next up previous
Next: Radius of Convergence Up: Finding when Power Series Previous: Finding when Power Series

Comparison and the Ratio Test

Our most basic tests for convergence are based on the comparison test:

Theorem 4.1 (Comparison Test)   If $0 \leq a_n \leq b_n$ for all n and $\sum b_n$ converges, then so does $\sum a_n$.

PROOF:

Since the terms an are non-negative, the sequence of partial sums

\begin{displaymath}S_k=\sum_{n=0}^k a_k\end{displaymath}

is monotone nondecreasing. To show that $\sum a_n$ converges we need only give an upper bound for all of the partial sums. Now since $a_n\leq b_n$ for all n,

\begin{displaymath}\sum_{n=0}^k a_n \leq \sum_{n=0}^k b_k \leq \sum_{n=0}^\infty b_n\end{displaymath}

Using the convergence of bounded monotone sequences, this tells us that $\sum a_n$ converges, though it does not tell us what it converges to.

Corollary 4.2   If $\sum \vert a_n\vert$ converges then so does $\sum a_n$.

PROOF:

Observe that

\begin{displaymath}0\leq a_n+\vert a_n\vert \leq 2\vert a_n\vert.\end{displaymath}

First note that if $\sum \vert a_n\vert$ converges to S then $\sum 2\vert a_n\vert$ will converge to 2S, since each partial sum is doubled. The comparison theorem then tells us that if $\sum 2\vert a_n\vert$ converges then so will $\sum (a_n + \vert a_n\vert)$. Subtracting $\sum \vert a_n\vert$ then tells us that $\sum a_n$ converges.

Definition 4.1   A series $\sum a_n$ is called absolutely convergent if $\sum \vert a_n\vert$ converges.

Since adding a finite sum to the beginning of a convergent series does not change the convergence, though it might change the first M terms, all that matters for determining convergence of a series is what happens in the tail, terms from aM on for any given M. We'll use this to provide the following comparison with geometric series:

Theorem 4.3 (Ratio Test)   Suppose that an>0 for all n and

\begin{displaymath}\lim_{n\to \infty}\frac{a_{n+1}}{a_n} = L.\end{displaymath}

Then if L<1 then $\sum a_n$ converges. If L>1 then the series diverges; if L=1 we don't know what happens.

PROOF:

We will prove the case L<1 by giving a comparison with a suitable geometric series. Since L<1 we can find an r with L<r<1 and for some M if $n\geq M$ then

\begin{displaymath}\frac{a_{n+1}}{a_n} < r.\end{displaymath}

We will show that the sequence $\displaystyle{\sum_{n=M}^\infty a_n}$ converges by comparison with the geometric series $\displaystyle{\sum_{n=0}^\infty a_M r^n}$. Certainly $a_M \leq a_M r^0$. For $n\geq M$ we have an+1< r an so using an induction hypothesis we get an+1 < rn+1-MaM. This completes the comparison, so $\sum a_n$ converges.

The proof for divergence if L>1 gives a comparison the other way with a divergent geometric series. For L=0 there are examples of both convergent and divergent series (p-series, which make their appearance later in these notes).




Example:

$\displaystyle{\sum \frac{(n!)^2}{(2n)!}}$ converges. We look at

\begin{eqnarray*}\lim_{n\to \infty} \frac{\left(\frac{((n+1)!)^2}{(2n+2)!}\right...
...&=& \lim_{n\to \infty}\frac{(n+1)^2}{(2n+2)(2n+1)} \\ &=& \frac14\end{eqnarray*}


Since $L=\frac14 < 1$ this tells us that this series converges.




Exercises: Try out the ratio test on the following series:

1.
$\displaystyle{\sum \frac{4^n}{n!}}$
2.
$\displaystyle{\sum \frac{n}{2^n}}$
3.
$\displaystyle{\sum \frac{n^{300}}{1.001^n}}$
4.
$\displaystyle{\sum \frac{n^2}{n!}}$
5.
$\displaystyle{\sum \frac{n^4}{4^n}}$
6.
$\displaystyle{\sum \frac{4^n}{n^3}}$
7.
$\displaystyle{\sum \frac{n!}{n^n}}$ Hint: $\displaystyle{\left(\frac{n+1}{n}\right)^n\to e}$


next up previous
Next: Radius of Convergence Up: Finding when Power Series Previous: Finding when Power Series
Larry Stout
1999-11-14