3
$\begingroup$

Let sequence $\{a_{n}\}$ such $$\lim_{n\to\infty}(4a_{n+2}-4a_{n+1}+a_{n})=2014$$

show that $$\lim_{n\to \infty}a_{n}$$ exist and find the limit value.

Now I use an ugly method to solve this. I use this follow lemma: if $$\lim_{n\to\infty}(a_{n+1}-\lambda a_{n})=a \Longleftrightarrow \lim_{n\to\infty}a_{n}=a,|\lambda|<1$$

I know this lemma proof is ugly, maybe anyone here has a simple method. Thank you.

$\endgroup$
3
$\begingroup$

Consider $b_0=a_0$, $b_1=a_1-a_0$, and $b_n=a_n-a_{n-1}+\frac14a_{n-2}$ for every $n\geqslant2$, and define the shift operator $\vartheta$ on sequences $c=(c_n)_{n\geqslant0}$ by $(\vartheta c)_n=c_{n-1}$ for every $n\geqslant1$ and $(\vartheta c)_0=0$.

Then the sequences $a=(a_n)_{n\geqslant0}$ and $(b_n)_{n\geqslant0}$ solve the identity $$ b=Q(\vartheta)(a),\qquad Q(x)=1-x+\tfrac14x^2, $$ and the inverse of the polynomial $Q(x)$ in the field of formal series is $$ Q(x)^{-1}=(1-\tfrac12x)^{-2}=\sum_{k\geqslant0}\frac{k+1}{2^k}x^k, $$ hence, $$ a=Q(\vartheta)^{-1}(b)=\sum_{k\geqslant0}\frac{k+1}{2^k}\vartheta^k(b), $$ that is, for every $n\geqslant0$, $$ a_n=\sum_{k=0}^n\frac{k+1}{2^k}b_{n-k}. $$ All this is algebra. Now, some analysis: the usual $N$-$\delta$ approach and the absolute convergence of the series $Q(x)^{-1}$ at $x=1$ show that, if $b_n\to\ell$ then $$a_n\to Q(1)^{-1}\ell=4\ell.$$ In the present case, one assumes that $b_n\to\frac14\cdot2014$ hence $a_n\to2014$.

Edit: The above is a bare-hands approach. If one can use the "ugly" lemma mentioned in the question, things are quicker. First, the lemma:

Assume that $a_{n+1}-\lambda a_{n}\to\ell$ for some $|\lambda|\lt1$, then $a_n\to\ell/(1-\lambda)$.

Now, its application in the case at hand: let $c_n=a_n-\frac12a_{n-1}$ and $b_n=c_n-\frac12c_{n-1}$, then $(b_n)$ is the sequence introduced in our answer above and $b_n\to\ell$ with $\ell=\frac14\cdot2014$ hence, using the lemma once with $\lambda=\frac12$, one gets $c_n\to\ell/(1-\frac12)=2\ell$, and using the lemma a second time, one gets $a_n\to(2\ell)/(1-\frac12)=4\ell$, QED.

Note finally that the proof of the "ugly" lemma is similar to the first approach in this post.

$\endgroup$
  • $\begingroup$ Nice proof! Thank you+1 $\endgroup$ – china math Jun 22 '14 at 10:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.