Basic Analysis - Proof Bolzano Wierestrass by Least Upper Bound

rakalakalili
Messages
13
Reaction score
0

Homework Statement


Let (an) be a boundedd sequence, and define the set
S= {x\in R : x < a_n for infinitely many terms a_n\}
Show that there exists a subsequence (a_n_k)converging to s = sup S


Homework Equations


This is supposed to be a direct proof of BW using the LUB property, so no monotonic convergence, Cauchy criterion, nested interval property etc...


The Attempt at a Solution


I am having trouble thinking of a way to define the subsequence. What I can show is that, if \epsilon > 0 there are finitely many terms a_n s.t. a_n> s+\epsilon
I thought of defining the subsequence to be a_n_k = min(a_n | a_n > s+\frac{1}{k}) But I was having trouble proving that this subsequence converges to s. I would greatly appreciate a tip or prod in the right direction of defining a subsequence that will work. Thank you!
 
Physics news on Phys.org
Note that the set S is bounded above, since (a_n) is bounded and any upper bound of
the sequence will be an upper bound of S. Therefore, sup S exists, define s = sup S.
Now, consider an arbitrary ε > 0. Since s + ε cannot be an element of the set S, there must be only a finite number of an such that an > s + ε. But by the definition of the supremum, for any ε > 0 there exists some x ∈ S with s − ε < x. Thus, for any ε > 0 there are an infinite number of terms an with s − ε < an and only a finite number of those terms satisfy an > s + ε. Therefore, we must have an infinite number of an with s−ε<an <s+ε.
Now, to construct a subsequence (a_nk ) → s, consider εk = 1/k. So, start by choosing some an1 so that s−1 < an1 < s+1, from this point on, we choose ank+1 so that nk+1 >nk ands−1/k<ank+1 <s+1/k.We know that we can always do this, since at every step k there are an infinite number of a_n with s−1/k < a_n <s+1/k.
 
That's perfect thank you! I had known that there was an infinite number of terms less than s+ \epsilon, but I did not connect that with making an interval, and simply choosing a term from each interval to form the subsequence. Thanks again.
 
There are two things I don't understand about this problem. First, when finding the nth root of a number, there should in theory be n solutions. However, the formula produces n+1 roots. Here is how. The first root is simply ##\left(r\right)^{\left(\frac{1}{n}\right)}##. Then you multiply this first root by n additional expressions given by the formula, as you go through k=0,1,...n-1. So you end up with n+1 roots, which cannot be correct. Let me illustrate what I mean. For this...
Back
Top