What is the proof for the limit superior?

5 replies · 2K views · last reply
NihalRi
Messages
134
Reaction score
12

Homework Statement


Printed homework problem 2: Let (xn) be a bounded real sequence and a in R. Prove that if a > limsup xn, then there exists K in N such that xn < a for n >= K. Hint: let s = limsup xn, show there exists K with Mk < a.


2. Relevant equation
Below is the definition of the limit superior
Typed definition DEF1 of the limit superior: for a bounded-above real sequence (x_n), M_n = sup{x_n, x_{n+1}, ...}, and lim sup x_n is defined as lim M_n as n tends to infinity, assuming that limit exists.

The Attempt at a Solution


I tried to start by considering two cases, case 1 in which the sequence does not converge and case 2 in which the sequence converges and got stuck with the second case.
Handwritten proof notes on limit superior: s = lim sup x_n = lim M_n, with s < a, split into Case 1 (for all n, M_n = s, so some M_k < a) and Case 2 (s = lim x_n).

I know intuitively that there exists a K such that Mk < a is the second case but I can not think of how to show this. I attempted to consider cases again, like monotone increasing, decreasing, or even use the definition of couchy sequence but was not getting anywhere. Is there a way to reach the same conclusion in my second case or is my approach for this proof completely off? I would gladly supply any additional information and would greatly appreciate any help.
 
Attachments
  • Printed homework problem 2: Let (xn) be a bounded real sequence and a in R. Prove that if a > limsup xn, then there exists K in N such that xn < a for n >= K. Hint: let s = limsup xn, show there exists K with Mk < a.
    ques.jpg
    20.9 KB · Views: 781
  • Typed definition DEF1 of the limit superior: for a bounded-above real sequence (x_n), M_n = sup{x_n, x_{n+1}, ...}, and lim sup x_n is defined as lim M_n as n tends to infinity, assuming that limit exists.
    def.jpg
    23.7 KB · Views: 768
  • Handwritten proof notes on limit superior: s = lim sup x_n = lim M_n, with s < a, split into Case 1 (for all n, M_n = s, so some M_k < a) and Case 2 (s = lim x_n).
    try.jpg
    28.5 KB · Views: 723
You don't need to consider cases and it doesn't matter whether the sequence converges.

Let ##u=(s+a)/2##. Can you show that for large enough ##n##, the amount ##M_n## must be less than ##u##? What does that tell us about the relationship between ##x_n## and ##a##?
 
  • Like
Likes   Reactions: NihalRi
andrewkirk said:
You don't need to consider cases and it doesn't matter whether the sequence converges.

Let ##u=(s+a)/2##. Can you show that for large enough ##n##, the amount ##M_n## must be less than ##u##? What does that tell us about the relationship between ##x_n## and ##a##?
for a large enough n, wouldn't ##M_n## = s?
a > s
a + s > s + s = 2s
(a+s)/2 > s
so s < u

I'm still trying to see the next part
 
NihalRi said:
for a large enough n, wouldn't ##M_n## = s?
Not necessarily. But it will be less than u. Why is that? (Consider the definition of lim sup)
 
  • Like
Likes   Reactions: NihalRi
It might be useful to observe that

$$\limsup a_n = \inf_{n=1}^\infty \{\sup_{k=n}^\infty a_k\}$$
 
  • Like
Likes   Reactions: NihalRi
I have more time now. Using my last formula, if ##a > \limsup a_n := L##, then there is ##n \geq 1## such that ##\sup_{k \geq n} a_k < a##. But then ##a_k \leq \sup_{k \geq n} a_k < a## for all ##k \geq n##, which was what we had to prove.

It remains to show that my last formula is correct. This is however easy, since ##\sup_{k\geq n} a_k## is a non-increasing sequence in ##n## and thus by the monotonuous convergence theorem, the sequence of suprema converges to its infinum.