January 31, 2014

Math 361: Homework 1 Solutions

1. We say that two norms | · | and | · |0 on a vector space V are equivalent or comparable if the topology
they define on V are the same, i.e., for any sequence of vectors {xk } and x in V ,

lim |xk − x| = 0 if and only if lim |xk − x|0 = 0.

k→∞ k→∞

(a) Show that | · | and | · |0 are equivalent if and only if there exist C1 , C2 > 0 such that

C1 |x|0 ≤ |x| ≤ C2 |x|0

for any x ∈ V .

“=⇒” We are given that | · | and | · |0 are equivalent, and we want to show that this implies that
∃C1 , C2 > 0 such that C1 |x|0 ≤ |x| ≤ C2 |x|0 .
Note that we can switch the definitions of the norms and get that C3 |x| ≤ |x|0 ≤ C4 |x|, which
means that if we prove there exists C1 |x|0 ≤ |x|, we will also have proven that there exists C3 |x| ≤
|x|0 . This implies (with C2 = C13 ) that there exist C1 , C2 > 0 such that C1 |x|0 ≤ |x| ≤ C2 |x|0 .
Therefore, we only need to show that one of the constants exists.
We do this by contradiction. Suppose that ∀C > 0, ∃x ∈ V such that C|x| > |x|0 . For each such
C, |x| is a constant and hence we can say that ∀C, ∃x ∈ V such that:
 0
1 1 x
C|x| > |x|0 =⇒ C|x| > |x|0 =⇒ C >   (1)
|x| |x| |x|
 0
We construct a sequence of vectors {xk } by picking the appropriate x so that for C = k1 , C >  |x|
x
.
In other words, for arbitrary k, we choose C = k1 and xk such that k1 > |xk |0 and |xk | = 1 ( |x|
x
is
a unit vector for each k under | · |).
Now, since limk→∞ k1 = 0 and k1 > |xk |0 for each k, we deduce that limk→∞ |xk |0 ≤ 0. Since
the norm is always positive, we know that limk→∞ |xk |0 = 0. As for the limit of the other norm,
because |xk | = 1 for each k, we can conclude that limk→∞ |xk | = 1.
Because the two norms are equivalent, we know that limk→∞ |xk −x| = 0 ⇐⇒ limk→∞ |xk −x|0 =
0 for any sequence of vectors {xk } in V . But for our sequence we just saw that limk→∞ |xk −0| = 0
and limk→∞ |xk − 0|0 = 1.
Therefore, we have a contradiction and we have shown that there must exist a C such that
C|x| < |x|0 , and by our earlier logic this is sufficient to prove the claim.

“⇐=” We are given that ∃C1 , C2 > 0 such that C1 |x|0 ≤ |x| ≤ C2 |x|0 , and we want to show that
this implies that | · | and | · |0 are equivalent.
We want to prove that limk→∞ |xk − x| = 0 =⇒ limk→∞ |xk − x|0 = 0 for any sequence of vectors
{xk } in V . Note that this will prove the entire if and only if statement because we can just switch
the definition of the norms to give us the other direction.
So, we begin by assuming that for any {xk }, limk→∞ |xk − x| = 0. This means that ∀ > 0, there
exists an N such that if k > N , |xk − x| < . We want to show that ∀0 > 0, there exists an N 0
such that if k > N 0 , |xk − x|0 < .
Next, we observe that xk − x is just some vector in V for any k. We are given that for any x ∈ V ,
C1 |x|0 < |x|. For each k, then, we get some C1,k so that C1,k |xk − x|0 ≤ |xk − x|.

1
 Now we fix 0 , and then we set k = 0 C1,k for each k. We get N such that for k > N , |xk − x| <
k = 0 C1,k . Now we have for N 0 = N :

0 C1,k
C1,k |xk − x|0 ≤ |xk − x| < 0 C1,k =⇒ |xk − x|0 < =⇒ |xk − x|0 < 0 (2)
C1,k

Therefore, we have shown that for k large enough, the value |xk − x|0 will approach 0. This means
that limk→∞ |xk − x| = 0 =⇒ limk→∞ |xk − x|0 = 0, and by our logic earlier, we have hence
proven the if and only if statement.

(b) Show that any norm on Rn is equivalent to the Euclidean norm | · |E . Thus all norms on a finite
dimensional vector space are equivalent, which means topologically there is no difference. (Hint:
for any other norm |·|, first show |·| ≤ C|·|E for some C > 0, which implies the function f (x) = |x|
is continuous with respect to the topology on Rn induced by the Euclidean norm. Recall that the
unit ball in the Euclidean space is compact and verify the inequality in the other direction.)

We begin by constructing the standard basis ei for Rn , and letting | · | be any norm on Rn . We
can write any ~x = (x1 , x2 , . . . , xn ) ∈ Rn as a linear combination of the standard basis elements:
n
X
~x = xi ei (1)
i=1

Now we try to obtain an upper bound for |~x|, the | · | norm of x:

 n  n n
X  X X
|~x| =  xi ei  ≤ |xi ei | = |xi ||ei | (2)
 
 
i=1 i=1 i=1

Note that we used the triangle inequality of | · | in (2). We make the expression in (2) bigger using
the Cauchy-Schwartz inequality:
v v v
n
X
u n u n u n
uX uX uX
|~x| ≤ |xi ||ei | ≤ t |xi |2 t 2
|ei | = |~x|E t |ei |2 (3)
i=1 i=1 i=1 i=1

pPn
Notice that the term multiplying |~x|E in (3) is a constant, so we define C2 = 2
i=1 |ei | .
Therefore, we shown that |~x| ≤ C2 |~x|E . We claim that this implies that f (x) = |x| is continuous
with respect to the topology on Rn induced by the Euclidean norm.
To show this, we need to prove that for all  > 0 there exists δ > 0 such that |x − y|E < δ =⇒
|f (x) − f (y)|E < . Fix  and let δ = C2 .

|f (x) − f (y)|E = ||x| − |y||E (def. of f ) (4)

≤ |x − y| (reverse triangle inequality) (5)
≤ C2 |x − y|E (|~x| ≤ C2 |~x|E ) (6)

≤ C2 · = (def. of δ) (7)
C2

Therefore, f (x) = |x| is continuous with respect to the Euclidean norm on Rn . We also know that
the unit ball in Rn is compact—that is, closed and bounded by the Heine-Borel theorem. We let
B be the unit ball, i.e., B = {x ∈ Rn : |x|E = 1}.

2
 Since | · | is continuous on Rn and B is compact, it must be the case that for all y ∈ B there exists
p ∈ B such that |y| ≥ |p|. Now, we can scale any ~x ∈ Rn using the Euclidean norm so that they
appear within B:  
~x  ~x 
~x ∈ Rn =⇒ ∈ B =⇒   ≥ |p| (8)
|~x|E |~x|E 
 
For a given ~x, |~x|E is just a constant, so if we look at  |~x~x|E , we can just move the constant outside
 
the norm:  
 ~x  1
 |~x|E  = |~x|E |~x| =⇒ |~x| ≥ |~x|E · |p| (9)
 

Finally, taking C1 = |p|, we have that for any ~x ∈ Rn , |~x| ≥ C1 |~x|E . Together with our earlier
upper bound for |~x|:
C1 |~x|E ≤ |~x| ≤ C2 |~x|E (10)
Based on our work in part (a) of the question, this implies that | · | is equivalent to | · |E , i.e., any
norm in Rn is equivalent to the Euclidean norm.

(c) Consider the norms

Z 1
|f |L1 = |f (t)|dt and |f |C 0 = max {|f (t)|}
0 t∈[0,1]

on the space of C 0 ([0, 1]) of continuous functions f : [0, 1] → R. Show that the two norms are not
equivalent. (Note that C 0 ([0, 1]) is a vector space of infinite dimension.)
Consider the following sequence of functions {f (t)k } defined by f (t)k = tk . Consider the following
two limits:
Z 1 Z 1 1
k k tk+1  1
lim |f (t)k |L1 = lim |t |dt = lim t dt = lim = lim =0 (1)
k→∞ k→∞ 0 k→∞ 0 k→∞ k + 1 0 k→∞ k + 1

lim |f (t)k |C 0 = lim max {|tk |} = lim 1 = 1 (2)

k→∞ k→∞ t∈[0,1] k→∞

However, these limits imply that for x = 0, limk→∞ |f (t)k − x|L0 6= limk→∞ |f (t)k − x|C 0 , which
means by the definition that the two norms are not equivalent.

2. Solve Pugh’s Chapter 5, Problem #9, 15, 16, 18.

9) Give an example of two 2 × 2 matrices such that the norm of the products is less than the product
of the norms.
Consider the following two matrices:
   
1 0 0 0
A= B= (1)
0 0 0 1

First, let us calculate kAk. Let ~x ∈ R2 .

     
 1 0 x1 
 = sup  x1  = sup x1 2 = sup x1 = 1
  p
kAk = sup |A~x| = sup  (2)
x|≤1| 0 0 x2 0
  
|~
x|≤1 |~ |~
x|≤1 |~
x|≤1 |~
x|≤1

Note that if ~x = (1, 0), then |~x| = 1 and we are restricted to vectors for which |~x| ≤ 1.

3
 Now, let us calculate kBk. Let ~x ∈ R2 .
     
 0 0 x1   0  p
kBk = sup |B~x| = sup    = sup   = sup x2 2 = sup x2 = 1 (3)
|~
x|≤1 x|≤1| 0 1
|~ x2  |~x|≤1  x2  |~x|≤1 |~
x|≤1

Note that if ~x = (0, 1), then |~x| = 1 and we are restricted to vectors for which |~x| ≤ 1.

Now that we have kAk and kBk, we can calculate the product of the norms:

kAk · kBk = 1 · 1 = 1 (4)

Next, we find the norm of the product. First we must find what A · B is:
     
1 0 0 0 0 0
A·B = · = (5)
0 0 0 1 0 0

Now, we finally calculate kA · Bk. Let ~x ∈ R2 .

     
 0 0 x1   0 
kA · Bk = sup |(A · B)~x| = sup  = sup   = sup 0 = 0 (6)
x|≤1| 0 0 x2 x|≤1 0
  
|~
x|≤1 |~ |~ |~
x|≤1

According to (4) and (6), we have found two 2 × 2 matrices A and B such that kA · Bk < kAk · kBk.

15) Show that the partial derivatives of the function

 xy

if (x, y) 6= (0, 0)
f (x, y) = x2 + y 2
 0 if (x, y) = (0, 0)

exist at the origin, but the function is not differentiable there.

First we show that the partial derivatives exist at (0, 0):
h·0
∂f f (h, 0) − f (0, 0) h2 +0 −0 0
(0, 0) = lim = lim = lim = lim 0 = 0 (1)
∂x h→0 h h→0 h h→0 h h→0

0·h
∂f f (0, h) − f (0, 0) 0+h2 −0 0
(0, 0) = lim = lim = lim = lim 0 = 0 (2)
∂y h→0 h h→0 h h→0 h h→0
That is, the partial derivatives exist at the origin and both are equal to 0 there. Now we must show
that, despite this, f is not differentiable at the origin.
Assume that f is differentiable. Let h = (h1 , h2 ). For some linear transformation A of R2 into R,
since f is differentiable:
|f (~x + ~h) − f (~x − A~h)|
lim =0 (3)
~
h→0 |~h|
Since we assumed f is differentiable at (0, 0), A must be equal to the matrix determined by its
partial derivatives in (1) and (2).
h i
A(0,0) = ∂f ∂f  
∂x ∂y = 0 0 (4)
(0,0)

Acting A(0,0) on ~h, we get:  

 h1
A(0,0)~h = 0

0 · =0 (5)
h2

4
 Plugging in ~x = (0, 0) and (5) into (3):

|f (0 + h1 , 0 + h2 ) − f (0, 0) − 0| | h21 h2 2 | h1 h2
lim p = lim ph1 +h2 = lim 2 2 3/2 (6)
2 2 2 2
~
h→0 | h1 + h2 | ~
h→0 | h1 + h2 | h→0 (h1 + h2 )
~

According to (3), the limit in (6) has to equal 0 from any direction. We choose h = (t, t) (approaching
along the line y = x) and get:
h1 h2 t2 1
lim 2 2 = lim 3/2 3
= lim 2/3 6= 0 (7)
h→0 (h1 + h2 )
~ 3/2 t→0 2 t t→0 2 t

Therefore, our assumption of the limit in (3) existing and equaling zero was wrong, which means
that despite the fact that the partial derivatives of f exist at the origin, f is not differentiable there.

16) Let f : R2 → R3 and g : R3 → R be defined by f = (x, y, z) and g = w where

w = w(x, y, z) = xy + yz + zx
x = x(s, t) = st y = y(s, t) = s cos t z = z(s, t) = s sin t
(a) Find the matrices that represent the linear transformations (Df )p and (Dg)q where p =
(s0 , t0 ) = (0, 1) and q = f (p).

First, note that f (s, t) = (x, y, z) = (st, s cos t, s sin t). Now, let us find the partial derivatives
Dj fi of f :
∂fx ∂x ∂fx ∂x
D1 f1 = = =t D2 f1 = = =s (1)
∂s ∂s ∂t ∂t
∂fy ∂y ∂fy ∂y
D1 f2 = = = cos t D 2 f2 = = = −s sin t (2)
∂s ∂s ∂t ∂t
∂fz ∂z ∂fz ∂z
D1 f3 = = = sin t D2 f3 = = = s cos t (3)
∂s ∂s ∂t ∂t
According to theorem 9.17 in Rudin, we construct Df and then evaluate at (0, 1):
     
D1 f1 D2 f1 t s 1 0
(Df )(0,1) = D1 f2 D2 f2  cos t −s sin t = cos(1) 0 (4)
D1 f3 D2 f3 (0,1) sin t s cos t (0,1) sin(1) 0

Next, we note that g(x, y, z) = w = xy + yz + zx. Now we proceed similarly to find the partial
derivatives Dj gi of g:
∂g
D1 g = =y+z (5)
∂x
∂g
D2 g = =x+z (6)
∂y
∂g
D3 g = =y+x (7)
∂z
As before, we construct Dg and evaluate at q = f (p) = f (0, 1) = (0 · 1, 0 · cos(1), 0 · sin(1)) =
(0, 0, 0):
h i
∂g ∂g ∂g    
(Df )(0,0,0) = ∂x ∂y ∂z = y + z x + z y + x (0,0,0) = 0 0 0 (8)
(0,0,0)

Therefore, we have constructed the two matrices (Df )(0,1) and (Dg)(0,0,0) .

5
 (b) Use the Chain Rule to calculate the 1 × 2 matrix [∂w/∂s, ∂w/∂t] that represents (D(g ◦ f ))p .
According to the Chain Rule:

∂w ∂w ∂x ∂w ∂y ∂w ∂z ∂w ∂w ∂x ∂w ∂y ∂w ∂z
= + + = + + (9)
∂s ∂x ∂s ∂y ∂s ∂z ∂s ∂t ∂x ∂t ∂y ∂t ∂z ∂t

This is equivalent to multiplying the two matrices we found in part (a):

   
h i D1 f1 D2 f1 1 0
∂g ∂g ∂g  
(D(g ◦ f ))(0,1) = ∂x ∂y ∂z
D1 f2 D2 f2  = 0 0 0 cos(1) 0 (10)
(0,0,0)
D1 f3 D2 f3 (0,1) sin(1) 0
 ∂w ∂w
  
=⇒ (D(g ◦ f ))(0,1) = ∂s ∂t (0,1)
= 0 0 (11)
In other words, both partial derivatives are equal to 0.

(c) Plug the functions x = x(s, t), y = y(s, t), and z = z(s, t) directly into w = w(x, y, z), and
recalculate [∂w/∂s, ∂w/∂t], verifying the answer given in (b).
First we find w in terms of s and t:

w = xy + yz + zx = (st)(s cos t) + (s cos t)(s sin t) + (s sin t)(st) = (12)

2 2 2
= s t cos t + s sin t cos t + s t sin t (13)
∂w ∂w
Now we find ∂s and ∂t directly and evaluate at (0, 1):

∂w 
= (2st cos t + 2s sin t cos t + 2st sin t)|(0,1) = 0 (14)
∂s (0,1)


∂w 
= ((−2st sin t + 2s cos t) + (−2s sin t sin t + 2s cos t cos t) + (2st cos t + 2s sin t)) (15)
∂t (0,1)
= (−2st sin t + 2s cos t − 2s sin2 t + 2s cos2 t + 2st cos t + 2s sin t)|(0,1) = 0 (16)

Therefore, we have confirmed our answers from (b) by finding the partials directly.

(d) Examine the statements of the multivariable Chain Rules that appear in your old calculus book
and observe that they are nothing more than the components of various product matrices.
Observed.

18) The directional derivative of f : U → Rm at p ∈ U in the direction u is the limit, if it exists,

f (p + tu) − f (p)
∇p f (u) = lim
t→0 t
(Usually, one requires that |u| = 1.)
(a) If f is differentiable at p, why is it obvious that the directional derivative exists in each direction
u?
If f is differentiable at p, we have by definition:

|f (p + h) − f (p) − Ah|
lim =0 (1)
h→0 |h|

6
 R(h)
We define R(h) such that limh→0 |h| = 0 (as in Rudin), and then we can write that:

f (p + h) − f (p) − A(h) = R(h) =⇒ f (p + h) = f (p) + A(h) + R(h) (2)

Now, we let h = tu and we rewrite the directional derivative:

f (p + tu) − f (p) f (p) + A(tu) + R(tu) − f (p) A(tu) + R(tu)
∇p f (u) = lim = lim = lim (3)
t→0 t t→0 t t→0 t

Now, since R(h) R(tu)

|h| → 0 as h → 0, we have that |tu| → 0 as tu → 0. Since tu → 0 as t → 0, we
can write that:
R(tu) R(tu) R(tu)
lim = lim = lim =0 (4)
t→0 t tu→0 tu tu→0 |tu|
Therefore, according to (3) and (4), we have that:

A(tu) + R(tu) A(tu) tA(u)

∇p f (u) = lim = lim = lim = A(u) (5)
t→0 t t→0 t t→0 t
That is, the directional derivative both exists and equals A(u).

(b) Show that the function f : R2 → R defined by

 3
 x y if (x, y) 6= (0, 0)
f (x, y) = x + y 2
4

0 if (x, y) = (0, 0)


has ∇(0,0) f (u) = 0 for all u but is not differentiable at (0, 0).

We begin by showing that the directional derivative of f at the point p = (0, 0) is 0 for any
u = (u1 , u2 ):

f ((0, 0) + t(u1 , u2 )) − f (0, 0) f (tu1 , tu2 ) − 0

∇(0,0) f (u) = lim = lim = (1)
t→0 t t→0 t
(tu1 )3 (tu2 )
(tu1 )4 +(tu2 )2 t3 u1 3 · tu2 t4 u1 3 u2
= lim = lim = lim = (2)
t→0 t t→0 (t4 u1 4 + t2 u2 2 )t t→0 t3 (t2 u1 4 + u2 2 )

tu1 3 u2 0
= lim = 2 =0 (3)
t→0 t u1 4 + u2 2
2 u2
That is, the directional derivative is 0 for any u. Now we show that this does not indicate that
f is differentiable at (0, 0).
First we show that the partial derivatives exist at (0, 0):
h3 ·0
∂f f (h, 0) − f (0, 0) h4 +0 −0 0
(0, 0) = lim = lim = lim =0 (4)
∂x h→0 h h→0 h h→0 h
0·h
∂f f (0, h) − f (0, 0) 0+h2 −0 0
(0, 0) = lim = lim = lim =0 (5)
∂y h→0 h h→0 h h→0 h

That is, the partial derivatives exist at the origin and both are equal to 0 there. Now, assume
that f is differentiable and let h = (h1 , h2 ). For some linear transformation A of R2 into R,
since f is differentiable:
|f (~x + ~h) − f (~x − A~h)|
lim =0 (6)
~
h→0 |~h|

7
We construct A(0,0) using the partial derivatives from (4) and (5):
h i
A(0,0) = ∂f ∂f  
∂x ∂y = 0 0 (7)
(0,0)

Therefore, we see that when we apply A to any vector ~h, we will get that A(0,0)~h = 0.
Plugging this result and ~x = (0, 0) into (6), we get:
3
|f (0 + h1 , 0 + h2 ) − f (0, 0) − 0| | h41 h2 2 | h1 3 h2
lim p = lim ph1 +h2 = lim p (8)
~
h→0 | h1 2 + h2 2 | ~ 2
h→0 | h1 + h2 |
2 ~ 4 2 2
h→0 (h1 + h2 ) h1 + h2
2

According to (6), the limit in (8) has to equal 0 from any direction. We choose h = (t, t2 )
(approaching along the line y = x2 ) and get:

h1 3 h2 t3 · t2 t5
lim p = lim p = lim √ = (9)
~ 4 2 2 2 t→0 (t4 + (t2 )2 ) t2 + (t2 )2 t→0 2t4 t2 + t4
h→0 (h1 + h2 ) h1 + h2

t5 1 1 1
= lim √ = lim √ = √ = 6= 0 (10)
t→0 2t5 1 + t2 t→0 2 1 + t2 2 1 2

Therefore, our assumption of the limit in (6) existing and equaling zero was wrong, which
means that despite the fact that the directional derivative of f exists at the origin, f is not
differentiable there.