M597K: Solution To Homework Assignment 7: 1. Show That The Sequence (X
M597K: Solution To Homework Assignment 7: 1. Show That The Sequence (X
M597K: Solution To Homework Assignment 7: 1. Show That The Sequence (X
The following problems are on the specified pages of the text book by Keener
(2nd Edition, i.e., revised and updated version)
Problems 3 and 4 of Section 2.1 on p.94;
Problem 1 of Section 3.1 on p.128;
Problem 1 of Section 3.2 on p.128.
Pn 1
1. Show that the sequence (xn ), xn = k=1 k! is a cauchy sequence using the
measure of distance d(x, y) = |x − y|.
Solution.
A sequence of points xn ∈ IR is Cauchy if for every > 0 there exists an N , such
that m > n > N implies |xn − xm | <
Let > 0 be given, choose N = √1 . Then for m > n > N , we have
|xn − xm | = 1 1 1
(n+1)! + (n+2)! + ... + m!
≤ 1 1
n(n+1) + n(n+1)(n+2) + ...
≤ 1 1
n(n+1) + n(n+1)(n+1) + ...
≤ 1 1 1
n(n+1) (1 + n+1 + (n+1)2 + ...)
≤ n(n+1) · 1− 1 = n2 = .
1 1 1
n+1
where
y(x − 1) , 0 ≤ y < x ≤ 1,
k(x, y) =
x(y − 1) , 0 ≤ x < y ≤ 1.
Solution.
We should first check the boundary:
Z
u(0) = k(0, y)f (y)dy = 0
Z
u(1) = k(1, y)f (y)dy = 0
And
R1
u(x) = 0 k(x, y)f (y)dy
Rx R1
= 0 y(x − 1)f (y)dy + x x(y − 1)f (y)dy
R1 Rx R1
= 0 xyf (y)dy − 0 yf (y)dy − x xf (y)dy
R1 Rx Rx
u0 (x) = 0 yf (y)dy − xf (x) + 1 f (y)dy + xf (x) = 1 f (y)dy
u00 (x) = f (x)
4. Show T (f ) = f (0), defined for all continuous functions on [−1, 1], is not a
bounded linear functional in the L2 norm, but it is a bounded linear functional with
the uniform norm.
Solution.
For T to be bounded on continuous functions in the L2 norm, there must exist
a constant C such that
The converse: i.e., for T not bounded, the statement is: For any finite number
C, there is always a continuous function f , which may depend on C, thus we use
fC (x) to denote it, such that
|T f | > Ckf kL2 . (1)
2
We prove that such an fC (x) exists. So let C be a positive number. Take fC (x)
be such that
fC (x) = C, x ∈ [−d, d],
and fC (x) = 0 for all other x, where d is a small number. Thus, we have |T fC (x)| =
√
|fC (0)| = C, and kfC (x)kL2 = C 2d. If we take d = 8C1 2 , then inequality (1)
holds like C > C · 12 . Thus this T is not a bounded linear functional on the set of
continuous functions measured in the norm L2 . (The function fC (x) can be made
continuous easily without changing is primary property.)
Since |f (0)| ≤ max |f (x)|, T is bounded with uniform norm (take C = 1).
5. Let `2 denote all the sequences (x1 , x2 , x3 , · · · , xn , · · ·) of real numbers. Let x
denote such a sequence. Use vector addition and scalar multiplication. Then it is a
vector space (no proof needed). Use the inner product
∞
X
< x, y > = xi y i
i=1
where y = (y1 , y2 , · · · , yn , · · ·). Show that this inner product is well defined in `2 and
it satisfies the four properties of the definition of inner product. (It is called “the
little ` two space”. (Reference: p. 59 of text book)
Solution.
By Cauchy-Schwarz inequality:
v v
∞ u∞ u∞
X uX uX
xi yi | ≤ t x2i t yi2 =< x, x > 2 < y, y > 2
1 1
| < x, y > | = |
i=1 i=1 i=1
3
Let g(t) be equal to 85 for t between 35 and 116, and equal to zero for all other
times t. This might represent another person’s social status score history, who at
age 35 invented a perpetual machine and enjoyed the fame he got throughout his
lifetime. (Unfortunately he lived only to 116. Thus his perpetual machine was just
somewhat perpetual, and that explains the score 85 instead of a higher number.)
Now a panel want to select one from the two to put into a Hall of Fame and want
you to give the panel a single measurement number from each of the two so that
the panel can decide. Do the maximum norm and the L2 norm calculation (serious
part) and decide which norm you want to use to give to the panel (a decision that
is purely up to you).
Solution.
100 , 19 ≤ t ≤ 20,
f (t) =
0 , else,
85 , 35 ≤ t ≤ 116,
g(t) =
0 , else,
k f (t) k∞ = 100, k g(t) k∞ = 85
R 20 1 R 116 1
k f (t) kL2 = ( 19 1002 dx) 2 = 100, k g(t) kL2 = ( 35 852 dx) 2 = 765
7. Solve Problem 5, p. 94, of the text book, where L2 is replaced by L1 .
Solution.
As fn (t) = 0 when 0 ≤ t < 1
2 − n1 , and fn (t) = 1 when t > 1
2 + 1
n
Suppose n < m, fn (t) − fm (t) = 0 when 0 ≤ t < 1
2 − 1
n or t > 1
2 + 1
n and
|fn (t) − fm (t)| ≤ |fn (t)| + |fm (t)| ≤ 2
R1 1 R 1
+1 1
Thus k fn (t) − fm (t) kL2 = ( 0 (fn (t) − fm (t)) ) ≤ (
2 2
1
2 n
−1
(fn (t) − fm (t))2 ) 2 ≤
R 1
+1 1
q 2 n
( 1
2 n
−1
(2)2 ) 2 = 8
n → 0 as n → ∞
2 n
That implies it is Cauchy in L1
8. Show that the functional T on the Banach space C[0, 1] defined by
Z 1 f (x)
Tf = √ dx, f (x) ∈ C[0, 1]
0 x
4
R1 αf (x)+βg(x) R1 f√
(x) R1 f√
(x)
T (αf + βg) = √ dx =α dx +β dx = αT (f ) + βT (g).
0 x 0 x 0 x
Thus T is linear.
R1 f√
(x) R1
And |T f | = | 0 x
dx| ≤ max |f (x)| 0
√1 dx
x
= 2 · max |f (x)|
Thus T is bounded.
9. Let a and b be two points in the interval [0, 1]. Show that the functional T on
the Banach space C[0, 1] defined by
is linear and bounded. (This functional is generally written as δ(x − a) − δ(x − b).)
(Hint: Lecture notes might help.)
Solution.
T (αf + βg) = α(f (a)) + βg(a) − (α(f (b)) + βg(b)) = α(f (a) − f (b)) + β(g(a) −
g(b)) = αT (f ) + βT (g).
Thus T is linear.
And |T f | = |f (a) − f (b)| ≤ |f (a)| + |f (b)| ≤ 2 · max |f (x)|
Thus T is bounded.
10. Find the adjoint operator T ∗ for the operator T defined on the Hilbert space
L2 [a, b] by
Z b
T u(x) = w(x) k(x, y)u(y)dy
a
where k(x, y) ∈ L2 ([a, b] × [a, b]) and w(x) ∈ C[a, b] are both given functions and u
is an arbitrary member in L2 [a, b]. (Hint: Read text book p.107.)
Rb
Solution. Starting with the definition of T u(x) = w(x) a k(x, y)u(y)dy and the
definition of T∗ we formulate
Rb Rb
< T u, v > = a (w(x) a k(x, y)u(y)dy)v(x)dx
R R
= ab ab w(x)k(x, y)v(x)u(y) dx dy
R R
= ab ( ab k(x, y)w(x)v(x) dx)u(y) dy
=< T ∗ v, u > .
Rb
Thus T ∗ v(y) = a k(x, y)w(x)v(x)dx or equivalently but more standard:
Z b
T ∗ u(x) = k(y, x)w(y)u(y) dy.
a
5
11. Verify that λ = (nπ)2 and u = cos(nπx) are eigenvalues and corresponding
eigenfunctions for the Sturm-Liouville eigenvalue problem:
d2 u
+ λu = 0, (0 < x < 1); u0 (0) = u0 (1) = 0
dx2
for all positive integer n.
Solution.
From λ = (nπ)2 and u = cos(nπx), we can get du
dx = −nπ sin(nπx), and
d2 u
+ λu = 0, (0 < x < 1)
dx2
And as u0 (x) = −nπ sin(nπx), u0 (0) = 0, u0 (1) = 0.