Calculus › Section J · Infinite series › Topic J.8

Power series

A power series is an infinite polynomial: fix a centre, stack up powers of (x - a), and let the series define a brand-new function. Where it converges is governed by one number, the radius, plus two delicate endpoint checks.

A polynomial that never ends

So far, a series had a fixed list of numbers as its terms, and its verdict was a single yes or no. Now let the terms depend on \(x\):

\sum_{n=0}^{\infty} c_n (x - a)^n = c_0 + c_1(x - a) + c_2(x - a)^2 + c_3(x - a)^3 + \cdots

This is a power series with centre \(a\): an endless polynomial. Substitute any particular \(x\) and you get an ordinary series to judge with the ordinary tests, and different \(x\) values can give different verdicts. So a power series carves the number line into a region where it converges, defining a function, and a region where it does not. The whole topic is about mapping that region.

The shape of the convergence region

Plug in \(x = a\) and every term with a power of \(x - a\) vanishes: the series always converges there, equal to \(c_0\). The structure theorem says the rest of the story is remarkably rigid. Exactly one of three things happens for a power series centred at \(a\): it converges only at \(x = a\); or there is a positive number \(R\) such that it converges absolutely for every \(x\) with \(|x - a| < R\) and diverges for every \(x\) with \(|x - a| > R\); or it converges absolutely for every \(x\). The number \(R\) is the radius of convergence, and \(R = 0\) and \(R = \infty\) cover the first and last cases. The convergence set is a symmetric interval around \(a\), with one twist: at the two boundary points \(x = a \pm R\) the theorem says nothing, and each must be tested by hand.

Why the tests see only distance

The radius-of-convergence behaviour is not a coincidence; it is baked into how the tests work. Apply the ratio test to a power series:

\lim_{n \to \infty} \left|\frac{c_{n+1}(x-a)^{n+1}}{c_n(x-a)^n}\right| = |x - a| \cdot \lim_{n \to \infty} \left|\frac{c_{n+1}}{c_n}\right|

The \(x\)-dependence factors out completely as the single quantity \(|x - a|\), the distance from the centre. So the limit comes out as distance times a constant, and the threshold \(L < 1\) becomes a statement about \(|x - a|\) alone: inside a cut-off distance the ratio is under 1, outside it is over 1, and at distance exactly on the cut-off the test is silent. That silence is exactly why the endpoints need separate hearings.

Worked example: radius and both endpoints

Find the convergence interval of \(\sum_{n=1}^{\infty} \frac{x^n}{n \cdot 3^n}\). The ratio is

\frac{|x|^{n+1}}{(n+1)3^{n+1}} \cdot \frac{n \cdot 3^n}{|x|^n} = \frac{|x|}{3} \cdot \frac{n}{n+1} \to \frac{|x|}{3}

which is below 1 exactly when \(|x| < 3\): the radius is \(R = 3\). Now the endpoints. At \(x = 3\): the series becomes \(\sum \frac{1}{n}\), the harmonic series, which diverges. At \(x = -3\): it becomes \(\sum \frac{(-1)^n}{n}\), the alternating harmonic series, which converges (conditionally, since the sign-stripped series diverges). So the interval of convergence is \([-3, 3)\): closed on the left, open on the right, one endpoint of each type, decided one at a time.

Worked example: both extremes

Two quick series show the whole range of radii. First, \(\sum_{n=0}^{\infty} \frac{x^n}{n!}\). The ratio is \(\frac{|x|^{n+1}}{(n+1)!} \cdot \frac{n!}{|x|^n} = \frac{|x|}{n+1}\), which approaches 0 for every \(x\). So \(R = \infty\): this series converges absolutely everywhere, which is the dream case. Second, \(\sum_{n=0}^{\infty} n!\, x^n\). The ratio is \((n+1)|x|\), which grows without bound for any \(x \ne 0\), so the ratio test condemns every point except the centre: \(R = 0\), and the series converges only at \(x = 0\), where every powered term vanishes and only the constant term, 1, survives. Factorials in the numerator push the radius down to nothing; factorials in the denominator push it to infinity.

Absolute inside, wild at the edges

Notice what the structure theorem promises inside the radius: absolute convergence, at every interior point. Conditional convergence, the delicate kind rescued by alternating signs, can occur only at the two endpoints, never inside. So a power series is stable in the interior: you can shuffle and combine terms there, because absolute convergence earns that freedom. The interesting drama, divergent endpoint, convergent endpoint, absolute or conditional, all plays out at the boundary. When a problem asks for the interval of convergence, it wants the full verdict: radius from the ratio test, plus both endpoint decisions.

Convergence region versus the function

The geometric series \(\sum x^n\) converges exactly for \(|x| < 1\), where it equals \(\frac{1}{1-x}\) (the formula from the series topic). But the expression \(\frac{1}{1-x}\) is a perfectly well-behaved number at \(x = 2\): it equals \(-1\). The series, meanwhile, is \(1 + 2 + 4 + \cdots\) there, which diverges spectacularly. The lesson: a series may represent a function only inside its radius, and the function can continue on past the boundary where the series dies. Convergence of a power series is about the series, not about any formula it happens to match. Where the representation is valid is part of the answer, and claiming a series equals a function without its interval is claiming too much.

Common mistakes to avoid

The most common lost points come from stopping at the radius: \(R = 3\) is not the answer to an interval-of-convergence question, because the endpoints are undecidable by the ratio test and must be substituted and judged individually, with the harmonic and alternating harmonic patterns being the usual outcomes. Second, do not assume symmetry means the endpoints behave the same: \([-3, 3)\) has one of each kind. Third, when the centre is not 0, keep the shift: \(\sum \frac{(x - 2)^n}{n \cdot 3^n}\) has radius 3 but its interval is centred at 2, running from -1 to 5, endpoints to be checked at those values. Fourth, remember that convergence inside the radius is always absolute: writing "conditionally convergent" about an interior point is never right.

Key takeaways

A power series \(\sum c_n(x-a)^n\) converges absolutely for \(|x - a| < R\), diverges for \(|x - a| > R\), converges at its centre always, and at \(x = a \pm R\) anything can happen: each endpoint is judged separately by substituting it in. The ratio test finds \(R\), because the \(x\)-dependence factors out as the distance \(|x - a|\). Radii run from 0 (converges only at the centre, like \(\sum n!x^n\)) to \(\infty\) (converges everywhere, like \(\sum \frac{x^n}{n!}\)). Conditional convergence can appear only at endpoints, never inside. The interval of convergence is the complete answer: radius plus both endpoint verdicts.

Flashcards

Flashcard

Power series

Σ cₙ(x - a)ⁿ, an endless polynomial

Centre a; each x gives a series to judge.

Test your understanding

Question 1 of 5

What is the radius of convergence of Σ xⁿ/(n·3ⁿ)?

Question 2 of 5

What is the interval of convergence of Σ xⁿ/(n·3ⁿ)?

Question 3 of 5

What is the radius of convergence of Σ xⁿ/n!?

Question 4 of 5

At which points can a power series converge conditionally?

Question 5 of 5

The geometric series Σ xⁿ equals 1/(1 - x) for which values of x?

Standards: 10 (BC)

← Previous topicNext topic →