2

The University of Sydney 3. Let X = [1, ∞) and let f : X → X be given by f (x) = x + 1/x. Show that
Pure Mathematics 3901 d(f (x), f (y)) < d(x, y) for all x, y ∈ X with x 6= y, and show that f has no
fixed point in X.
Metric Spaces 2000 Solution.
This time the Contraction Mapping Theorem will not apply since it is not in
Tutorial 9
fact true that there is an α < 1 with d(f (x), f (y)) ≤ αd(x, y) for all x, y ∈ X.
(For all subspaces of R, use the usual (Euclidean) metric.) (So f is not a contraction mapping in the sense of the theorem.) It is clear
that x + (1/x) ≥ 1 whenever x ≥ 1 (since 1/x > 0), and so the formula does
1. Let X = (0, 1/4) and let f : X → X be given by f (x) = x2 . Prove that f define a function X → X. There is no fixed point, since x + (1/x) = x gives
is a contraction mapping with no fixed point in X. Reconcile this with the 1/x = 0, and hence 1 = 0. Now for x, y ≥ 1,
Contraction Mapping Theorem.
1 1
Solution. d(f (x), f (y)) = |(x − y)(1 − xy )| = d(x, y)(1 − xy ) < d(x, y),
A closed subset of a complete space is a complete space, but (0, 1/4) is not as claimed.
closed in R. And indeed it is not complete, since (for example) the sequence
(1/n)∞n=1 is a Cauchy sequence in (0, 1/4) which has no limit in (0, 1/4). 4. Let X = [1, ∞) and let f : X → X be given by f (x) = 2526 (x + 1/x). Show that
Completeness of the space is an important hypothesis of the Contraction d(f (x), f (y)) ≤ 25
26 d(x, y) (whence f is a contraction mapping). By solving
Mapping Theorem; since the hypotheses are not all satisfied here, the theorem the equation algebraically, show that 5 is the unique fixed point of f .
cannot be applied.
If 0 < x < 1/4 then 0 < x2 < 1/16 < 1/4; hence f (x) = x2 does define a map Solution.
from (0, 1/4) to (0, 1/4). The only solutions in R of x2 = x are x = 0 and d(f (x), f (y)) = 25 1 25
26 d(x, y)(1 + xy ) ≤ 26 d(x, y) (cf. Question 3). It is important
x = 1, neither of which lie in (0, 1/4). So f has no fixed points. Finally, if that the given formula does define a map X → X; so let us check that x ≥ 1
25
x, y ∈ (0, 1/4) then implies 26 (x + (1/x)) ≥ 1. If x ≥ 26 25 25
25 then 26 (x + (1/x)) > 26 x ≥ 1. If
1 ≤ x ≤ 25 then 26 ≤ x ≤ 1 and 1 + 26 ≤ x + x , giving 26 (x + x ) ≥ 1275
26 25 1 25 1 25 1
676 > 1.
|x2 − y 2 | = |x − y| |x + y| ≤ |x − y|(|x| + |y|) ≤ |x − y|( 14 + 14 ); Note that X is a complete metric space, being a closed subset of the complete
space R. So the Contraction Mapping Theorem guarantees that there is a
that is, d(f (x), f (y)) ≤ 21 d(x, y), which shows that f is a contraction mapping. unique x ∈ X with f (x) = x. In this case we can easily confirm this by just
solving the equation. If f (x) = x then 26 1 2
25 x = x + x , giving x = 25, so that
2. Let X = { x ∈ Q | x ≥ 1 } and let f : X → X be defined by f (x) = x2 + x1 . x = 5 is the only possible solution in X. (And f (5) = 5 is easily checked.)
Show that f is a contraction mapping and that f has no fixed point in X.
5. Let X = [0, 1] and let f : X → X be given by f (x) = 17 (x3 + x2 + 1). Show
Solution.
that d(f (x), f (y)) ≤ 57 d(x, y). Calculate f (0), f (2) (0), f (3) (0), . . . , and hence
The question asserts that given formula for f defines a mapping X → X; find, to three decimal places, the fixed point of f .
let us check that it is true. If x ∈ Q then x/2 ∈ Q and 1/x ∈ Q; so
(x/2) + (1/x) ∈ Q. If 1 ≤ x ≤ 2 then 1/2 ≤ 1/x ≤ 1 and 1/2 ≤ x/2 ≤ 1; so Solution.
(x/2) + (1/x) ≥ (1/2) + (1/2) = 1. If x > 2 then (x/2) + (1/x) > x/2 > 1. If 0 ≤ x ≤ 1 then 0 ≤ x3 + x2 + 1 ≤ 3, and so 0 ≤ 17 (x3 + x2 + 1) ≤ 37 < 1.
So if x ∈ Q and x ≥ 1 then (x/2) + (1/x) ≥ 1, as required. So again the question has not lied: we do have a function X → X. Now
√
If f (x) = x then (x/2) + (1/x) = x, which gives 1/x = x/2, and x = ± 2.
So f (x) = x has no solution in X. And if x, y ∈ X then d(f (x), f (y)) = 17 |(x3 −y 3 )+(x2 −y 2 )| = 17 d(x, y)|x2 +xy+y 2 +x+y| ≤ 57 d(x, y)

d(f (x), f (y)) = |(x − y)( 12 − 1

xy )| ≤ 12 d(x, y) whenever x, y ∈ [0, 1] (since x2 , xy etc. are all in [0, 1]). Again X is a
complete space, being a closed subset of R. According to my calculator:
since x, y ≥ 1 gives − 21 ≤ 12 − xy
1
< 12 . So f is a contraction mapping. (Again f (0) ≈ 0.142857, f (2) (0) ≈ 0.146189, f (3) (0) ≈ 0.1463565, f (4) (0) ≈ 0.1463650,
the space X is not complete.) . . . —but this is not a course in numerical analysis.
 3 4

6. Let X = [1, 2] ∩ Q and let f : X → X be defined by f (x) = − 14 (x2 − 2) + x. Solution.

Rx Rx
Prove that f is a contraction mapping and that f has no fixed point in X. (i ) (F f )(x) − (F g)(x) = 0 (f (t) − g(t)) dt ≤ 0 d(f, g) dt ≤ x d(f, g).
Z x
Solution. (ii ) (F (2) f )(x) − (F (2) g)(x) = ((F f )(t) − (F g)(t)) dt
The space is not complete, of course. If X were defined simply to be [1, 2] 0
Z x
then the Contraction Mapping Theorem would apply; however, the solution x2
≤ td(f, g) dt = d(f, g).
of f (x) = x in [1, 2] turns out to be irrational, and so not in the set X as 0
2

actually defined. Indeed, f (x) = x if and only if x2 − 2 = 0, and this has no Thus d(F f, F g) = sup |(F f )(x) − (F g)(x)| ≤ sup x d(f, g) ≤ d(f, g), and
solution in Q. The quadratic − 41 (x2 − 2) + x has its turning point at 2, where x∈[0, 1] x∈[0,1]
the function value is 23 . So f (1) = 45 ≤ f (x) ≤ f (2) = 32 for all x ∈ [1, 2]. therefore
This confirms that f (x) ∈ [1, 2], as the question asserts. And it is also clear
x2
that x ∈ Q implies − 14 (x2 − 2) + x ∈ Q. For all x, y ∈ [1, 2], d(F (2) f, F (2) g) = sup |(F (2) f )(x) − (F (2) g)(x)| ≤ sup 2 d(f, g) ≤ 12 d(f, g).
x∈[0,1] x∈[0,1]

d(f (x), f (y)) = |x − y| |1 − 14 (x + y)| ≤ 12 d(x, y)

Hence F (2) is a contraction mapping. But F is not, since by taking f = 1
(since − 41 ≤ 1 − 14 (x + y) ≤ 12 ); so f is a contraction. and g = 0, we have d(F f, F g) = supx∈[0,1] x = 1 = d(f, g).

9. (Square
Pn Pnsum 2criterion) Show that if the n × n matrix C (over R) satisfies
7. Let f : [a, b] → [a, b] be differentiable over [a, b]. Show that f is a contraction n
c < 1, then for any b ∈ R the linear system x = Cx + b has a
mapping if and only if there exists a number K < 1 such that |f 0 (x)| ≤ Kfor j=1 k=1 jk
unique solution. (Imitate thepproof of Theorem 2.1 of Choo’s notes, using the
all x ∈ (a, b). Pn 2 ) instead of the one used there.)
Euclidean metric (d(x, y) = (x
i=1 i − y i )
Solution. Solution.
Suppose first of all that f is a contraction mapping. Then there exists K < 1 Define f : Rn → Rn by f (x) = Cx + b. Let x, y ∈ Rn and
such that |f (x) − f (y)| ≤ K|x − y| for all x, y ∈ [a, b]. So if x ∈ (a, b) is Pn put z = f (x)
and w = f (y). Then z − w = C(x − y), and so zi − wi = j=1 cij (xj − yj )
arbitrary, then for each i. Apply the Cauchy-Schwarz inequality (which says that the dot
product of two vectors in Rn is at most the product of their lengths—here
f (x) − f (y)
f 0 (x) = lim ≤ lim K = K. the two vectors in question
qP are q the i-th row of C and x − y). We de-
y→x x−y y→x n 2
Pn 2
duce that |zi − wi | ≤ j=1 cij j=1 (xj − yj ) . Squaring and summing
Pn n n n
Conversely, suppose that K < 1 and |f 0 (x)| ≤ K for all x ∈ (a, b). Then 2
P P 2

P 2


on i gives i=1 (zi − wi ) ≤ i=1 Pj=1 cP ij j=1 (xj − yj ) . That is,
for arbitrary x, y ∈ [a, b] we have that f (x) − f (y) = f 0 (t)(x − y) for some n n
d(f (x), f (y)) ≤ Kd(x, y), where K = i=1 j=1 c2ij < 1. Thus f is a con-
t ∈ (x, y) ⊆ (a, b) (by the Mean Value Theorem). This gives traction mapping, and so has a unique fixed point.

|f (x) − f (y)| = |f 0 (t)| |x − y| ≤ K|x − y|,

as required.

8. Let C be the set of continuous functions [0, 1] → R, and let d be Rgiven by

x
d(f, g) = supx∈[0, 1] |f (x) − g(x)|. Define F : C → C by (F f )(x) = 0 f (t) dt
(for all f ∈ C). Show that for all f, g ∈ X and all x ∈ [0, 1],
(i ) (F f )(x) − (F g)(x) ≤ x d(f, g), and
x2
(ii ) (F (2) f )(x) − (F (2) g)(x) ≤ 2 d(f, g),
and deduce that F (2) is a contraction mapping. Show, however, that F is not
a contraction mapping.