Suprema + Density Practice

1.3.1, 1.3.2, 1.3.3, 1.3.8, 1.3.11, 1.4.1, 1.4.2, 1.4.3, 1.4.4, 1.4.5, 1.4.8

1.3.1

Question

a. Write a formal definition in the style of the suprema for the infimum or greatest lower bound of a set.
b. Now, state and prove a version of suprema lemma for greatest lower bounds.

(a): Here it is:

Infimum, Greatest Lower Bounds

Some sR is the greatest lower bound for a set AR if it meets:

  1. s is a lower bound for A.
  2. If b is any lower bound for A, then sb.
    Denote s=inf(A).

(b):

Infima Lemma

Suppose sR is a lower bound for a set AR. Then s=inf(A) iff, for all ε>0 there is an element aA such that s+ε>a.

Proof
(): Suppose s=inf(A). Then s is obviously a lower bound for A. Furthermore, we know for all other lower bounds b for A that sb. Now, let ε>0. Notice that s+ε>s, so since s is the greatest lower bound, then we know that s+ε cannot be a lower bound for A. Hence, there is some aA such that a<s+ε.

(): Suppose that ε>0aA(s+ε>a). We'll show the two properties of inf:

  1. It is given that s is a lower bound.
  2. Let b be any lower bound for A, so aA(ab). We want to show that sb. Notice that if we assume that b>s is a greater lower bound, then clearly bs>0 so then plugging into our given then aA where s+(bs)>a So then b>a. But we know that b is a lower bound and found that b>a! That's a contradiction, so then our assumption is false. Hence bs as required.

1.3.2

Question

Give an example of each of the following, or state that the request is impossible:
(a): A set B with inf(B)sup(B)
(b): A finite set that contains its infimum but not its supremum
(c): A bounded subset of Q that contains its supremum but not its infimum.

Proof
(a): Simple. Have B={2}. Clearly 2=inf(B)=sup(B).

(b): Impossible. Finite sets with only finite elements must have to contain their elements (since then you can order the elements into an ordered list, and the infimum/suprema are the endpoints of the list).

(c): This is pretty easy. Chose B={qQ|4q22}. Notice that 2=inf(B)B but 2=sup(B)B since 2 is irrational.

1.3.3

Question

a. Let A be nonempty and bounded below, and define B={bR|b is a lower bound for A}. Show that sup(B)=inf(A).
b. Use (a) to explain why there is no need to assert that greatest lower bounds exist as part of the axiom of completeness.

Let's prove (a):

Proof
Since A is nonempty and bounded below, then b as a lower bound such that aA(ba), and further aA. We immediately can say ba. Since b is a lower bound for A then clearly bB.

Say β=inf(A). Then ε>0aA(β+ε>a) using our infima lemma. Furthermore, we know that β clearly is a lower bound for A.

If we show β is the suprema for B then we are done:
(i) We should show β is an upper bound for B, or that bB(βb). Let bB be arbitrary. Now assume that β<b, for contradiction. Then clearly ε=bβ>0 so then via our infima lemma given, then aA such that β+(bβ)>a so then b>a. But then that means b is not a lower bound for A, so our assumption is false. Thus bB(βb).
(ii). We should show β is specifically the least upper bound for B, or that βb for any upper bound b for B. Let b be that arbitrary upper bound for B. Assume that β>b. Then βB since b is an upper bound for B. then β is not a lower bound for A, so aA where β>a. But β is a lower bound for A since it's the infimum, so this is a contradiction. Thus βb so β is the smallest upper bound.

(b) Let A be nonempty and bounded below. Then (a) suggests that sup(B)=inf(A). But clearly by the axiom of completeness, because B is nonempty (A is bounded below so b exists as a lower bound for A so bB exists) and bounded above (if there were always larger lower bounds for A, then A wouldn't be bounded below), then sup(B) exists. Then clearly inf(A) exists by consequence.

1.3.8

Question

Compute, without proofs, the suprema and infima (if they exist) of the following sets:
a. {mn|m,N;m<n}
b. {(1)mn|m,nN}
c. {n3n+1|nN}
d. {mm+n|m,nN}

(treat them like 'limits')
a. inf(A)=0. sup(A)=1.
b. inf(B)=1. sup(A)=1.
c. inf(B)=13. sup(A)=14.
d. inf(B)=0. sup(A)=12.

1.3.11

Question

Decide if the following statements about suprema and infima are true or false. Give a short proof for those that are true. For any that are false, supply an example where the claim in question does not appear to hold.
a. If A,B are nonempty, bounded, and satisfy AB, then sup(A)sup(B).
b. If sup(A)<inf(B) for sets A,B, then cR satisfying a<c<b for all aA,bB
c. If cR where a<c<b for all aA,bB, then sup(A)<inf(B).

a. True.

Proof
A,B are bounded and nonempty so then both their suprema and infima exist. Let α=sup(A),β=sup(B). If we assume for contradiction that sup(A)>sup(B) then ε=αβ>0 so then aA where α(αβ)<a, or just β<a. But since aA and AB then aB. Thus then clearly β is not an upper bound for B, which is a contradiction as β is the suprema of B. Therefore, we must have sup(A)sup(B).

b. True.

Proof
Suppose sup(A)<inf(B), denoted α<β for simplicity here. Choose c=α+β2R. Let aA,bB be arbitrary.

First let's show a<c. Notice that a<sup(A)<α+β2=c. Similarly c=α+β2<inf(B)<b by the definitions of sup and inf.

c. False. You can have equality. Let A={qQ|q2<2} and B={qQ|q2>2}. Here you can always choose some c=a+b2 where aA,bB and have a<c<b. However, clearly sup(A)=2=inf(B).

1.4.1

Question

Recall I stands for the irrational numbers.
a. Show that if a,bQ then ab,a+b are both elements of Q as well.
b. Show that if aQ and tI then a+tI and atI as long as a0.
c. (a) can be summarized by saying Q is closed under addition and multiplication. Is I closed under addition and multiplication? Given two irrationals s,tI what can we say about s+t and st?

Proof
a. Here a=mana and b=mbnb where all the miZ and niN. Then:

a+b=mana+mbnb=manb+nambnanb

Now clearly nanbN by the closure of multiplication over N. Similarly manb,nambZ by the same idea, and by closure of addition on Z thne manb+nambZ. Therefore then a+b has the correct form of a rational number so a+bQ. Furthermore:

ab=manbmbna=mambnanb

Here mambZ by closure via multiplication so then ab is also of the right form. abQ.

b. Use a=mn and tI, assume for contradiction that a+tQ. Then it's rational, so a+t=mn. But then:

a+t=mnt=mnmnQ

because Q is closed under addition and scalar multiplication. As such then tQ which contradicts tI. Hence a+tQ so a+tI. Similarly for at if we suppose a0 then m0. Thus:

at=mnt=man=mnnmQ

which contradicts tQ. Thus atI.

c. I is not closed under addition nor scalar multiplication. For multiplication we know that 2I via Irrationality of sqrt(2). But:

(2)(2)=2I

Thus I isn't closed under scalar multiplication. Furthermore, we know 2+1I (without proof here, but it's easy to verify) but:

2+(2+1)=1I

so I isn't closed under addition neither.

1.4.2 (check, didn't use s+1/n)

Question

Let AR be nonempty and bounded above, and let sR have the property that for all nN, s+1n is an upper bound for A and s1n is not an upper bound for A. Show s=sup(A).

Proof
If we show ε>0aA(sε<a) then we are done by the Suprema Lemma. Let ε>0. Notice that since ε>0 then the archimedian property gives that nN where 1n<ε. For later, this has:

1n<εε<1nsε<s1n

Then notice that since s1n is not an upper bound for A then aA where s1n<a. But:

sε<s1n<asε<a

1.4.3

Question

Prove that n=1(0,1n)=. Notice that this demonstrates that the intervals in the Nested Interval Property must be closed for the conclusion of the theorem to hold.

Proof
Assume for contradiction . Then an=1(0,1n). That means that for all nN then a(0,1n), so 0<a<1n. Notice that this is just the opposite of the The Density of Q in R#^46383a, where we've shown aRnN(1n>a), so we've hit a contradiction (since we know the Archimedean property is true). Thus, the set must be the empty set.

1.4.4 (check)

Question

Let a<b be real numbers and consider T=Q[a,b]. Show sup(T)=b.

Proof

  1. (b is an upper bound). We want to show that tT(bt). Let tT and assume instead b<t. Since tQ[a,b] then mZ,nN where t=mn. Furthermore, we know that atb. But look! tb cannot be true while t>b! Thus our assumption is false so then bt as required.
  2. (b is the least upper bound). We want to show that for all upper bounds u for T that bu. It's true that ub, because if we had assumed that u<b then choose bT to show that u is not a valid upper bound.

1.4.5

Question

Using Suprema + Density Practice#1.4.1, prove The Density of Q in R#^ca78a0 by considering the real numbers a2 and b2.

Proof
Given two reals a,bR, then since Q is dense in R and a2,b2R then qQ where:

a2<q<b2

adding 2 to both sides:

a<q+2<b

Now choose t=q+2. Notice that by Suprema + Density Practice#1.4.1 that tI because qQ.

1.4.8

Question

Give an example of each or state that the request is impossible. When impossible, provide a compelling argument for why it's the case.
a. Two sets A,B with AB= and supA=supB where supAA and supBB
b. A sequence of nested open intervals J1J2J3 with n=1Jn nonempty but containing only a finite number of elements.
c. A sequence of nested unbounded closed intervals L1L2L3 with n=1Ln= where the unbounded closed intervals are of the form [a,) where aR.
d. A sequence of closed bounded (not necessarily nested) intervals I1,I2,I3,... with the property that n=1NIn for all NN, but n=1In=

Proof
a. A=Q(0,1) and B=I(0,1). Here AB=, and supA=supB=1A,B.
b. Define Ji=(ai,bi) for all iN where A={aj|jNi,j(ai<bj} and B={bj|jNi,j(ai<bj)}. Notice that supA,infBn=1Jn.
c. Have each Li be of the form [i,). Clearly we have the nested interval property. Now assume for contradiction there was some αn=1Ln to our contrary. Then since αL1 then 1α. Indeed for any j we have that jα. But since αR then it must bounded between two integers by the Archimedian Property:

nα<n+1

But then α<n+1N which implies that αLi+1αn=1Ln which is a contradiction.
d. This is impossible. Assume you could do such a thing. Then there is αn=1NIn for any nN. We have to have αn=1In so then Ii such that αIi. But since iN then clearly αn=1iIn as our given, so then that's our contradiction.