View Full Version : Limit of a sequence
matrix_204
Mar7-05, 10:17 PM
i had this problem in my book that i wasn't able to do. I kinda had the idea of what it involved but just wanted to clear it up with you guys.
So the problem is:
Suppose that an ->L and bn ->L. Show that a1,b1,a2,b2,... converges to L.
So here it seems to me like i can obviously define a new sequence cn, like that in the pinching theorem. But using other definitions(epsilon, etc.) of sequences, how do i come up with a proof of this.
Fix an epsilon. That a_n and b_n converge to L implies there is some N (take each sequence individually, choose an M and M' for each, take the bigger of the two and double that) such that whenever n>N both a_(n/2) and b_(n/2) are within epsilon of L (and therefore c_n is within epsilon of L).
matrix_204
Mar8-05, 09:35 PM
could someone check this proof for me and tell me what is missing as i m not sure if i know anymore:(problem stated above)
Proof:
If a_n ->L and b_n ->L for some L, then for any eps>0 there is a K such that for all n, if n>K, then |a_n - L|<eps/2 and |L - b_n|<eps/2.
Since n>K, |a_n - b_n|</= |a_n - L| + |L - b_n|<eps/2 + eps/2 =eps.
(now to show that it is bounded)
If eps=1, then |a_n - b_n|<1 for some n>K.
This means |a_n - b_K+1|<1 for all n>K.
Thus {a_n:n>K} and {b_n:n>K} are bounded.
So a_n ->L and b_n->L.
-------
I don't think you have to show anything is bounded. That will follow automatically once you prove the limit is L. What you want to do is define the sequence by \{c_n\}_{n\in \mathbb{N}} by c_n=a_{[(n+1)/2]} if n is odd and c_n=b_{(n/2)} if n is even. Now fix an epsilon > 0. There are N and M such that whenever n>N, a_(n+1)/2 is within epsilon of L and whenever n>M b_(n/2) is within epsilon of L. Take the bigger of the two, let's say it is P. Then whenever n>P, both a_(n+1)/2 and b_(n/2) are within epsilon of L. That is, whenever n>P, c_n is within epsilon of L.
(sorry for the sloppier version of this above)
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.