|L1(g(x) − L2) + L2(f(x) − L1) + (f(x) − L1)(g(x) − L2)|
≤ |L1(g(x) − L2)| + |L2(f(x) − L1)| + |(f(x) − L1)(g(x) − L2)|
= |L1||g(x) − L2| + |L2||f(x) − L1| + |f(x) − L1||g(x) − L2|
< |L1|
3(1 + |L1|)
+ |L2|
3(1 + |L2|)
+ /3 /3 From (5)
=
3
|L1|
1 + |L1|
+
3
|L2|
1 + |L2|
+
3
<
3
+
3
+
3
= Since
|L1|
1 + |L1|
< 1 and
|L2|
1 + |L2|
< 1
which shows that (4) holds for the δ selected. I
Do not be alarmed if the proof of part
(c) seems difficult; it takes some expe-
rience with proofs of this type to de-
velop a feel for choosing a valid δ. Your
initial goal should be to understand the
ideas and the computations.
PROOF OF A BASIC CONTINUITY PROPERTY
Next we will prove Theorem 2.5.5 for two-sided limits.
C.2 theorem (Theorem 2.5.5). If limx →c g(x) = L and if the function f is con-
tinuous at L, then limx →c f(g(x)) = f(L). That is,
lim
x →c
f(g(x)) = f lim
x →c
g(x)
proof. We must show that given > 0, we can find a number δ > 0 such that
|f(g(x)) − f(L)| < if 0 < |x − c| < δ (6)
Since f is continuous at L, we have
lim
u→L
f(u) = f(L)

A24 Appendix C: Selected Proofs
and hence we can find a number δ1 > 0 such that
|f(u) − f(L)| < if |u − L| < δ1
In particular, if u = g(x), then
|f(g(x)) − f(L)| < if |g(x) − L| < δ1 (7)
But limx →c g(x) = L, and hence there is a number δ > 0 such that
|g(x) − L| < δ1 if 0 < |x − c| < δ (8)
Thus, if x satisfies the condition on the right side of statement (8), then it follows that g(x)
satisfies the condition on the right side of statement (7), and this implies that the condition
on the left side of statement (6) is satisfied, completing the proof. I
PROOF OF THE CHAIN RULE
Next we will prove the chain rule (Theorem 3.6.1), but first we need a preliminary result.
C.3 theorem. If f is differentiable at x and if y = f(x), then
y = f (x) x + x
where →0 as x →x and = 0 if x = 0.
proof. Define
=



f(x + x) − f(x)
x
− f (x) if x = 0
0 if x = 0
(9)
If x = 0, it follows from (9) that
x = [f(x + x) − f (x)] − f (x) x (10)
But
y = f(x + x) − f(x) (11)
so (10) can be written as
x = y − f (x) x
or
y = f (x) x + x (12)
If x = 0, then (12) still holds, (why?), so (12) is valid for all values of x. It remains to
show that →0 as x →0. But this follows from the assumption that f is differentiable
at x, since
lim
x →0
= lim
x →0
f(x + x) − f(x)
x
− f (x) = f (x) − f (x) = 0 I
We are now ready to prove the chain rule.
C.4 theorem (Theorem 3.6.1). If g is differentiable at the point x and f is differ-
entiable at the point g(x), then the composition f ◦g is differentiable at the point x.
Moreover, if y = f(g(x)) and u = g(x), then
dy
dx
=
dy
du
·
du
dx

proof. Since g is differentiable at x and u = g(x), it follows from Theorem C.3 that
(u) = g (x) x + 1 x (13)
where 1 →0 as x →0. And since y = f(u) is differentiable at u = g(x), it follows from
Theorem C.3 that
y = f (u) u + 2 u (14)
where 2 →0 as u→0.
Factoring out the u in (14) and then substituting (13) yields
y = [f (u) + 2][g (x) x + 1 x]
or
y = [f (u) + 2][g (x) + 1] x
or if x = 0,
y
x
= [f (u) + 2][g (x) + 1] (15)
But (13) implies that u→0 as x →0, and hence 1 →0 and 2 →0 as x →0. Thus,
from (15)
lim
x →0
y
x
= f (u)g (x)
or
y
x
= f (u)g (x) =
dy
du
·
du
dx
I
PROOF THAT RELATIVE EXTREMA OCCUR AT CRITICAL POINTS
In this subsection we will prove Theorem 5.2.2, which states that the relative extrema of a
function occur at critical points.
C.5 theorem (Theorem 5.2.2). Suppose that f is a function defined on an open
interval containing the point x0. If f has a relative extremum at x = x0, then x = x0 is
a critical point of f ; that is, either f (x0) = 0 or f is not differentiable at x0.
proof. Suppose that f has a relative maximum at x0. There are two possibilities—either
f is differentiable at a point x0 or it is not. If it is not, then x0 is a critical point for f and we
are done. If f is differentiable at x0, then we must show that f (x0) = 0. We will do this
by showing that f (x0) ≥ 0 and f (x0) ≤ 0, from which it follows that f (x0) = 0. From
the definition of a derivative we have
f (x0) = lim
h→0
f(x0 + h) − f(x0)
h
so that
f (x0) = lim
h→0+
f(x0 + h) − f(x0)
h
(16)
and
f (x0) = lim
h→0−
f(x0 + h) − f(x0)
h
(17)
Because f has a relative maximum at x0, there is an open interval (a, b) containing x0 in
which f(x) ≤ f(x0) for all x in (a, b).
Assume that h is sufficiently small so that x0 + h lies in the interval (a, b). Thus,
f(x0 + h) ≤ f(x0) or equivalently f(x0 + h) − f(x0) ≤ 0
Thus, if h is negative,
f(x0 + h) − f(x0)
h
≥ 0 (18)

and if h is positive,
f(x0 + h) − f(x0)
h
≤ 0 (19)
But an expression that never assumes negative values cannot approach a negative limit and
an expression that never assumes positive values cannot approach a positive limit, so that
f (x0) = lim
h→0−
f(x0 + h) − f(x0)
h
≥ 0 From (17) and (18)
and
f (x0) = lim
h→0+
f(x0 + h) − f(x0)
h
≤ 0 From (16) and (19)
Since f (x0) ≥ 0 and f (x0) ≤ 0, it must be that f (x0) = 0.
A similar argument applies if f has a relative minimum at x0. I
PROOFS OF TWO SUMMATION FORMULAS
We will prove parts (a) and (b) of Theorem 6.4.2. The proof of part (c) is similar to that of
part (b) and is omitted.
C.6 theorem (Theorem 6.4.2).
(a)
n
k=1
k = 1 + 2 + · · · + n =
n(n + 1)
2
(b)
n
k=1
k2
= 12
+ 22
+ · · · + n2
=
n(n + 1)(2n + 1)
6
(c)
n
k=1
k3
= 13
+ 23
+ · · · + n3
=
n(n + 1)
2
2
proof (a). Writing n
k=1
k
two ways, with summands in increasing order and in decreasing order, and then adding, we
obtain
n
k=1
k = 1 + 2 + 3 + · · · + (n − 2) + (n − 1) + n
n
k=1
k = n + (n − 1) + (n − 2) + · · · + 3 + 2 + 1
2
n
k=1
k = (n + 1) + (n + 1) + (n + 1) + · · · + (n + 1) + (n + 1) + (n + 1)
= n(n + 1)
Thus,
n
k=1
k =
n(n + 1)
2

proof (b). Note that
(k + 1)3
− k3
= k3
+ 3k2
+ 3k + 1 − k3
= 3k2
+ 3k + 1
So,
n
k=1
[(k + 1)3
− k3
] =
n
k=1
(3k2
+ 3k + 1) (20)
Writing out the left side of (20) with the index running down from k = n to k = 1, we have
n
k=1
[(k + 1)3
− k3
] = [(n + 1)3
− n3
] + · · · + [43
− 33
] + [33
− 23
] + [23
− 13
]
= (n + 1)3
− 1 (21)
Combining (21) and (20), and expanding the right side of (20) by using Theorem 6.4.1 and
part (a) of this theorem yields
(n + 1)3
− 1 = 3
n
k=1
k2
+ 3
n
k=1
k +
n
k=1
1
= 3
n
k=1
k2
+ 3
n(n + 1)
2
+ n
So,
3
n
k=1
k2
= [(n + 1)3
− 1] − 3
n(n + 1)
2
− n
= (n + 1)3
− 3(n + 1)
n
2
− (n + 1)
=
n + 1
2
[2(n + 1)2
− 3n − 2]
=
n + 1
2
[2n2
+ n] =
n(n + 1)(2n + 1)
2
Thus,
n
k=1
k2
=
n(n + 1)(2n + 1)
6
The sum in (21) is an example of a tele-
scoping sum, since the cancellation of
each of the two parts of an interior
summand with parts of its neighbor-
ing summands allows the entire sum
to collapse like a telescope.
PROOF OF THE LIMIT COMPARISON TEST
C.7 theorem (Theorem 10.5.4). Let ak and bk be series with positive terms
and suppose that
ρ = lim
k →+ϱ
ak
bk
If ρ is ﬁnite and ρ > 0, then the series both converge or both diverge.
proof. We need only show that bk converges when ak converges and that bk
diverges when ak diverges, since the remaining cases are logical implications of these
(why?). The idea of the proof is to apply the comparison test to ak and suitable multiples
of bk. For this purpose let be any positive number. Since
ρ = lim
k →+ϱ
ak
bk

it follows that eventually the terms in the sequence {ak/bk} must be within units of ρ; that
is, there is a positive integer K such that for k ≥ K we have
ρ − <
ak
bk
< ρ +
In particular, if we take = ρ/2, then for k ≥ K we have
1
2
ρ <
ak
bk
<
3
2
ρ or
1
2
ρbk < ak <
3
2
ρbk
Thus, by the comparison test we can conclude that
ϱ
k=K
1
2
ρbk converges if
ϱ
k=K
ak converges (22)
ϱ
k=K
3
2
ρbk diverges if
ϱ
k=K
ak diverges (23)
But the convergence or divergence of a series is not affected by deleting ﬁnitely many terms
or by multiplying the general term by a nonzero constant, so (22) and (23) imply that
ϱ
k=1
bk converges if
ϱ
k=1
ak converges
ϱ
k=1
bk diverges if
ϱ
k=1
ak diverges I
PROOF OF THE RATIO TEST
C.8 theorem (Theorem 10.5.5). Let uk be a series with positive terms and
suppose that
ρ = lim
k →+ϱ
uk+1
uk
(a) If ρ < 1, the series converges.
(b) If ρ > 1 or ρ = +ϱ, the series diverges.
(c) If ρ = 1, the series may converge or diverge, so that another test must be tried.
proof (a). The number ρ must be nonnegative since it is the limit of uk+1/uk, which is
positive for all k. In this part of the proof we assume that ρ < 1, so that 0 ≤ ρ < 1.
We will prove convergence by showing that the terms of the given series are eventually
less than the terms of a convergent geometric series. For this purpose, choose any real
number r such that 0 < ρ < r < 1. Since the limit of uk+1/uk is ρ, and ρ < r, the terms
of the sequence {uk+1/uk} must eventually be less than r. Thus, there is a positive integer
K such that for k ≥ K we have
uk+1
uk
< r or uk+1 < ruk
This yields the inequalities
uK+1 < ruK
uK+2 < ruK+1 < r2
uK
uK+3 < ruK+2 < r3
uK
uK+4 < ruK+3 < r4
uK
...
(24)

But 0 < r < 1, so
ruK + r2
uK + r3
uK + · · ·
is a convergent geometric series. From the inequalities in (24) and the comparison test it
follows that
uK+1 + uK+2 + uK+3 + · · ·
must also be a convergent series. Thus, u1 + u2 + u3 + · · · + uk + · · · converges by The-
orem 10.4.3(c).
proof (b). In this part we will prove divergence by showing that the limit of the general
term is not zero. Since the limit of uk+1/uk is ρ and ρ > 1, the terms in the sequence
{uk+1/uk} must eventually be greater than 1. Thus, there is a positive integer K such that
for k ≥ K we have uk+1
uk
> 1 or uk+1 > uk
This yields the inequalities
uK+1 > uK
uK+2 > uK+1 > uK
uK+3 > uK+2 > uK
uK+4 > uK+3 > uK
...
(25)
Since uK > 0, it follows from the inequalities in (25) that limk →+ϱ uk = 0, and thus the
series u1 + u2 + · · · + uk + · · · diverges by part (a) of Theorem 10.4.1. The proof in the
case where ρ = +ϱ is omitted.
proof (c). The divergent harmonic series and the convergent p-series with p = 2 both
have ρ = 1 (verify), so the ratio test does not distinguish between convergence and diver-
gence when ρ = 1. I
PROOF OF THE REMAINDER ESTIMATION THEOREM
C.9 theorem (Theorem 10.7.4). If the function f can be differentiated n + 1 times
on an interval I containing the number x0, and if M is an upper bound for |f (n+1)
(x)|
on I, that is, |f (n+1)
(x)| ≤ M for all x in I, then
|Rn(x)| ≤
M
(n + 1)!
|x − x0|n+1
for all x in I.
proof. We are assuming that f can be differentiated n + 1 times on an interval I con-
taining the number x0 and that
|f (n+1)
(x)| ≤ M (26)
for all x in I. We want to show that
|Rn(x)| ≤
M
(n + 1)!
|x − x0|n+1
(27)
for all x in I, where
Rn(x) = f (x) −
n
k=0
f (k)
(x0)
k!
(x − x0)k
(28)

In our proof we will need the following two properties of Rn(x):
Rn(x0) = Rn(x0) = · · · = R(n)
n (x0) = 0 (29)
R(n+1)
n (x) = f (n+1)
(x) for all x in I (30)
These properties can be obtained by analyzing what happens if the expression for Rn(x)
in Formula (28) is differentiated j times and x0 is then substituted in that derivative. If
j < n, then the jth derivative of the summation in Formula (28) consists of a constant term
f (j)
(x0) plus terms involving powers of x − x0 (verify). Thus, R
(j)
n (x0) = 0 for j < n,
which proves all but the last equation in (29). For the last equation, observe that the nth
derivative of the summation in (28) is the constant f (n)
(x0), so R(n)
n (x0) = 0. Formula (30)
follows from the observation that the (n + 1)-st derivative of the summation in (28) is zero
(why?).
Now to the main part of the proof. For simplicity we will give the proof for the case
where x ≥ x0 and leave the case where x < x0 for the reader. It follows from (26) and (30)
that |R(n+1)
n (x)| ≤ M, and hence
−M ≤ R(n+1)
n (x) ≤ M
Thus,
x
x0
−M dt ≤
x
x0
R(n+1)
n (t) dt ≤
x
x0
M dt (31)
However, it follows from (29) that R(n)
n (x0) = 0, so
x
x0
R(n+1)
n (t) dt = R(n)
n (t)
x
x0
= R(n)
n (x)
Thus, performing the integrations in (31) we obtain the inequalities
−M(x − x0) ≤ R(n)
n (x) ≤ M(x − x0)
Now we will integrate again. Replacing x by t in these inequalities, integrating from x0 to
x, and using R(n−1)
n (x0) = 0 yields
−
M
2
(x − x0)2
≤ R(n−1)
n (x) ≤
M
2
(x − x0)2
If we keep repeating this process, then after n + 1 integrations we will obtain
−
M
(n + 1)!
(x − x0)n+1
≤ Rn(x) ≤
M
(n + 1)!
(x − x0)n+1
which we can rewrite as
|Rn(x)| ≤
M
(n + 1)!
(x − x0)n+1
This completes the proof of (27), since the absolute value signs can be omitted in that
formula when x ≥ x0 (which is the case we are considering). I
PROOF OF THE TWO-VARIABLE CHAIN RULE
C.10 theorem (Theorem 14.5.1). If x = x(t) and y = y(t) are differentiable at t,
and if z = f(x, y) is differentiable at the point (x(t), y(t)), then z = f(x(t), y(t)) is
differentiable at t and dz
dt
=
∂z
∂x
dx
dt
+
∂z
∂y
dy
dt

proof. Let x, y, and z denote the changes in x, y, and z, respectively, that corre-
spond to a change of t in t. Then
dz
dt
= lim
t →0
z
t
,
dx
dt
= lim
t →0
x
t
,
dy
dt
= lim
t →0
y
t
Since f(x, y) is differentiable at (x(t), y(t)), it follows from (5) in Section 14.4 that
z =
∂z
∂x
x +
∂z
∂y
y + ( x, y) x2 + y2 (32)
where the partial derivatives are evaluated at (x(t), y(t)) and where ( x, y) satisﬁes
( x, y)→0 as ( x, y)→(0, 0) and (0, 0) = 0. Dividing both sides of (32) by t
yields
z
t
=
∂z
∂x
x
t
+
∂z
∂y
y
t
+
( x, y) x2 + y2
t
(33)
Since
lim
t →0
x2 + y2
| t|
= lim
t →0
x
t
2
+
y
t
2
= lim
t →0
x
t
2
+ lim
t →0
y
t
2
=
dx
dt
2
+
dy
dt
2
we have
lim
t →0
( x, y) x2 + y2
t
= lim
t →0
| ( x, y)| x2 + y2
| t|
= lim
t →0
| ( x, y)| · lim
t →0
x2 + y2
| t|
= 0 ·
dx
dt
2
+
dy
dt
2
= 0
Therefore,
lim
t →0
( x, y) x2 + y2
t
= 0
Taking the limit as t →0 of both sides of (33) then yields the equation
dz
dt
=
∂z
∂x
dx
dt
+
∂z
∂y
dy
dt
I

Appendex c

More Related Content

What's hot (19)

Viewers also liked (20)

Similar to Appendex c (20)

More from swavicky (9)

Appendex c