Convergence of a Recursive Sequence: Proving a_n*c*n\rightarrow 1 for Positive c

  • Level: Graduate 
  • Thread starter Thread starter happyxiong531
  • Start date Start date
  • Tags Tags
    Sequence
Join the discussion
Registration is free. Start your own thread to ask a follow-up.
4 replies · 2K views
happyxiong531
Messages
3
Reaction score
0
I want to prove that
if the sequence [itex]a_n[/itex] satisfy that
[itex]a_{n+1}=a_n\left(1-c\frac{a_n}{1+a_n}\right)[/itex]
then [itex]a_n*c*n\rightarrow 1[/itex] for all positive [itex]c[/itex].

Like when [itex]c=1[/itex], then [itex]a_n*n\rightarrow 1[/itex],
but if [itex]c\neq 1[/itex], it's difficult to prove.
 
Physics news on Phys.org
What makes you believe it is true? Your question implies [itex]a_n*n ->\frac{1}{c}[/itex]. Doesn't look right, especially for large c.
 
Last edited:
mathman said:
What makes you believe it is true? Your question implies a_n*n -> 1/c. Doesn't look right, especially for large c.

Thank you for you reply.
I think it's correct.
First, I can have [itex]a_n*n\rightarrow 1[/itex] when [itex]c=1[/itex], from
[itex]a_{n+1}=\frac{a_n}{1+a_n}=\frac{a_{n-1}}{1+2a_{n-1}}=\cdots=\frac{a_1}{1+(n+1)a_1}[/itex]

Then, let [itex]ca_n=b_n[/itex] if [itex]c\neq 1[/itex], [itex]c[/itex] is some constant. we can have [itex]b_{n+1}=b_n\left(1-\frac{b_n}{1+b_n/c}\right)[/itex].

Actually, it's easy to prove [itex]a_n[/itex] and [itex]b_n[/itex] will go to zero,
so, [itex]\frac{b_n}{1+b_n/c}\sim\frac{b_n}{1+b_n}[/itex], then[itex]b_n*n\rightarrow 1[/itex].

I have made a plot, it's correct no matter [itex]c[/itex] is larger or less than 1.
But I think my proof is not strict.
Thanks for your concern.
 
mathman said:
What makes you believe it is true? Your question implies [itex]a_n*n ->\frac{1}{c}[/itex]. Doesn't look right, especially for large c.

Oh, I forget there is a condition that
the sequence should satisfy that[itex]1-c\frac{a_1}{1+a_1}[/itex]>0,
so that all the elements in this sequence should be positive, and c cannot be too large.
I have made some plots like [itex]c=0.5, c=2[/itex], the conclusion is correct.
Thanks