Alternating series test

The alternating series test is a method used to prove that infinite series of terms converge. It was discovered by Gottfried Leibniz and is sometimes known as Leibniz's test or the Leibniz criterion.

A series of the form

 \sum_{n=0}^\infty (-1)^n a_n\!

where all the an are non-negative, is called an alternating series. If the sequence an approaches 0 as n approaches infinity, and the sequence an is monotone decreasing (i.e. each an is smaller than an−1), then the series converges. If L is the sum of the series,

\sum_{n=0}^\infty (-1)^n a_n = L\!

then the partial sum

S_k = \sum_{n=0}^k (-1)^n a_n\!

approximates L with error

\left | S_k - L \right \vert \le \left | S_k - S_{k-1} \right \vert = a_k\!

It is perfectly possible for a series to have its partial sums Sk fulfill this last condition without the series being alternating. For a straightforward example, consider:

\sum_{n=0}^\infty \left(\frac{1}{3}\right)^n  = \frac{3}{2}\!

Contents

Proof

[1]

We are given a series of the form \sum_{n=0}^\infty (-1)^n a_n\!. The limit of the sequence a_n equals 0 as n approaches infinity, and each a_n is smaller than a_{n-1} (i.e. the sequence a_n is monotone decreasing).

Proof of convergence

The (2n+1)-th partial sum of the given series is W_{2n%2B1}  = a_0  %2B \left( { - a_1  %2B a_2 } \right) %2B \left( { - a_3  %2B a_4 } \right) %2B \ldots %2B \left( { - a_{2n - 1}  %2B a_{2n} } \right) - a_{2n%2B1} . As every sum in brackets is non-positive, and as a_{2n%2B1}  \geq 0, then the (2n+1)-th partial sum is not greater than a_0.

That very (2n+1)-th partial sum can be written as W_{2n%2B1}  = \left( {a_0  - a_1 } \right) %2B \left( {a_2  - a_3 } \right) %2B \ldots %2B \left( {a_{2n}  - a_{2n%2B1} } \right). Every sum in brackets is non-negative. Therefore, the series S_{2n%2B1} is monotonically increasing: for any n \in N the following holds: W_{2n%2B1}  \le S_{2n %2B 3} .

From the two paragraphs it follows by the monotone convergence theorem that there exists such a number s that \lim_{n \to \infty } W_{2n%2B1}  = s.

As W_{2n}  = W_{2n%2B1}  - a_{2n %2B 1} and as \lim_{n \to  %2B \infty } a_n  = 0, then \lim_{n \to \infty } W_{2n}  = s. The sum of the given series is \lim_{n \to \infty}W_{2n}  = \lim_{n \to \infty}W_{2n %2B 1}  = s, where s is a finite number. Thus, convergence is proved.

Another way to prove this is showing that the sequence of partial sums are a cauchy sequence.

Proof of partial sum error

In the proof of convergence we saw that S_{2n%2B1} is monotonically increasing. Since S_{2n} = a_0 %2B\left(-a_1 %2B a_2\right) %2B \ldots %2B \left(-a_{2n-1} %2B a_{2n}\right), and every term in brackets is non-positive, we see that S_{2n} is monotonically decreasing. By the previous paragraph, \lim_{n \to \infty}S_{2n} = L, hence S_{2n} \geq L. Similarly, since S_{2n%2B1} is monotonically increasing and converging to L, we have S_{2n%2B1} \leq L. Hence we have S_{2n%2B1} \leq L \leq S_{2n} for all n.

Therefore if k is odd we have |L - S_k| = L - S_k \leq S_{k%2B1} - S_k = a_{k%2B1} \leq a_k, and if k is even we have |L-S_k| = S_k - L \leq S_k - S_{k-1} = a_k.

See also

Literature

References

  1. ^ Beklemishev, Dmitry V. (2005). Analytic geometry and linear algebra course (10 ed.). FIZMATLIT. 

External links