On this concept: Formula

Not sure which lesson to do first? Find your level and get a plan →

Math · Introduction to Calculus · Grade 9-12

Power Series

A power series is an infinite sum ∑an(x−c)n centered at c with coefficients an; it behaves like an 'infinite polynomial' and defines a function for every x where it converges, which is an interval (c−R,c+R) of radius R.

∑n=0∞an(x−c)n
Radius of convergence: R=1lim sup⁡n→∞∣an∣1/n or use the ratio test: R=lim⁡n→∞∣anan+1∣.

Quick Answer

A power series is an infinite sum ∑an(x−c)n centered at c with coefficients an; it behaves like an 'infinite polynomial' and defines a function for every x where it converges, which is an interval (c−R,c+R) of radius R. Use one when a function is given by its coefficients rather than a formula, when you want to differentiate or integrate a function term by term, or when you need to know which inputs give a number at all. Inside the radius the sum behaves like an ordinary function; the two endpoints must be tested separately.

Why This Matters

Once you know where a power series converges, you can treat it as a function: plug in values, differentiate, integrate, or recognize it as 11−x or ln⁡(1+x). That is how the arctan series produces digits of π, and how many differential equations are solved when no formula exists. Ignore the interval and you get nonsense, such as 1+2+4+8+⋯=−1 from putting x=2 into 11−x.

Intuitive Explanation

Try values in 1+x+x2+x3+⋯. At x=0.5 the partial sums are 1, 1.5, 1.75, 1.875, closing in on 2. At x=−0.5 they are 1, 0.5, 0.75, 0.625, 0.6875, closing in on 23. At x=2 they are 1, 3, 7, 15, growing without limit. The same expression gives a number for some inputs and nothing for others, and the inputs that work form the interval of convergence. Where it works, the sum agrees with 11−x.

That set is always a single stretch centered at c, never scattered pieces. The size of each term is controlled by ∣x−c∣: if ∣an+1an∣→L, the terms behave roughly like (L∣x−c∣)n, which shrink when ∣x−c∣<1L and grow when it is larger. That is why the ratio test finds R. At the two endpoints this comparison gives exactly 1 and tells you nothing, so each endpoint needs its own test, and the two can disagree: ∑xnn converges at x=−1 but not at x=1.

Compare three look-alikes. ∑1n2 has no x, so it converges or it doesn't, and that is one answer, not an interval. 3x2−x+1 has finitely many terms, so it never fails to give a number. A Taylor series is a power series whose coefficients came from a known function's derivatives, but a power series can be given by its coefficients alone, as ∑xnn2 is.

Think of it like this. A power series behaves like a radio station broadcasting from the center c: within range R you get a clear signal, a number; beyond it, only noise. Close to the edge the signal fades, so the sum converges more slowly, and right at the edge reception can go either way. On the number line the range is just the stretch from c−R to c+R.

Core idea. A power series ∑an(x−c)n defines a function on the interval of x where it converges.


When to Use

Treat a problem as a power series problem when you see infinitely many terms of the form an(x−c)n and a question about the radius or interval of convergence, or a request to differentiate, integrate or identify the function a series represents. Find the radius with the ratio test, then test each endpoint. If there is no x, it is a numeric series with a single yes-or-no answer. If the terms stop, it is a polynomial. If you must build the coefficients from a function's derivatives, start with Taylor series.

Tip. Ask: Is this a series whose terms are coefficients times powers of (x−c), with convergence depending on the value of x?

How to Recognize It

Find the center c, see what the coefficients do as n grows, and note whether the question asks for an interval, a function or a value.

  1. Do the coefficients contain a factorial, like xnn!?

    A factorial in the denominator usually makes the radius infinite, so the series converges for every x, as for ex. A factorial on top, as in ∑n! xn, gives radius 0: the series converges only at its center.

  2. Does the series have the form ∑(something)n, a geometric series in disguise?

    Then the radius comes straight from ∣something∣<1: ∑(x−14)n converges for ∣x−1∣<4, to 45−x. Otherwise use the ratio test on the whole term, x included.

  3. Are you asked to build the series from a given function's derivatives?

    Yes is Taylor series work, with coefficients f(n)(c)n!. Power series questions usually go the other way, handing you the coefficients and asking where the sum makes sense or which function it equals.

  4. Does the series stop after finitely many terms?

    Then it is a polynomial, defined for every x, with no convergence question at all. Only infinitely many terms make a radius of convergence necessary.

Power Series compared with Taylor series, Polynomial, and Numeric (constant) series

All four rows are sums of terms, and three use powers of x, so they are easy to mix up. Ask whether there is a variable x, whether the terms stop, and whether the coefficients come from a function's derivatives.

Power Series

Meaning
Use this when you have a series in powers of (x−c) and need to find where it converges or treat it as a function of x.
Key test
Is this a series whose terms are coefficients times powers of (x−c), with convergence depending on the value of x?
Formula
∑n=0∞an(x−c)n
Radius of convergence: R=1lim sup⁡n→∞∣an∣1/n or use the ratio test: R=lim⁡n→∞∣anan+1∣.
Example
Find the radius and interval of convergence of ∑n=1∞xnn.

Taylor series

Meaning
A SPECIAL power series whose coefficients are f(n)(c)n! from a known function.
Key test
Use 'Taylor' when the coefficients come from a function's derivatives.
Formula
an=f(n)(c)n!
Example
ex=∑xn/n!

Polynomial

Meaning
A FINITE sum of powers; converges (is defined) for all x trivially.
Key test
Use when there are finitely many terms and no convergence question.
Formula
a0+⋯+akxk
Example
3x2−x+1

Numeric (constant) series

Meaning
Has no variable x; convergence is a single yes/no, not an interval.
Key test
Use when the terms are numbers, not functions of x.
Formula
∑an
Example
∑1/n2 converges to π2/6

Formula & Notation

∑n=0∞an(x−c)n
Radius of convergence: R=1lim sup⁡n→∞∣an∣1/n or use the ratio test: R=lim⁡n→∞∣anan+1∣.
∑n=0∞an(x−c)n has radius of convergence R=1/lim sup⁡n→∞∣an∣1/n. The series converges absolutely for ∣x−c∣<R and diverges for ∣x−c∣>R. Within (c−R,c+R): ddx∑an(x−c)n=∑nan(x−c)n−1 and ∫∑an(x−c)n dx=∑ann+1(x−c)n+1+C.

How to read it: R = radius of convergence. Interval of convergence = (c−R,c+R), with endpoints checked separately.

Worked Examples

Example 1 — Find the radius of convergence

Easy

Find the radius and interval of convergence of ∑n=1∞xnn.
Show the steps and the answer
  1. Read the parts: center c=0, coefficients an=1n, terms xnn.

    The center is where the interval of convergence will be centered, here at 0.

  2. Apply the ratio test: ∣xn+1/(n+1)xn/n∣=∣x∣⋅nn+1→∣x∣.

    The series converges when this limit is below 1 and diverges when it is above 1, so it converges for ∣x∣<1 and R=1.

  3. Test the endpoint x=1: the series becomes 1+12+13+⋯, the harmonic series, which diverges.

    At an endpoint the ratio limit is exactly 1, which settles nothing, so a separate test is needed.

  4. Test the endpoint x=−1: the series becomes −1+12−13+14−⋯, which converges by the alternating series test (its sum is −ln⁡2).

    The signs alternate and the sizes 1n shrink to 0, which is enough for an alternating series to converge.

  5. Assemble the interval: include −1, exclude 1.

    A square bracket marks an endpoint where the series converges and a round one an endpoint where it diverges, giving [−1,1).

Answer: R=1, interval [−1,1)

Takeaway: The ratio test gives the radius; each endpoint needs its own test, and the two endpoints can behave differently.

Example 2 — A finite polynomial

Standard

Where does 3x2−x+1 converge?
Show the steps and the answer
  1. Count the terms of 3x2−x+1: there are three, so every coefficient an with n≥3 is 0.

    Convergence is a question about infinite sums, and this sum stops after three terms.

  2. Try a large input, x=100: 3(10000)−100+1=29901.

    A finite sum of numbers is always a number, however large x is, so no input makes it fail.

  3. Viewed as a power series with a0=1, a1=−1, a2=3 and all other an=0, its radius is R=∞.

    The ratio test would divide by zero here; the accurate statement is that the interval is the whole real line.

Answer: All real x (no radius needed)

Takeaway: If the terms stop, there is nothing to converge: a polynomial is defined for every real x. Radius and endpoints matter only when infinitely many terms are nonzero.

Example 3 — Find the mistake: the wrong center

Application

Find the interval of convergence of ∑n=0∞(x−3)n2n. Lee writes: "The ratio test gives ∣x−3∣2<1, so R=2 and the interval is (−2,2)." Find the mistake and correct it.
Show the steps and the answer
  1. Keep Lee's correct work: the ratio of neighbouring terms is ∣x−3∣2, so the series converges when ∣x−3∣<2, and R=2.

    The radius is right; the mistake is where Lee placed the interval.

  2. Solve ∣x−3∣<2: −2<x−3<2, so 1<x<5.

    Powers of (x−3) put the center at c=3, so the interval runs from 3−2 to 3+2.

  3. Test a point in Lee's interval, x=−1: the terms are (−42)n=(−2)n, which grow without bound.

    One input inside Lee's interval where the series diverges is enough to show that interval is wrong.

  4. Test the endpoints: at x=5 every term is 1n=1, and at x=1 the terms are (−1)n.

    In neither case do the terms shrink to 0, so both endpoints diverge and are excluded.

  5. Confirm a point inside, x=4: the series is 1+12+14+⋯=2.

    It is geometric with ratio x−32=12, and the sum matches the closed form 25−x=2.

Answer: The interval is (1,5): centered at 3 with R=2, and both endpoints diverge.

Takeaway: Solve ∣x−c∣<R for x, so the interval is centered at c, not at 0. Then test both endpoints.

Where You'll See It

  • Computing digits of pi

    The power series arctan⁡x=x−x33+x55−⋯ converges for −1≤x≤1. At the endpoint x=1 it gives π4, but so slowly that 1000 terms still leave an error of about 0.001 in π. In 1706 John Machin used π4=4arctan⁡15−arctan⁡1239, where each input sits well inside the interval and terms shrink about 25 times per step, to compute 100 digits.

  • Pricing payments that never end

    A perpetuity pays $100 a year forever when interest is 5%. A payment n years away is worth 100xn today, with x=11.05, so the value is ∑n=1∞100xn. Because ∣x∣<1 this converges to 100x1−x=1000.05=2000, or $2000. At 0% interest x=1 and the series diverges: no finite price is fair.

  • Counting ways to make change

    The number of ways to make n cents from pennies and nickels is the coefficient of xn in 1(1−x)(1−x5)=(1+x+x2+⋯ )(1+x5+x10+⋯ ). For 12 cents, x12 arises as x12⋅1, x7⋅x5 and x2⋅x10: three ways, using 0, 1 or 2 nickels. Multiplying in series for dimes and quarters too, the coefficient of x100 shows there are 242 ways to make a dollar.

Common Mistakes

  1. The slip: Reporting only the radius and skipping endpoints

    Instead: test x=c−R and x=c+R separately to get the full interval.

  2. The slip: Confusing the center

    Instead: powers are of (x−c), so the interval is centered at c, not at 0 unless c=0.

  3. The slip: Assuming it equals a function everywhere

    Instead: a power series only defines a function inside its interval of convergence.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. Find the radius and interval of convergence of ∑n=0∞xn3n.

    Hint: This is a geometric series with ratio x3.

  2. Find the interval of convergence of ∑n=1∞(x−2)nn2.

    Hint: The ratio test gives the radius. Then test x=1 and x=3 separately.

  3. Find the interval of convergence of ∑n=1∞(x+1)n2n n.

    Hint: The centre is −1, not 0. Find the radius, then test both endpoints.

  4. Write 11+2x as a power series centred at 0, and say where it converges.

    Hint: Compare with 11−u=1+u+u2+⋯ and choose u.

  5. Is ∑n=1∞1n2 3n a power series? Is ∑n=1∞xnn2 3n?

    Hint: Look for a variable raised to the power n.

  6. A student finds R=1 for ∑n=1∞(x−5)nn and writes the interval as (−1,1). Correct the work.

    Hint: Where is the series centred? Then check both endpoints.

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Why is the set where a power series converges always an interval centered at c?

If the series converges at some x0x_0, its terms are bounded there. At any xx closer to cc, each term is smaller by the factor ∣x−cx0−c∣n\left|\frac{x-c}{x_0-c}\right|^n, which shrinks like a geometric series. So convergence at one distance forces convergence at every smaller distance, on both sides. That is why the answer is always a radius RR around cc, possibly 0 or infinite, plus a check of the two edge points.

If I differentiate a power series term by term, does the radius change?

No, the radius stays the same, though the endpoints can change. ∑xnn2\sum\frac{x^n}{n^2} converges at both x=1x=1 and x=−1x=-1, but its derivative ∑xn−1n\sum\frac{x^{n-1}}{n} diverges at x=1x=1. Term-by-term differentiation and integration are what turn 11−x=∑xn\frac{1}{1-x}=\sum x^n into series for 1(1−x)2\frac{1}{(1-x)^2} and for −ln⁡(1−x)-\ln(1-x).

Can a power series define a function that has no ordinary formula?

Yes, and several important functions are defined exactly this way. Bessel functions, which describe the vibration of a drumhead, are written as power series such as J0(x)=∑n=0∞(−1)n(n!)2(x2)2nJ_0(x)=\sum_{n=0}^{\infty}\frac{(-1)^n}{(n!)^2}\left(\frac{x}{2}\right)^{2n}, which converges for every xx. Engineers compute their values straight from the series.

Learning Path

Power Series

You are here

Next →

Nothing after this yet
You need Sigma Notation to read ∑an(x−c)n, Convergence and Divergence (the ratio test, the harmonic and alternating series) to judge each input, and Taylor Series as the main source of examples. Power series add the idea of a function defined by its coefficients on an interval. Next you can differentiate and integrate series term by term, for example turning 11+x2 into the series for arctan⁡x, and solve differential equations with series.

See Also