Access full Solution Manual only here

http://www.book4me.xyz/solution-manual-fourier-transforms-hansen/

CHAPTER 1

Review of Prerequisite Mathematics

1-1.
1
kvk2 C kwk2 wk2

v
w D kvk kwk cos  D kv
2
1 2
v C vy2 C wx2 C wy2 wx /2 wy /2

D .vx .vy
2 x
D vx wx C vy wy :

1-2. (a) Begin with v10 e01 C v20 e02 D v1 e1 C v2 e2 . Because e01
e02 D 0, to isolate v10 on the left
hand side, take the dot product of both sides with e01 . Then

v10 D v1 e1
e01 C v2 e2
e01

H) a11 D e1
e01 ; a12 D e2
e01
and similarly, a21 D e1
e02 ; a22 D e2
e02 :

(b) In the one basis,

v
w D .v1 e1 C v2 e2 /
.w1 e1 C w2 e2 / D v1 w1 C v2 w2

by the orthonormality of the basis vectors. Similarly, using the other basis,

v
w D v10 w10 C v20 w20 :

(c) In the primed basis,

v0 C w0 D v10 e01 C v20 e02 C w10 e01 C w20 e02 D v10 C w10 e01 C v20 C w20 e02

Substitute for the primed coefficients from part (a),

v0 C w0 D .v1 C w1 / e1
e01 e01 C .v2 C w2 / e2
e01 e01

C .v1 C w1 / e1
e02 e02 C .v2 C w2 / e2
e02 e02

and collect terms,

D .v1 C w1 / e1
e01 e01 C e1
e02 e02 C .v2 C w2 / e2
e01 e01 C e2
e02 e02


 


„ ƒ‚ … „ ƒ‚ …
De1 De2
2 Review of Prerequisite Mathematics

(d) Apply the scalar to the vector in the primed basis and see what happens to the
coefficients after transformation to the original basis. In the primed basis, c 0 v D
c 0 v10 e01 C v20 e02 D c 0 v10 e01 C c 0 v20 e02 . Substitute for the primed coefficients from part

(a),

c 0 v10 e01 C c 0 v20 e02 D c 0 v1 e1
e01 C v2 e2
e01 e01 C c 0 v1 e1
e02 C v2 e2
e02 e02

D c 0 v1 e1
e01 e01 C e1
e02 e02 Cc 0 v2 e2
e01 e01 C e2
e02 e02 D c 0 v

„ ƒ‚ … „ ƒ‚ …
De1 De2

The scalar is unchanged by the change of basis.

1-3. (a) Write the function f .x/ D fe .x/ C fo .x/. By the definition of even and odd,
f . x/ D fe .x/ fo .x/. Adding these two equations gives fe , and subtracting gives
fo (1.16).
Z a Z 0 Z a
(b) Write f .x/ dx D f .x/ dx C f .x/ dx. With f odd, changing variables in
a Za 0 0Z
a Z a
the first integral gives f .x/ dx D f . x/ dx D f .x/ dx, which cancels
a 0 0
the second integral, leading to (1.17).
Z a Z 0 Z a
(c) Write f .x/ dx D f .x/ dx C f .x/ dx. With f even, changing variables
a aZ 0 Z
0 a Z a
in the first integral gives f .x/ dx D f . x/ dx D f .x/ dx, which is the
a 0 0
same as the second integral and leads to (1.18).

1-4. (a) f .x/ is a step function that jumps from 0 to 1 at x D 1. f . x/ is a step function
that jumps from 1 to 0 at x D 1. Adding the two (make a sketch),
8
1
ˆ 2 ; jxj > 1
ˆ
ˆ
<
fe .x/ D 14 ; jxj D 1
ˆ
ˆ
:0; jxj < 1
ˆ

and subtracting, fo .x/ D fe .x/ sgn.x/.

(b) f . x/ D e x U. x/. Adding the two gives
1 jxj
fe .x/ D e ;
2
a two-sided exponential. Subtracting gives fo .x/ D fe .x/ sgn.x/.
(c) f .x/ is a ramp with unit slope between 0 and 1, and zero otherwise (make a sketch).
f . x/ is a ramp with slope 1 between 1 and 0, and zero otherwise. Adding the
two, 8
jxj
ˆ2;
ˆ
ˆ jxj < 1
<
fe .x/ D 14 ; jxj D 1
ˆ
ˆ
:0; jxj > 1
ˆ
 Access full Solution Manual only here

http://www.book4me.xyz/solution-manual-fourier-transforms-hansen/
Review of Prerequisite Mathematics 3

Subtracting gives
8
ˆ
ˆ
ˆ x; jxj < 1
<
fo .x/ D fe .x/ sgn.x/ D ˙ 14 ; x D ˙1
ˆ
ˆ
:0; jxj > 1
ˆ

It may be tempting to think, from these problems, that the odd part is always the
even part times a signum, but this is not so.

f .x/ f . x/ f 0 .x/ . 1/f 0 . x/

1-5. The odd part of a function is fo .x/ D . Its derivative is fo0 .x/ D D
2 2
f 0 .x/ C f 0 . x/
, which is, by definition, the even part of f 0 . Therefore fo0 is even. The
2
same reasoning, beginning with the definition of fe .x/, shows that fe0 .x/ is odd.

1-6. (a) Let A D B in (1.20). Then sin 2A D sin A cos A C cos A sin A D 2 sin A cos A, and
similarly, cos 2A D cos2 A sin2 A.
(b) Change B to B in (1.20). Then sin.A B/ D sin A cos. B/ C cos A sin. B/ D
sin A cos B B/ D cos A cos B C sin A sin B.
cos A sin B, and similarly, cos.A
(c) Write cos 2A D cos A sin A (part (a)) D 1 sin2 A
2 2
sin2 A D 1 2 sin2 A.


2 1 2 1
Solve for sin A D 2 .1 cos 2A/. Similarly, cos A D 2 .1 C cos 2A/.

1-7.

C sin.2t C '/ D C sin.2t / cos ' C C cos.2t / sin '

D .C sin '/ cos.2t / C .C cos '/ sin.2t / :
p
In this form, A D C sin ' and B D C cos '. We still have C D A2 C B 2 , but tan ' D
A=B.

x2 x3 x4
1-8. The Taylor series for e x is e x D 1 C x C 2Š
C 3Š
C 4Š
C : : :. (All derivatives of e x are
x
e , and evaluate to 1 at x D 0.). Substitute x D i, and

.i /2 .i /3 .i /4 2 4 3

e i D 1 C i C C C C ::: D 1 C C : : : C i i C :::
2 3Š 4Š „ 2Š ƒ‚4Š … „ 3Š
ƒ‚ …
cos  i sin 
4 Review of Prerequisite Mathematics

To verify the Taylor series coefficients for sin  ,

sin 0 D 0
ˇ
d sin  ˇˇ
D cos 0 D 1
d ˇ0
d 2 sin  ˇˇ
ˇ
D sin 0 D 0
d 2 ˇ0
d 3 sin  ˇˇ
ˇ
D cos 0 D 1
d 3 ˇ0
::
:

and similarly for cos  .

1-9. The easier way is

j exp.i /j2 D exp.i / exp. i / D exp.0/ D 1;

but using the Euler equations,

j exp.i /j2 D j cos  C i sin j2 D cos2  C sin2  D 1

N
X1 1 xN
1-10. We want to show SN D xn D . It is easy to see that S1 D 1 and S2 D
nD0
1 x
2 K
X1 K
X2
1 x
D 1 C x. Now, SK D xn D xn C xK 1
D SK 1 C xK 1
, and assuming
1 x nD0 nD0
1 xK 1
SK 1 D ,
1 x
1 xK 1
1 xK 1
C .1 x/x K 1
1 xK
SK D C xK 1
D D :
1 x 1 x 1 x

1 e 2x
1-11. (a) is of the form 0=0 as x ! 0, so apply L’Hospital’s rule,
x
1 e 2x 2e 2x
lim D lim D 2
x!0 x x!0 1
1 cos x
(b) is of the form 0=0 as x ! 0, so apply L’Hospital’s rule,
x
1 cos x  sin x
lim D lim D0
x!0 x x!0 
Review of Prerequisite Mathematics 5

1-12. (a) Let u D x, dv D e ax dx, then

1 1 ax 1 1 ax ax 1 ax
Z Z
xe ax dx D xe ax e dx D xe ax 2
e D e :
a a a a a2
1
(b) Let u D x 2 , dv D cos bx, then du D 2xdx and v D b
sin bx, and
x2 2
Z Z
x 2 cos bx dx D sin bx x sin bx dx:
b b
2x 2 1
Now let u D b
, dv D sin bx dx, then du D b
dx and v D b
cos bx, and

x2 2x 2
Z Z
x 2 cos bx dx D sin bx C cos bx cos bx dx
b b2 b2
x2 2x 2 b2x2 2 2x
D sin bx C cos bx sin bx D sin bx C 2 cos bx :
b b2 b 3 b3 b

1=2
1-13. (a) x grows more slowly than 1=x as x ! 0, so it is integrable. The integral is
Z 1 ˇ1
dx 1=2 ˇ
ˇ
1=2
D 2x ˇ D 2 :
0 x 0
1
(b) decays more rapidly than 1=x as jxj ! 1, so it is integrable. Using an
1 C x2
integral table or symbolic integrator,
Z 1 ˇ1
dx ˇ   
2
D arctan x ˇ D D
1 1Cx 2 2
ˇ
1

1-14. (a) The integrand is positive and even, and grows faster than 1=x as x ! 0, so it is not
integrable.
(b) The integrand grows faster than 1=x as x ! 0, but because it is the integral of an
odd function over a symmetric interval, we expect the integral to be zero. Calculate
the Cauchy principal value,
Z 1 Z  Z 1 
dx dx dx
P 3
D lim 3
C 3
1 x !0 1 x  x
    
1 1 1 1 1
D lim 1 D0
!0 2  2 . 1/2 2 2
(c) Using the result of part (b), we may write
1 ˇˇ2
Z 2 Z 1 Z 2 ˇ
dx dx dx 3
P 3
D P 3
C 3
D D :
1 x 1 x 1 x 2x 2 ˇ1 8
„ ƒ‚ …
0
 Access full Solution Manual only here

http://www.book4me.xyz/solution-manual-fourier-transforms-hansen/

CHAPTER 2

Vector Spaces

2-1. (a) It is not a vector space, because there cannot be an additive inverse if all the signals
are positive. Other than that, it works.
(b) This is a vector space. The key point is closure under addition, and the sum of two
signals A cos .!t C ˛/ C B cos .!t C ˇ/ has frequency ! (use Equation 1.21).
(c) This is not a vector space. The set is not closed under addition (can get pixel values
greater than 255). Also, there is no additive inverse.
(d) Invalid problem, don’t assign. This appears to be better that (c), because the set
is now closed under addition and under scalar multiplication. However, there is an
algebraic problem that was glossed over in the text. The definition of vector space
used here effectively assumes that the scalars are either real or complex numbers.
Any other set of scalars must comprise what is called a field, with certain addition
and multiplication properties that these numbers lack.
(e) One-sided signals do comprise a vector space. The key point is closure under addi-
tion and scalar multiplication. Adding two one-sided signals results in a one-sided
signal; multiplication by a scalar results in a one-sided signal.
(f) One-sided signals that decay exponentially do comprise a vector space. Again,
closure is the key point. The sum of two exponentially decaying functions decays
exponentially (dominated by the slower decay).
(g) Piecewise-constant functions comprise a vector space. The sum of two piecewise-
constant functions is also piecewise constant, and the product of a scalar with a
piecewise-constant function is piecewise constant. This is the key point.
(h) Waves on a 1m string with fixed ends are functions f .x/ defined on an interval Œ0; 1
with f .0/ D f .1/ D 0. This set of functions is closed under addition and scalar
multiplication, because the boundary conditions are still satisfied. This is the key
point in establishing the set as a vector space.
(i) If two signals f and g have zero average value, their sum also has zero average
value, and the product with a scalar, cf , also has zero average value. This is the
key point in establishing the set as a vector space.
(j) The set of functions with unit area is not a vector space. The area under the sum
of two such functions is 2, not 1, violating closure under addition.

2-2. (a) This is a vector space. Consider two such vectors, .x1 ; x2 ; x3 / and .y1 ; y2 ; y3 /. The
product with a scalar c is .cx1 ; cx2 ; cx3 /, and cx1 Ccx2 Ccx3 D c.x1 Cx2 Cx3 / D 0.
8 Vector Spaces

The sum of two vectors is .x1 C y1 ; x2 C y2 ; x3 C y3 /, and .x1 C y1 / C .x2 C y2 / C

.x3 C y3 / D .x1 C x2 C x3 / C .y1 C y2 C y3 / D 0.
One of the vector components is determined by the other two, e.g., x3 D x1 x2 ,
which reduces the dimensionality of the space to two. Two basis vectors are .1; 1; 0/
and .0; 1; 1/. They are linearly independent, but not orthogonal.
(b) This is not a vector space. The sum of two such vectors, .x1 ; x2 ; x3 / and .y1 ; y2 ; y3 /,
is .x1 C y1 ; x2 C y2 ; x3 C y3 /, and .x1 C y1 / C .x2 C y2 / C .x3 C y3 / D .x1 C x2 C
x3 / C .y1 C y2 C y3 / D 2, not 1.
(c) fReal m  n matrices Ag is a vector space. The key point is closure under addition
and scalar multiplication, which holds.
The space is mn-dimensional. A basis consists of mn matrices, with 1 in the ij
location and 0 in all the other locations, for i D 1; 2; : : : m and j D 1; 2; : : : n.
(d) fReal n  n matrices A with det A D 0g is not a vector space, because
" # the determi-
" #
1 0 0 0
nant is not a linear operation. For a counterexample, consider and .
0 0 0 1
" #
1 0
Their determinants are each 0, but their sum is , which has determinant 1.
0 1
(e) Square symmetric matrices are a vector space. If you add two symmetric matrices,
the result is symmetric: .A C B/0 D A0 C B 0 D A C B. If you multiply a symmetric
matrix by a scalar, the result is symmetric: .cA/0 D c 0 A0 D cA0 D cA. The additive
inverse, A, is also symmetric. These are the key points.
For a basis, we seek a set of square symmetric matrices that are linearly independent
and span the space of square symmetric matrices. One such set, for n D 3, is

82 3 2 3 2 3 2 3 2 3 2 39
< 1
ˆ 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 >=
40 0 05 ; 40 1 05 ; 40 0 05 ; 41 0 05 ; 40 0 15 ; 40 0 05
6 7 6 7 6 7 6 7 6 7 6 7
ˆ >
0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 1 0 0
: ;

Here there are six basis matrices and the dimension of the space is, correspondingly,
six. In general, for nn matrices, there are nC.n 1/C.n 2/C: : :C1 D n.nC1/=2
basis matrices (dimensions).
(f) nth -degree polynomials are a vector space. The sum of two such polynomials is an
nth -degree polynomial. Multiplying by a scalar results in an nth -degree polynomial.
The additive inverse, obtained by multiplying all coefficients by 1, is an nth -degree
polynomial. These are the key points.
For a basis, we need linearly independent nth -degree polynomials. Here is a set that
works: x n ; x n C 1; x n C x; x n C x 2 ; : : : ; x n C x n 1 (n dimensions).
˚

(g) These polynomials do not comprise a vector space. The sum of two such polynomi-
als, f C g, at x D 0, is .f C g/.0/ D f .0/ C g.0/ D 2 ¤ 1.
(h) These polynomials do comprise a vector space. The sum of two such polynomials,
f C g, at x D 1, is .f C g/.1/ D f .1/ C g.1/ D 0. Multiplying by a scalar,
cf .1/ D c
0 D 0. The additive inverse, f , has f .1/ D 0. These are the key
points.
Vector Spaces 9

For a basis, we seek linearly independent polynomials of degree  n, with p.1/ D 0.

Here is a set that works: x n 1; x n 1 1; : : : ; x 1 . The dimension of the space is
˚

n. There is no nontrivial n D 0 (constant) member of the space and no constant basis

element. If there were, because p.1/ D 0, the constant would be zero everywhere,
which is trivial.

2-3. For continuous functions f; g; h 2 C Œ0; 1: (a) The sum .f C g/.x/ is continuous and the
product .cf /.x/ D cf .x/ is continuous. Because these functions are real-valued, at every
x they satisfy commutative, associative and distributive properties (b-d) and then, by
(a), these combinations are continuous. (e) The additive identity element is 0. (f) Each
function f has an additive inverse . f /.x/ D 1
f .x/ which is continuous, by (a). (g)
The multiplicative identity element is 1.

q
jv1 j2 C jv2 j2 D v1 v1 C v2 v2 , applying Definition 2.2,
p
2-4. For kvk D

(a) The squared magnitudes jv1 j2 and jv2 j2 are nonnegative, so the square root of their
sum must be nonnegative.
(b) Clearly, if v = 0 then kvk D 0. On the other hand, if kvk D 0, then jv1 j2 Cjv2 j2 D 0,
and because jv1 j2 and jv2 j2 are nonnegative, they can sum to zero only if they are
both zero, hence v D 0. r 
q  q
(c) kcvk D jcv1 j2 C jcv2 j2 D c 2 jv1 j2 C jv2 j2 D jcj jv1 j2 C jv2 j2 D jcjkvk:
(d) To verify the triangle inequality, write

kv C wk2 D hv C w; v C wi D kvk2 C kwk2 C hv; wi C hv; wi

Now, by the triangle inequality (since each term is a real or complex number),

kv C wk2 D ˇkv C wk2 ˇ  kvk2 C kwk2 C 2 jhv; wij  kvk2 C kwk2 :

ˇ ˇ

2-5.

kxk1 D j1j C j2i j D 1 C 2 D 3

p p p
kxk2 D j1j2 C j2ij2 D 1 C 4 D 5
kxk1 D max fj1j; j2ijg D 2

2-6. Write x D .x y/ C y and apply the triangle inequality,

kxk D k.x y/ C yk  kx yk C kyk

10 Vector Spaces

then rearrange terms,

kx yk  kxk kyk :

By similar reasoning, exchanging x and y,

ky xk  kyk kxk D .kxk kyk/

Together, these imply that

ˇ ˇ
kx yk  ˇkxk kykˇ :
ˇ ˇ

n
X
2-7. In Rn , d1 .x; y/ D jxk yk j. Squaring both sides,
kD1

n
!2 n X
n
X X
d12 .x; y/ D
ˇ ˇ
jxk yk j D ˇxj yj ˇ jxk yk j
kD1 j D1 kD1
n
X Xˇ
yk j2 C
ˇ
D jxk ˇxj yj ˇ jxk yk j
kD1 j ¤k
„ ƒ‚ … „ ƒ‚ …
d22 .x;y/ 0

H) d1 .x; y/  d2 .x; y/

Next,
n
X  2
d22 .x; y/ D jxk yk j2 D max jxk yk j C n 1 other
ƒ‚ terms
k „ …
kD1 „ ƒ‚ … 0
2
d1 .x;y/

H) d2 .x; y/  d1 .x; y/

2-8.  For d1 .0; x/ D jx1 j C jx2 j D 1, the “unit circle” is a diamond:

8
<˙ .1 x / ; 1  x  0
1
x2 D ˙ .1 jx1 j/ D
:˙ .1 C x / ; 0  x  1
1

p
 For d2 .0; x/ D jx1 j2 C jx2 j2 D 1, the “unit circle” is the unit circle.
 For d1 .0; x/ D max fjx1 j; jx2 jg D 1, the “unit circle” is a square: jx2 j D 1 for
jx1 j < 1, and jx1 j D 1 for jx2 j < 1.
Vector Spaces 11

x2

d∞
d2
d1

x1

Comparing with Problem 2-7, d1 .0; x/  d2 .0; x/  d1 .0; x/, the “unit circles” are
ordered with d1 inside d2 inside d1 . d1 is the smallest of the metrics, so the points for
which d1 .0; x/ D 1 will be the farthest from the origin.

2-9. (a) hv; cwi D hcw; vi D c  hw; vi D c  hv; wi

(b) hu; v C wi D hv C w; ui D hv; ui C hw; ui D hu; vi C hu; wi
(c) For any vector u, 0u D 0, so hv; 0i D hv; 0ui D 0 hv; ui D 0, and similarly for h0; vi.
(d) If hu; vi D hu; wi, then hu; vi hu; wi D hu; v wi D 0. That is, v w is orthogonal
to every u 2 V , which implies v w D 0, hence v D w.

2-10. With reference to Figure 2.2, using the Law of Cosines,

ku vk2 D kuk2 C kvk2 2kuk kvk cos .angle opposite u v/

ku C vk2 D kuk2 C kvk2 2kuk kvk cos .angle opposite u C v/

Adding the two,

ku C vk2 C ku vk2 D 2kuk2 C 2kvk2 2kukkvk Œcos .angle opposite u C v/ C cos .angle opposite u v/ :

In a parallelogram, the angles sum to 360ı , so the angle opposite u C v is 180ı the
angle opposite u v. Now, cos. A/ D cos A, so the sum of the two cosines is 0,
leaving the parallelogram law.

2-11. (a) In an inner product space, hu; ui D kuk2 . Then,

ku C vk2 D hu C v; u C vi D hu; ui C hu; vi C hv; ui C hv; vi

ku vk2 D hu v; u vi D hu; ui hu; vi hv; ui C hv; vi

Now add these two equations.

12 Vector Spaces

(b) In the parallelogram law, ku C vk2 Cku vk2 D 2kuk2 C2kvk2 . Since ku vk2  0,
subtracting it from the left hand side makes the left hand side  the right hand
side.

2-12. (a) For real u; v, use the equations for ku C vk2 and ku vk2 from the previous proof,
and subtract them instead of adding, giving

ku C vk2 ku vk2 D 2 hu; vi C 2 hv; ui D 4 hu; vi :

For complex u; v, we cannot equate the two inner products,

ku C vk2 ku vk2 D 2 hu; vi C 2 hv; ui

Derive a similar expression for ku C ivk2 ku ivk2 ,

ku C ivk2 ku ivk2 D 2i hu; vi C 2i hv; ui

Then eliminate the hv; ui term by combining the two expressions,

ku C vk2 ku vk2 C i ku C ivk2 ku ivk2 D 2 hu; vi C 2 hv; ui C i . 2i hu; vi C 2i hv; ui/

D 4 hu; vi

(b) Apply the triangle inequality, then the parallelogram law,

j4 hu; vi j  ku C vk2 C ku vk2 C ku C ivk2 C ku ivk2

„ ƒ‚ … „ ƒ‚ …
2kuk2 C2kvk2 2kuk2 C2kvk2

H) j hu; vi j  kuk2 C kvk2

For real u; v, begin with the real result,

j4 hu; vi j  ku C vk2 C ku vk2

„ ƒ‚ …
2kuk2 C2kvk2

kuk C kvk2
2
H) j hu; vi j 
2

2-13. Assume that the parallelogram law holds. Then it holds in particular for the vectors
u D .a; 0; 0; : : : ; 0/ and v D .0; b; 0; : : : ; 0/. Calculating the various norms,

kuk1 D k.a; 0; 0; : : : ; 0/k1 D jaj

kvk1 D k.0; b; 0; : : : ; 0/k1 D jbj
ku C vk1 D k.a; b; 0; : : : ; 0/k1 D jaj C jbj
ku vk1 D k.a; b; 0; : : : ; 0/k1 D jaj C jbj

and substituting into the parallelogram law,

ku C vk21 C ku vk21 D 2kuk21 C 2kvk21 ;

we obtain
2.jaj C jbj/2 D 2jaj2 C 2jbj2 ;
Vector Spaces 13

which is only true in the trivial case a D 0 or b D 0. A similar calculation shows that
kuk1 also cannot be calculated as an inner product.

2-14. (a) hau; bvi D ab  hu; vi D 0

(b) hu C v; u vi D hu; ui C hv; ui hu; vi hv; vi D 0.
„ƒ‚… „ƒ‚… „ƒ‚… „ƒ‚…
kuk2 D0 D0 kvk2

2-15. (a) Norms of u:

p p p
kuk1 D j2j C j1 C i j C ji j C j1 i j D 2 C 2 C 1 C 2 D 3 C 2 2
p p
kuk2 D j2j2 C j1 C ij2 C ji j2 C j1 i j2 D 4 C 2 C 1 C 2 D 3
kuk1 D 2

(b) Norms of v:

kvk1 D j1j C ji j C j 1j C ji j D 4
p
kvk2 D 4 D 2
kvk1 D 1

(c) hu; vi D 2  1 C .1 C i /  i C i  1 C .1 i /  i D 2 3i
(d) hv; ui D 1  2 C i  .1 i / 1  . i / C i  .1 C i / D 2 C 3i

Note kuk1 > kuk2 > kuk1 , kvk1 > kvk2 > kvk1 , and hu; vi D hv; ui .

2-16. If the vectors u and v are orthogonal, then

ku C vk2 D hu C v; u C vi D hu; ui C hu; vi C hv; ui C hv; vi D kuk2 C kvk2 :

„ƒ‚… „ƒ‚…
0 0
2 2 2
Conversely, if ku C vk D kuk C kvk , then

hu; ui C hu; vi C hv; ui C hv; vi D hu; ui C hv; vi

H) hu; vi C hv; ui D 2Re hu; vi D 0

If u and v are real, then their inner product has no imaginary part, and 2Re hu; vi D
2 hu; vi D 0, which implies u and v are orthogonal. If u and v are complex, then their
inner product can still have a nonzero imaginary part and satisfy 2Re hu; vi D 0; that
is, they can be nonorthogonal.

2-17.

ke1 e2 k22 D he1 e2 ; e1 e2 i D he1 ; e1 i he1 ; e2 i he2 ; e1 i C he2 ; e2 i D 2

„ ƒ‚ … „ ƒ‚ … „ ƒ‚ … „ ƒ‚ …
1 0 0 1
p
H) ke1 e2 k2 D 2
 Access full Solution Manual only here

http://www.book4me.xyz/solution-manual-fourier-transforms-hansen/
14 Vector Spaces

2-18.

kem en k2 D hem en i D kem k2 C ken k2 hem ; en i

en ; em hem ; en i
ˇ 1 i mx ˇ2
Z ˇ ˇ Z 
1
kem k2 D ˇ p e ˇ dx D dx D 1

ˇ 2 ˇ
 2
Z  ˇ
1 i.m n/x 1 1 i.m n/x ˇ
ˇ
hem ; en i D e dx D e
 2 2 i.m n/ ˇ

8
sin .m n/ <1; m D n
D D
.m n/ :0; otherwise
8
<0; mDn
H) kem en k D p
: 2; otherwise

This is the same as the distance between ex and ey in the plane.

2-19. (a) In R4 , four vectors are needed for a basis.

(b) The four chosen vectors must be linearly independent in order to comprise a basis.

2-20. By the triangle inequality,

jhf; gi C hg; f ij  jhf; gij C jhg; f ij D 2 jhf; gij D 2kf k2 kgk2 ;

using the Cauchy-Schwarz inequality (Theorem 2.1).

2-21.

kx C yk2 D hx C y; x C yi D hx; xi C hy; yi C hx; yi C hx; yi

 kxk2 C kyk2 C jhx; yij C ˇhx; yi ˇ
ˇ ˇ
(triangle inequality)
 kxk2 C kyk2 C 2kxkkyk D .kxk C kyk/2 (Cauchy-Schwarz)
H) kx C yk  kxk C kyk

X
2-22. For `1 , kxk1 D jxn j. The first three properties of the norm are obvious. For the
n
triangle inequality, consider x; y 2 `1 , and
N
X N
X
jxn C yn j  jxn j C jyn j (triangle inequality in C)
nD1 nD1
N
X N
X
D jxn j C jyn j
nD1 nD1