assume the nested interval property is true and use it to prove the completeness property in R
here is what I know so far. there exists a an element in all these nested closed intervals. a1<a2<a3<..an<x<bn<...b3<b2<b1
What is the definition of completeness you are using? "Every set A bounded above has a l.u.b./sup"?
yes every non empty set of reals that has an upper bound also has a supremum in R. I got that the nested inervals provides that the intervals are non empty. do i use the fact that this element in all the sets must be an upper bound of all the a_n?
You can do this constructively by writing down the intervals such that intersection of them all is x = lim bn, and that number is the sup. Or you can argue by contradiction. I think I would use the constructive method. It's not hard to see that's possible to write down a strictly monotonically increasing series (an); just be careful with the (bn) so that the intervals are nested.
or rather "in the intersection of them all is lim bn = x and that number is the sup"
actually, no, we need this to be sharper and have lim an = lim bn = x and then show that x = sup A. Of course in the construction (an) are all in A and the (bn) are all greater than every member of A.
and in that case the (an) is increasing, but not necessarily strictly.
i like the use of the montone cnvg thm. but my prof won't allow that because it is in a future chap. so contradiction would have to be the way to go. Assume that not every non empty set of real that has an upper bound also has a supremum and get a contradiction?
Ok. So yes, take such a set A.
Note first that A must have infinitely many members, other max A would exist and be = sup A
Now let (an) be a strictly increasingly series in A such that ______ and (bn) be a strictly decreasingly series such that bn > a for all a in A such that ______
I'm just trying to figure out exactly what to put in these blanks.
Oh, I see. ...such that lim(an - bn) = 0
but doesn't that mean that we don't have a common element since an and bn are becominmg closer and closer?
I'm not sure exactly what you mean by 'common element' and I'm going to have to chew on this for a couple of minutes to see how to get a sharp proof given where you are in the material. Let me ask you have any other properties from completeness yet that we can use?
I.e., have you shown Completeness <=> P for some statement P ?
...so we can use not P to help us in our proof by contradiction.
we have got the axioms for reals, the completness property and the nested interval prop. we know that the completeness property implies the nested interval property.
If the intervals are nested we know that they share at least one common element.
By hypothesis yes. The question is how do we construct the intervals so that number is sup A.
Let's try this. Let B(x,r) be the closed ball of radius r i.e., B(x,r) = [x-r, x+r]
but if lim (an - bn) = 0 our contradiction because as the distace between an and bn shortens it makes that common element impossible.
Yes, but how do we know there are such series (an) and (bn)? So that's what I'm going to try and construct now.
Now for every integer n there is an upper bound for A, call it x_n (not in A of course), such that |x_n - a| < 1/n for some n. Let the (bn) be a decreasing sequence of such x_n.
Then B(b_n,1/n) intersects A by construction. Let a_n = max(a_{n-1}, a) for an arbitrary \[a \in B(b_n,1/n)\] and a1 = some arbitrary member of A.
Then the interviews [an,bn] are nested and by construction \[|b_n -a_n| < 1/n \rightarrow 0\]
I've been waiting 7 minutes for that to post!! lag on this site is terrible.
Now, the intersection of all the interviews in non-empty by hypothesis; call it {x}. Now show x must necessarily be the sup.
So actually, this is a constructive argument after all.
Draw a diagram: this will help a lot.
Join our real-time social learning platform and learn together with your friends!