MATH 131B, SPRING 2007

HOMEWORK 1 SOLUTIONS

Sec. 5.2, ex. 7: Let f : [0, 1] R be defined by

(
1,
if x [0, 1] Q
f (x) =
1, if x [0, 1] \ Q.
Then |f | = 1 is clearly continuous. Let us show that f is discontinuous at every point of [0, 1].
If x [0, 1] is rational, let (xn ) be a sequence of irrationals in [0, 1] converging to x. Such a
sequence exists because the set of irrational numbers is dense in R. Then
lim f (xn ) = 1 6= 1 = f (x).

By the characterization of continuity via sequences, it follows that f is discontinuous at x.

If y is irrational, let (yn ) be a sequence of rational numbers converging to y. Such a sequence
exists because the set of rational numbers is dense in R. Then
lim f (yn ) = 1 6= 1 = f (y).

Therefore, f is discontinuous at y, hence at every point of [0, 1].

5.3, ex. 6: Let


1
.
g(x) = f (x) f x +
2
It suffices to show that g(c) = 0, for some 0
c 1/2.
Since f is continuous, so is x 7 f x + 12 , as the composition of two continuous functions.
It follows that g is continuous on its domain [0, 12 ]. Furthermore, g(0) = f (0) f (1/2) and
 
 
 
1
1
1
g
=f
f (1) = f
f (0).
2
2
2


If g(0) = 0, then f (0) = f (1/2), so we can take c = 0. Otherwise, g(0) 6= 0, so g(0) and g(1/2)
are of opposite sign. By the Intermediate Value Theorem (IVT), it follows that there exists
c (0, 1/2) such that g(c) = 0.

5.3, ex. 17: Suppose f is not constant. Then it takes at least two distinct values, say u < v.
Since f is continuous, by the Intermediate Value Theorem, every number in (u, v) is a value of
f . But every open interval (u, v) contains both rational and irrational numbers, contradicting
the assumption that f takes only rational or only irrational values.

Sec. 6.1, ex. 4: By definition of f , we have
0 |f (x)| x2 ,
for all x R. Therefore,



 f (x) 
 |x|,
0 
x 
for all x 6= 0. By the Squeeze Theorem, |f (x)/x| 0, as x 0 and therefore, f (x)/x =
[f (x) f (0)]/(x 0) 0. Thus f is differentiable at 0 and f 0 (0) = 0.

1

Remark: It is sufficient to show that f (x)/x 0, as Q x 0 and Q 6 x 0 separately, but

you need to prove the sufficiency.
Sec. 6.1, ex. 13: Suppose first that f is differentiable at c. By definition, the limit
f (c + h) f (c)
(1)
h0
h
exists and equals f 0 (c). By the sequential characterization of limits, f 0 (c) equals the limit in
(1) along any sequence (hn ) such that hn 0, as n . Taking hn = 1/n gives the desired
conclusion.
To see that the existence of the limit of n[f (c + n1 ) f (c)] does not guarantee the existence
of f 0 (c), take
(
x sin 2
x , if x 6= 0
f (x) =
0,
if x = 0.
Then the limit of
n[f (1/n) f (0)] = sin 2n = 0
is 0, yet f is not differentiable at c = 0, since
lim

lim

x0

f (x) f (0)
2
= lim sin
x0
x
x

does not exist.

Sec. 6.1, ex. 16: Denote by f the restriction of tangent to the interval (/2, /2). Observe
that f (/2, /2) = R, since the limits of f at the endpoints are and +. Furthermore,
f is differentiable and
1
f 0 (x) =
> 0.
cos2 x
Therefore, f : (/2, /2) R has a differentiable inverse, f 1 = arctan : R (/2, /2),
whose derivative is

0
1
1
f 1 (y) = 0
= 1 = cos2 x,
f (x)
cos2 x
where x = f 1 (y), i.e., y = tan x. Using cos2 + sin2 = 1, we obtain
1
1
cos2 x
cos2 x
=
=
. 
=
2
2
2
1
1 + y2
1 + tan x
cos x + sin x
Sec. 6.2, ex. 4: Taking the derivative of
n
n
X
X
2
f (x) =
(ai x) =
(x ai )2 ,
cos2 x =

i=1

i=1

we obtain, for any x R,

f 0 (x) =

n
X

2(x ai ).

i=1

Solving f 0 (x) = 0, we obtain the only critical point of f

n

c=

1X
ai .
n
i=1

Since

f 0 (x)

= 2n(x c), observe that

(
< 0 if x < c
f (x)
> 0 if x > c.
0

(2)

By the First Derivative Test, f has a minimum at 0. The minimum is absolute (or global) since
by (2) f is strictly decreasing on (, c] and strictly increasing on [c, ). Since f has no other
critical points, the Interior Extremum Theorem implies that c is the unique extremum point of
f.

Sec. 6.2, ex. 16: (a) Let h > 0 be fixed. By the Mean Value Theorem, there exists a point
cx (x, x + h) such that f (x + h) f (x) = f 0 (cx )h. Observe that cx , as x .
Therefore,
f (x + h) f (x)
lim
= lim f 0 (cx ) = b.
x
x
h
(b) Suppose that f (x) a, as x . Then using (a), we have:
f (x + h) f (x)
x
h
limx [f (x + h) f (x)]
=
h
aa
=
h
= 0.

b = lim

(c) First assume b > 0. Let > 0 be arbitrary. Without loss of generality, we can assume that
< 1 and < 2b (since, after all, we only care for very small values of ). Since f 0 (x)
 b, as

0
x , there exists a > 0 such that |f (x) b| < /2, for all x a. Put K = max a , 2f(a) .
By the Mean Value Theorem, for every x > K(> a), there exists a number (a, x) such that
f (x) f (a) = f 0 ()(x a). Therefore,
f (x)
f (x) f (a) f (a)
=
+
x
x
x


a
f (a)
= f 0 () 1
+
.
x
x
Observe that b /2 < f 0 () < b + /2, 1 < 1 xa < 1, and /2 < f (a)
x < /2. Using these
estimates, we obtain two inequalities:
f (x) 


< b+
1 + = b + ,
x
2
2
and
f (x) 


2
> b
(1 ) = b (1 + b) +
> b (1 + b).
x
2
2
2
In other words, for all x > K, we have
f (x)
< b + .
x
Since can be arbitrarily small, it follows that f (x)/x b, as x . If b < 0 the argument
is symmetric, and for b = 0 it is similar.

b (1 + b) <

Remark: You may complain that we didnt show that f (x)/x was greater than b , only
b (1 + b). However, it is not necessary to get exactly on both sides. Heres a useful lemma:
Lemma. Let f : I R and a I. Suppose that there exist functions u, v continuous at 0, with
u(0) = v(0) = 0, such that for every > 0, there exists > 0 such that
L u() < f (x) < L + v(),
for all x I, 0 < |x a| < . Then f (x) L, as x a.

If u(x) = v(x) = x, we get our usual definition of the limit at f at a. Proving this lemma is
a good exercise in - analysis.
Sec. 6.2, ex. 20: (a) Applying the Mean Value Theorem to f on [0, 1], we obtain a point
c1 (0, 1) such that f (1) f (0) = f 0 (c1 )(1 0), i.e., f 0 (c1 ) = 1.
(b) Applying Rolles Theorem to f on [1, 2] yields a point c2 (1, 2) such that f 0 (c2 ) = 0.
(c) Since f 0 (c1 ) = 1, f 0 (c2 ) = 0, and 0 <
between c1 and c2 such f 0 (c) = 1/3.

1
3

< 1, by Darbouxs theorem there exists a point c