The problem was:
$x_1 \gt 0$ and $x_{n+1} = \log (1 + x_n)$.
Find $\lim_{n \to \infty} nx_n$.
Note: $x_n$ is monotonically decreasing and bounded below by $0$, and $x_n \to L = 0$ (solution of $L = \log(1+L)$).
Solution
Did you know there was something like a l'Hopital rule for sequences?
It is called the Stolz–Cesàro theorem. That can be used here, and easily gives us the limit (in fact by using the fact that $x_n \to 0$ and applying the normal l'Hopital rule twice!)
We will not use that though.
Consider the sequence $$ y_n = \dfrac{1}{x_{n+1}} = \dfrac{1}{\log(1+x_n)}$$
Now $\log (1 + x_n) = x_n - x_n^2/2 + O(x_n^3)$
Thus $$y_n = \frac{1}{x_n - x_n^2/2 + O(x_n^3)} = \frac{1}{x_n(1 - x_n/2 + O(x_n^2))}$$
$$ = \frac{1}{x_n} - \frac{1}{2} + O(x_n)$$
(Using $\dfrac{1}{1-t} = 1 + t + t^2 + \dots$).
Thus $c_n = y_n - y_{n-1} \to \dfrac{1}{2}$ (we can apply Stolz Cesaro here if we want).
Now if $a_n \to L$, then $S_n = \dfrac{a_1 + a_2 + \dots + a_n}{n} \to L$.
Thus applying the above to $c_n$, we get $\dfrac{y_n}{n} \to \dfrac{1}{2}$.
Thus $$n x_n \to 2$$
$x_1 \gt 0$ and $x_{n+1} = \log (1 + x_n)$.
Find $\lim_{n \to \infty} nx_n$.
Note: $x_n$ is monotonically decreasing and bounded below by $0$, and $x_n \to L = 0$ (solution of $L = \log(1+L)$).
Solution
Did you know there was something like a l'Hopital rule for sequences?
It is called the Stolz–Cesàro theorem. That can be used here, and easily gives us the limit (in fact by using the fact that $x_n \to 0$ and applying the normal l'Hopital rule twice!)
We will not use that though.
Consider the sequence $$ y_n = \dfrac{1}{x_{n+1}} = \dfrac{1}{\log(1+x_n)}$$
Now $\log (1 + x_n) = x_n - x_n^2/2 + O(x_n^3)$
Thus $$y_n = \frac{1}{x_n - x_n^2/2 + O(x_n^3)} = \frac{1}{x_n(1 - x_n/2 + O(x_n^2))}$$
$$ = \frac{1}{x_n} - \frac{1}{2} + O(x_n)$$
(Using $\dfrac{1}{1-t} = 1 + t + t^2 + \dots$).
Thus $c_n = y_n - y_{n-1} \to \dfrac{1}{2}$ (we can apply Stolz Cesaro here if we want).
Now if $a_n \to L$, then $S_n = \dfrac{a_1 + a_2 + \dots + a_n}{n} \to L$.
Thus applying the above to $c_n$, we get $\dfrac{y_n}{n} \to \dfrac{1}{2}$.
Thus $$n x_n \to 2$$
No comments:
Post a Comment