Royden Review:LimSup, LimInf are the Largest/Smallest Limit Points of {a_n}

  • Level: Graduate 
  • Thread starter Thread starter Bacle
  • Start date Start date
  • Tags Tags
    Limit Points
Join the discussion
Registration is free. Start your own thread to ask a follow-up.
2 replies · 6K views
Bacle
Messages
656
Reaction score
1
Hi, All:

Let {a_n}; n=1,2,... be sequence. I am trying to show LimSup and LimInf are the largest and smallest limit points of {a_n}. This is what I got so far:

i) If {a_n} converges, to, say, a<oo, then LimSup=LimInf, and we're done, since we have a unique limit point L. If a=oo, then oo is the limit point.

I think I can show ( here in ii) below ) that Lim Sup, Lim Inf are both limit points of {a_n}, but I cannot show they are the largest, smallest respectively.

LimSup is a limit point of {a_n} if {a_n} does not converge:
Proof:
ii)If {a_n} does not converge, then it is not strictly monotone, so we can extract monotone

non-increasing and monotone non-decreasing subsequences ( by using, e.g., the lim sup

and lim ii) lim sup, lim inf are both limit points; in the case of lim sup, the sequence: a_n' :

{ sup_k>n(a_n)} is monotone non-increasing; by the LUB property,L= LimSup{a_n} :=inf_n

(a_n') exists, and it is a limit point of {a_n}; by contradiction, if L were not a limit point of

{a_n}, there would be e>0 with L> a_k-e for all a_k. But then L is not the lub of the

monotone-decreasing subsequence; a_k-e is the LUB.

I am stuck trying to show that Lim Sup is the largest limit point; I am sure the proof that

Lim Inf is the smallest limit point is automatic after knowing this one.

Thanks.
 
Physics news on Phys.org
What is your definition of "limsup" and "liminf". The definition I learned was "[/itex]a= limsup a_n[/itex] if and only if a is the supremum of the set of all subsequential limits of [itex]a_n[/itex]" and "[itex]b= liminf a_n[/itex] if and only if b is the infimum of the set of all subsequential limits of [itex]a_n[/itex]".

But notice that I used "sup" and "inf", NOT "largest" and "smallest". A sequence may not have "largest and smallest limit points".
 
Let's say that L is a limit point of the sequence. Can you prove that

[tex]\sup_{n\geq k}~{a_n}\geq L[/tex]

for every k?