Sei sulla pagina 1di 5

Week 1: Selected Solutions

Rohan Ghanta June 30, 2011

I am going to present the solutions to some selected problems, namely those problems that will be used for calculating your marks on the assignment. The rst chapter in Rudin lays the foundation for modern analysis through introducing you to the real and complex systems. Among other things, you should have built some intuition for why the rationals, by the themselves, are inadequate, and the real number system is necessary to rigorously do calculus. I will show the formal solution, and (if necessary) say a little bit about how to develop intuition for the formal solution.

Problem 6
Fix

(a)

b > 1. If m, n, p, q

are integers,

n > 0, q > 0,and r = (bm ) n = (bp ) q


1 1

m n

p q , prove that

so that

br = (bm ) n

, or writing rational exponents,is a well-dened convention.

Solution:
use to say

There are two things to recognize. First note that it is legal for

p mq pn m n = q nq = nq , so note that nq = want to use this fact to show that the two expressions are equivalent.

(bm ) n = (b n ).
1

Also,

pn.

We

(bm ) n = ([(bm ) n ]nq ) nq = (bmq ) nq = (bpn ) nq = (bp ) q

(b)

Prove that

br+s = br bs

if

and

are rational.

Solution:

The technique here is to use what we know about integer exponents

and then extend that to rational exponents. Let me rst state another fact.

Lemma: If r, s Z(the eld of Proof: The proof of the above


dealing with integers. Since Then,

integers), then note that

br+s = br bs .

lemma is very straightforward, because we are

and

are rational, let me write


m n p q mq nq

r=

(br bs ) = (b b ) = (b

np nq

) = (bmq )
1

1 nq

p m n and s = q , where m, n, p, q Z. 1 (bnp ) nq . Then by the Corollary

to

Theorem 1.21
1 nq 1 nq

and the fact that


1 nq

nq

is an integer, we know that

(br bs ) =

(bmq ) (bnp ) = (bmq bnp ) mq and np are integers, we


(c)
and If

. Now using the above


1

Lemma
m p

and the fact that

see

(br bs ) = (bmq+np ) nq = b n + q = br+s . bt ,


where

x is real, dene B(x) t x. Prove that br =

to be the set of all numbers sup B(r) when

is rational

is rational. Hence it makes sense

to dene

bx = sup B(x)
for every real

x. b > 1.
The rst part is very easy to

Solution:

This is why we specied that

br B(r), as r is a rational. Now, for any other rational t such that t < r , we have r t > 0, which means brt > 1 (since b > 1). Then, brt = br bt > 1. This implies that br > bt . So, t r r for any rational t r , we have b b . Thus, b = sup B(r).
show, and we will use First note that

part(b).

(d)

Prove that

bx+y = bx by

for all real

and

y.

Proof:
lems.

This should not be any more dicult than one of the worksheet probFrom

part(c), it suces to prove that sup B(x + y) = [sup B(x)] [sup B(y)]. First I will prove that sup B(x + y) [sup B(x)] [sup B(y)]. Given > 0, choose b B(x) such that b > sup B(x) and choose b B(y) such that b > sup B(y) . Such b , b exist by the denition of supremum (and we can nd those , Q, because Q is dense in R). So, sup B(x + y) ba+b = ba bb > [sup B(x) ] [sup B(y) ]= (sup B(x))(sup B(y)) sup B(x) sup B(y) 2 . Now since is arbitrary (that means we can send 0), we see that sup B(x + y) [sup B(x)] [sup B(y)]. Now I will prove that sup B(x + y) [sup B(x)] [sup B(y)]. Given > 0, choose + Q such that we can nd b+ B(x + y) and sup B(x + y) < b+ . Such b+ can be found, because of the denition of supremum and the fact that Q is dense in R(so we can nd the necessary and ). sup B(x + y) < b+ = b b [supB(x)] [sup B(y)]. Since is arbitrary, we see that sup B(x + y) [sup B(x)] [sup B(y)]. Thus, we x+y conclude that sup B(x + y) = [sup B(x)] [sup B(y)], which means b = bx by .
Note that there may be other techniques to argue the point, but give special attention to the technique I have used in this proof, because it is a precursor to the

technique you will encounter in Chapter 3 that is the foundation of

any analytic argument. There is an important point to take away from this problem. Just like we

have proven properties of the rationals through thinking about the rationals as a quotient of two integers, we prove properties of the real numbers using the

supremum property
set

and the fact that the rationals

are

dense

in in the

of real-numbers. Because

is dense in

R,

we can construct a sequence of

points in

Q that converge to an element in R,


2

in such a way that the supremum

of the sequence of points in is why the second part of

is a real-number that may not be rational. This holds true.

part(c)

Now is also a good time to reect on the relationship between maximum and supremum. Why is it that supremum of a set is a stronger characterization than the maximum of a set? For example, can you think of a set where the Also, while maximum doesn't exist, but the supremum exists? Do you think there exists a set where the supremum doesn't exist but a maximum exists? you're reading this, you are probably studying topology in Chapter 2 of Rudin, so here is another small question: What topological conditions can you put on a set to ensure that its maximum and supremum are identical?

Worksheet Problem 1
Prove or disprove the following of the supremum:

(a)

For a non-empty subset

holds: inf A

= sup{A},

where

A R, which is bounded A := {x| x A}.

below, the following

Proof:

When you are trying to prove expressions like the one above, you want

to rst show that the entity sup{A} exists! such that

Since A is bounded below, l l x for all x A. Then, we know l x for all x A and {x} A. Thus, A is bounded above by l. Theorem 1.19 (also called the Completeness Axiom in other texts) tells us that sup{A} exists. Let me name a0 =sup{A}. So, a0 x for all x A. Then, we see that for

A. Now b x for all x A. Then b x for all (x) A. But a0 =sup{A}, so b a0 , which implies b a0 . So, I have shown a0 = sup{A} = inf A.
I have just shown that is the lower bound of I must show that it is the greatest lower bound. Suppose we have Through my proof, I have shown a corollary to why?

a0 x

x A.

a0

Theorem 1.19

in Rudin by Do you see

showing that if a set is bounded from below, it has an inmum.

(b)

For two nonempty subsets of Sup(A

R,

call them

and

B:

B) = Sup(A) Sup(B)

where

A B = {x y| x A and y B}.
It suces to give a counterexample to disprove the above statement. and and

Disproof:

A and B with negative elements so that A = {5, 3, 2} B = {6, 3}. Then, Sup(A B) = (5)(6) = 30, but Sup(A) = 3 Sup(B) = 3. But 30 = 9.
Consider a sets

It is more interesting to see under what conditions the above statement works. If

A and B
6(d)

didn't contain any negative elements, would the above statement hold?

How would you go about proving it? You may want to look at my technique in or the proof of statement below.

(c)

For two sets like in

part(b),
sup (A

+ B) = sup A + sup B y B}.

where

A + B := {x + y| x A

and

First I will show that sup (A + B) sup A + sup B . Given an > 0, a A and b B such that a > sup A and b > sup B . We know that such a and b exist by the denition of a supremum (the least upper bound). So, sup (A + B) a + b > sup A + sup B 2 . Since is arbitrary, we have sup (A + B) sup A + sup B . choose Now I will show that sup (A+B)

Proof :

sup A+ sup B . Given > 0, choose (a+b) A + B such that a + b > sup (A + B) . Then we have sup A + sup B < a + b sup (A + B) . But since is arbitrary, we have sup (A + B) sup A + sup B and we conclude that sup (A + B) = sup A + sup B .

Worksheet Problem 3
We consider the eld automorphism

: R R.

(a)

Show that if

r Q,

then

(r) = r.

Proof:

To prove this property for rationals, I am going to rst prove that

the property holds for all integers. This strategy should not be too surprising, because this is the same strategy I used in

6(b)

from

[Rudin].

So, for all

m Z, I will show (m) = m using the fact that (1) = 1. So, (m) = (11 + 12 + 13 + ... + 1m ) = (11 ) + ... + (1m ) = m. Now, consider any rational m number r Q such that r = n , where m, n Z. Note that given any n Z and 1 1 1 1 1 n = 0, we have (1) = (n n ) = (n) ( n ) = 1. Thus, ( n ) = (n) = n for 1 m m any n Z. So ( ) = (m) ( ) = n n n . Thus, when r Q, we have (r) = r .
(b)
If

xR

and

x > 0,

then

(x) > 0.
exists a unique

Proof:

Since Thus,

y 2 = x.
(c)

x > 0, by Theorem 1.21, there (x) = (y 2 ) = ((y))2 > 0.


part(b)
that if

y R,

such that

Deduce using

a<b

then

(a) < (b).

Proof: (d)

This shouldn't be too hard. If

we see that

a < b, then b a > 0, and from part(b), (b a) = (b) (a) > 0. So, (a) < (b). x R,
we have

Prove that for all

(x) = x. x
arbitrarily

Proof:

The intuition behind proving this is that we sandwich

close between two rational numbers (such two rational numbers exist, because

Q is dense in R). Given q1 < x < q2 and q2 q1 <

> 0,

we can nd two rationals

. Then, by

part(c),

we know

q1 , q2 such that (q1 ) < (x) < (q2 ).

By

part(a),

we see that

q1 < (x) < q2 . Cthat

Since

is arbitrary, we let

0,

and

we have

(x) x.
is not the identity. I suggest complex conjugation,

(e)

Give an automorphism on

Proof:
on

It helps to think geometrically here.

which is just the reection about the real axis. Interestingly, the automorphism

is still identity here.

Do you think there exists an automorphism on

that doesn't preserve this identity condition on

R?

Potrebbero piacerti anche