Section 4.

5
Exercise 4.5.3: There is no continuous function f : R → R with f (R) = Q. If there were, then we could find
a, b ∈ R with f (a) = 1 and f (b) = 2. Either a < b or b < a. Let’s suppose a < b. Since f√
(x) is continuous on
R it is also continuous on [a, b].
√ By the√intermediate value theorem, and the fact that 1 < 2 < 2, there exists
a c ∈ (a, b) such that f (c) = 2. But 2 6∈ Q, hence f (R), which includes f (c), cannot just be Q. The same
argument works if b < a, but we find c ∈ (b, a) instead.

Exercise 4.5.4: Let f (x) : [a, b] → R be an increasing function (x < y implies f (x) ≤ f (y)) such that for
all x < y in [a, b] and L ∈ R between f (x) and f (y) there is a cL ∈ (x, y) with f (cL ) = L. Pick v ∈ [a, b]. We
wish to show that f (x) is continuous at v. Pick  > 0. Now let x = a and y = v. Since f (x) is increasing on
[a, v] we have h = f (v) − f (a) ≥ 0. Use the intermediate value property to find d1 so that f (d1 ) = f (v) − h̃ where
h̃ = 12 min{h, }. (Note: if h = 0, then f (x) = f (v) for every x ∈ (a, v), so d1 can be any value in (a, v)). Then
for all x ∈ (d1 , v) we have f (d1 ) ≤ f (x) ≤ f (v), i.e. f (v) − h̃ ≤ f (x) ≤ f (v). However, h̃ <  so this implies
− < f (x)−f (v) ≤ 0 for all x ∈ (d1 , v). We now do the same on the other side. Let h̃ = 12 min{f (b)−f (v), }. Then
find d2 such that f (d2 ) = f (v) + h̃. Again we’ll find that for x ∈ (v, d2 ) we have f (v) ≤ f (x) ≤ f (v) + h̃ < f (v) + .
Putting the two sides together, we get for x ∈ (d1 , d2 ) − < f (x) − f (v) <  which is just |f (x) − f (v)| < . Now
just choose (v − δ, v + δ) ⊂ (d1 , d2 ), and we have shown that f (x) is continuous at v.

Exercise 4.5.7: First, note that f ([0, 1]) ⊂ [0, 1] implies f (0) ≥ 0 and f (1) ≤ 1. If either f (0) = 0 or f (1) = 1,
we have found a fixed point. The remaining case to consider is when f (0) > 0 and f (1) < 1. Let g(x) = f (x) − x.
Then g(x) is continuous on [0, 1] since f (x) and y = x are continuous on [0, 1]. In the case we are considering,
g(0) = f (0) − 0 > 0 and g(1) = f (1) − 1 < 0. By the intermediate value theorem, there is a point c ∈ (0, 1) with
g(c) = 0, since g(1) < 0 < g(0). But for this value of c g(c) = 0 means f (c) − c = 0 or f (c) = c.
Section 5.2
Exercise 5.2.2:
(i) Let f (x) = x1 . To compute the derivative, f 0 (a), for a 6= 0 we evaluate
1
f (x) − f (a) −1 a−x
a−x −1 −1
lim = lim x a = lim x a = lim = lim = 2
x→a x−z x→a x − a x→a x − a x→a x a(x − a) x→a x a a
d
which is in agreement with the usual formula: dx xn = n xn−1 for n = −1. Thus f (x) is differentiable on
(−∞, 0) ∪ (0, ∞).
1
(ii) First, use the product rule for f (x)·h(x) where h(x) = g(x) . Note that the domain of h(x) is exactly the
subset of the domain of g(x) where g(x) 6= 0. Provided g(c) 6= 0 we can consider the value of f (x) h(x) at
c. The product rule tells us that if f (x) and h(x) are both differentiable at x = c then so is f (x)h(x) with
derivative f 0 (c)h(c) + f (c)h0 (c). However, we do not yet know that h(x) is differentiable at x = c. This is
where we use the chain rule. g(x) is differentiable at x = c and since g(c) 6= 0, q(x) = x1 is differentiable at
g(c). The chain rule then tells us that q(g(x)) = g(x)1
is differentiable at x = c with derivative q 0 (g(c))g 0 (c).
−1 −g 0 (c)
By the preceding part q 0 (g(c)) = g(c)2
. So h0 (c) = g(c)2
. Putting these two together gives

d f (x) f 0 (c) g 0 (c) f 0 (c)g(c) − f (c)g 0 (c)

= − f (c) =
dx g(x) x=c g(c) g(c)2 g(c)2

(iii) To directly verify the formula use

f (x) f (c)
g(x) − g(c) f (x)g(c) − g(x)f (c)
lim = lim
x→c x−c x→c g(x)g(c)(x − c)
1
 . Rewrite f (x)g(c) − g(x)f (c) = (f (x) − f (c))g(c) − (g(x) − g(c))f (c), so that
 
d f (x) 1 f (x) − f (c)
g(x) − g(c)

= lim g(c) − f (c)
dx g(x) x→c g(x)g(c) x−c x−c
From here it is just a matter of calculating the limit, noting that lim g(x) = g(c) since, g(x) being
x→c
differentiable at x = c, g(x) is also continuous at x = c. The rest of the limits follow from the definition of
f 0 (c) and g 0 (c) and the algebraic limit theorem.

Exercise 5.2.3: Consider the following function

x2 x ∈ Q

f (x) =
0 x∈ 6 Q
We first verify that the function is differentiable at x = 0 where f (x) = 0.
f (x) − 0 f (x)
lim = lim
x→0 x − 0 x→0 x
f (x) x2
But 0 ≤ x ≤ x = |x|, so as x → 0 we must have
f (x)
=0lim
x x→0
Thus f 0 (0) = 0. To see that the function is not differentiable anywhere else we will show that it is not even
continuous except when x = 0. To that end let c 6= 0. If c ∈ Q, let xn → c be a sequence of irrationals con-
verging to c. Then lim f (xn ) = 0 but f (c) = c2 6= 0. If c 6∈ Q, use a sequence of rationals qn → c. Then
n→∞
lim f (qn ) = lim qn2 = c2 (since y = x2 is continuous at x = c), but f (c) = 0.
n→∞ n→∞

Exercise 5.2.5: First note that

   
1 1
ga0 (x) = ax a−1
sin −xa−2
cos
x x
when x 6= 0. For x = 0 we need to evaluate
 
g(x) − 0 1
lim = lim xa−1 sin
x→0 x − 0 x→0 x
This limit will exist and equal 0 provided a > 1 and xa−1 is defined on R (no square roots allowed, for example).
(i) Note that g(x) differentiable at x = 0 means that the first part of g(x) will have a limit as x → 0,
and otherwise g 0 (x) is bounded. So to get something unbounded requires that we make the second part
unbounded, and that means lim xa−2 must be unbounded. This happens when a < 2. So we need 1 < a < 2
x→0
to meet both conditions. Choose a = 35 , say (we use a cubed root to ensure that ga (x) is defined on all of
R.
(ii) To have ga0 (x) continuous we need to have a > 2 so that both terms will go to 0 as x → 0, and the
derivative at 0 will also equal 0. This will make xa−1 have exponent bigger than 1 which will allow the
first term of ga0 (x) to have a derivative at 0 equal to 0, but as long as a − 2 ≤ 1 the second term will not
have a derivative at x = 0. So we choose a = 37 .
(iii) By the reasoning in part (ii), if a > 3 then ga0 (x) will have a derivative at x = 0 and will be differentiable
at every other x as well. Now lets consider the derivative of ga0 (x). We have for x 6= 0 that
       
1 1 1 1
ga00 (x) = a(a − 1)xa−2 sin − a xa−3 cos − (a − 2)xa−3 cos + xa−4 sin
x x x x
Let a = 4. Then ga (x) is differentiable at x =
0 by our preceding calculations, but g 00 (a) will not have a
limit as x → 0 due to the appearance of sin x1