Scribd
Upload
30 views

CourseII 2016

This document contains 4 math problems regarding analysis and vector spaces. Problem 1 involves proving properties of functions defined as series over a bounded interval. Problem 2 analyzes the contraction properties of an iterated linear map on R^2. Problem 3 considers a function on an interval and its fixed points. Problem 4 presents an identity for sums of norms of vectors in an inner product space.

Uploaded by

arvind lakshmi ranjan
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
30 views

CourseII 2016

This document contains 4 math problems regarding analysis and vector spaces. Problem 1 involves proving properties of functions defined as series over a bounded interval. Problem 2 analyzes the contraction properties of an iterated linear map on R^2. Problem 3 considers a function on an interval and its fixed points. Problem 4 presents an identity for sums of norms of vectors in an inner product space.

Uploaded by

arvind lakshmi ranjan
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

MAT3004. Unassessed Coursework II. Deadline: May 3, 2016.

Problem 1∗∗ . Consider the space Cb (0, 1) of bounded and continuous functions
f : (0, 1) → R defined on the open interval x ∈ (0, 1) endowed by the sup-norm:

∥f ∥Cb := sup |f (x)|.


x∈(0,1)

Let also xn := 2−n−1 and the spike-like function φn (x) be such that φn (x) = 0 for
x∈/ [xn , xn−1 ], φn ( xn−12+xn ) = 1 and φn (x) is linear on the segments [xn , xn−12+xn ]
and [ xn−12+xn , xn−1 ].
Finally, for any bounded sequence a := {an }∞ n=1 ∈ l∞ , define a function
∞
∑
fa (x) := an φn (x).
n=1

a) Prove that the function fa ∈ Cb (0, 1) for any a ∈ l∞ .


b) Prove that ∥fa ∥Cb (0,1) = ∥a∥l∞ .
c) Deduce from the previous result that the space Cb (0, 1) is not separable. You
may use that l∞ is not separable without proving this fact. Hint: prove that any
subset of a separable metric space is separable.

Problem 2. Let X := R2 with the Euclidean norm and let the map F : X → X
be defined via
(1 )
1
F (x) = Ax + b, A = 2 , x = (x1 , x2 ) ∈ R2
0 12

and b ∈ R2 is some fixed (independent of x) vector.


a) Check that F (x) and F (2) (x) := F (F (x)) both are not contractions on X.
b) Prove that the third iteration F (3) (x) is a contraction on X.

Problem 3. Let√X := [−10, 10] ⊂ R with the standard norm and f : X → R is


defined by f (x) := x2 + 1.
a) Prove that, for any x, y ∈ X, |f (x) − f (y)| ≤ √10 |x
101
− y|.
b) Prove that there are no fixed points of f .
c) Does this result contradict the Banach contraction theorem? Justify your
answer.

√
Problem 4. Let V be an inner product space and ∥x∥ := (x, x). Prove that,
for any three vectors x, y, z ∈ V , the following identity holds:

∥x + y + z∥2 + ∥x∥2 + ∥y∥2 + ∥z∥2 = ∥x + y∥2 + ∥y + z∥2 + ∥x + z∥2 .


Solutions
Problem 1. a) We need to check that fa (x) is continuous and bounded. The
continuity of fa at any x0 ∈ (0, 1) follows from the fact that in the series for fa (x) for
a fixed x, no more than one term is non-zero and, for x belonging to the sufficiently
small neighbourhood of x0 , no more than two subsequent terms in the series are
non-zero. So, the continuity of every spike φn (x) implies the continuity of fa . Let
us check boundedness. Indeed, by the triangle inequality,
∞
∑ ∞
∑
∥fa ∥Cb ≤ |an |φn (x) ≤ ∥a∥l∞ φn (x) ≤ ∥a∥l∞ .
n=1 n=1

Thus, boundedness is also proved and we have verified that fa ∈ Cb (0, 1).

b) We have already proved that ∥fa ∥Cb ≤ ∥a∥l∞ . Let us prove the opposite
inequality. Indeed,
1
∥fa ∥Cb ≥ sup |fa ( (xn−1 + xn ))| = sup |an | = ∥a∥l∞
n∈N 2 n∈N

and the opposite inequality is also proved.

c) Since fαa+βb (x) = αfa (x) + βfb (x) for any α, β ∈ R, a, b ∈ l∞ and x ∈ (0, 1),
the space l∞ is isometrically embedded in Cb (0, 1). The desired statement follows
now from a general fact that any subspace of a separable metric space is separable.
Indeed, let (X, d) be a separable metric space, {zn }∞
n=1 be a countable dense set in
X and Y ⊂ X. Let dn = dn (zn , Y ) := inf y∈Y d(zn , Y ) be a distance from zn to the
set Y . By definition, it means that there exists yn ∈ Y such that d(yn , zn ) ≤ 2dn .
We claim that {yn }∞ n=1 ⊂ Y is a countable dense set in Y . Let y0 ∈ Y be arbitrary.
Then, by assumptions, there exists a subsequence znk ∈ X such that znk → y0 in
X. In particular, this means that dnk → 0 as k → ∞. Then, by triangle inequality

d(ynk , y0 ) ≤ d(znk , y0 ) + d(znk , ynk ) ≤ d(znk , y0 ) + 2dnk → 0

Thus, ynk → y0 in X and {yn }∞


n=1 is a countable dense set in Y .

Problem 2. Obviously F (2) (x) = A(Ax + b) + b = A2 x + b1 , F (3) (x) = A(A2 x +


b1 ) + b = A3 x + b2 , where the constant vectors b1 := Ab + b and b2 = A2 b + Ab + b.
Computing the powers of the matrix A, we see
(1 ) (1 3)
2 4 1 3
A = , A = 8 41 .
0 14 0 8

a) To prove that both F (x) and F (2) (x) are not contractions, we take x = (0, 1)
and y = (0, 0). Then, ∥x − y∥ = ∥x∥ = 1 and
√
1 5
∥F (x) − F (y)∥ = ∥Ax∥ = ∥(1, )∥ = >1
2 2
and F is not a contraction. Analogously,
√
1 17
∥F (2) (x) − F (2) (y)∥ = ∥A2 x∥ = ∥(1, )∥ = >1
4 4
and F (2) is also not a contraction.
b) Let us prove that F (3) (x) is a contraction. We use that F (3) (x) − F (3) (y) =
A (x − y) and denote z := x − y = (z1 , z2 ). Then
3

1 3 1 1 2 3
∥A3 z∥2 = ( z1 + z2 )2 + z22 = (z1 + z22 ) + z1 z2 +
8 4 64 64 16
9 2 1 9 2 3 1 43 2
+ z2 ≤ ( + (z1 + z22 ) + ( (z12 + z22 )) = (z + z22 ).
16 64 16 16 2 64 1
√
Thus, F (3) is a contraction with a contraction factor κ = 843 . Actually, more
√
accurate computations show that the sharp contraction factor is κ = 18 10 + 38 ,
but we need not this to prove that F (3) is a contraction.

Problem 3. a) Due to the mean value theorem, |f (x) − f (y)| = |f ′ (ξ)||x − y|.
Since f ′ (ξ) = √ ξ2 , we see that
ξ +1

10
|f ′ (ξ)| ≤ ∥f ′ ∥C[−10,10] = √
101

and |f (x) − f (y)| ≤ √10 |x


101
− y| for all x, y ∈ X.

b)√There are no fixed points for this map since any fixed point p must satisfy
p = p2 + 1 which gives p2 = p2 + 1 and 0 = 1.

c) This does not contradict the Banach


√ contraction theorem since f maps X to
R, but not X to X. Indeed f (10) = 101 ∈ / X.

Problem 4. This is just a routine calculation

∥x + y + z∥2 = (x + y + z, x + y + z) = ∥x∥2 + ∥y∥2 + ∥z∥2 + 2(x, y) + 2(y, z) + 2(x, z)

and analogously

∥x + y∥2 + ∥y + z∥2 + ∥x + z∥2 = 2∥x∥2 + 2∥y∥2 + 2∥z∥2 + 2(x, y) + 2(y, z) + 2(x, z).

You might also like