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

Taylor Series

A Taylor series represents a smooth function as an infinite sum $\sum\frac{f^{(n)}(a)}{n!

f(x)=∑n=0∞f(n)(a)n!(x−a)n=f(a)+f′(a)(x−a)+f′′(a)2!(x−a)2+⋯

Quick Answer

A Taylor series represents a smooth function as an infinite sum ∑f(n)(a)n!(x−a)n, built so that every derivative of the polynomial matches the function's at the center a (when a=0 it's a Maclaurin series). Use it when a function is hard to evaluate, integrate or take a limit of directly, but its derivatives at one point are easy. Near the center, a few terms give a polynomial that is very close to the function, and more terms widen the range where it stays close. Far from the center, or outside the interval of convergence, the series can fail.

Why This Matters

Polynomials need only adding and multiplying, so a Taylor series turns e0.1, sin⁡0.2 or ∫01e−x2 dx into arithmetic. It is also the source of the standard shortcuts in science, such as sin⁡θ≈θ for small angles. Drop the n! or ignore where the series converges, though, and you get a polynomial that looks reasonable and is badly wrong.

Intuitive Explanation

Build a polynomial that copies ex near x=0, one feature at a time. The constant 1 matches the height. 1+x also matches the slope: it is the tangent line. 1+x+x22 also matches how fast the slope is changing. Each new term fixes one more derivative without spoiling the earlier ones, because xn and its first n−1 derivatives are all zero at x=0. At x=1 the partial sums run 1, 2, 2.5, 2.667, 2.708, 2.717, closing in on e≈2.71828.

The n! comes from differentiating. Taking three derivatives of x3 gives 3x2, then 6x, then 6=3!. So for the nth derivative of the term cn(x−a)n to equal f(n)(a), the coefficient has to be cn=f(n)(a)n!. Forget the factorial and every term from x2 on is too big.

Matching derivatives at one point does not guarantee agreement far away. The series ln⁡(1+x)=x−x22+x33−⋯ fits beautifully near 0, but at x=2 its terms 2nn grow and the sum never settles, although ln⁡3 is a perfectly good number. The series equals ln⁡(1+x) only for −1<x≤1. A tangent line, by contrast, is the same idea stopped after two terms, and it is trustworthy only very close to a.

Think of it like this. A Taylor series is like giving directions from one landmark: first where you are (the value), then which way you are heading (the slope), then how the road bends (the second derivative), and so on. Near the landmark a few instructions get you almost exactly there. Far away the errors add up, and for some functions no amount of detail reaches past a certain distance.

Core idea. A Taylor series rebuilds a function as an infinite polynomial whose derivatives at a match the function's.


When to Use

Use a Taylor series when a problem asks you to approximate a function near a point, asks for a Maclaurin or Taylor polynomial of a given degree, or needs a limit or integral with no easy antiderivative, such as ∫00.5e−x2 dx. If only the tangent line is wanted, stop after two terms. If the series is already given and the question is where it converges, treat it as a power series problem. If neighbouring terms have a constant ratio, sum it as a geometric series instead.

Tip. Ask: Am I building an infinite polynomial whose successive derivatives at one center match the function's?

How to Recognize It

Look at whether you start from a function or from a given series, where the approximation is needed, and how many terms the question asks for.

  1. Are you given a function and asked to produce its series or polynomial?

    Yes is the Taylor direction: compute f(n)(a) and divide by n!. If a series ∑an(x−c)n is handed to you and the question is where it converges, treat it as a power series problem.

  2. Is the point where you need a value close to a point where the function is easy?

    Yes means center there: ln⁡1.1 is best done about a=1, where ln⁡1=0. A center far from the point needs many more terms, and outside the interval of convergence no number of terms helps.

  3. Does the question want only the tangent line, the first two terms?

    Then stop at f(a)+f′(a)(x−a), the linear approximation. Taylor polynomials of degree 2 and higher add the bending, and a question naming a degree tells you exactly where to stop.

  4. Can the function be built from a series you already know, like ex or 11−x?

    Then substitute instead of differentiating: replacing x by −x2 in the series for ex gives e−x2=1−x2+x42−⋯ in one line, far faster than many derivatives by hand.

Taylor Series compared with Power series (general), Linear approximation / tangent line, and Infinite geometric series

All four rows write a function or a number as a sum of terms, so they share notation. Ask where the coefficients come from: the function's derivatives, any given rule, only the first two terms, or a constant ratio between terms.

Taylor Series

Meaning
Use this when you need to approximate a smooth function near a point, or convert it to an infinite polynomial built from its derivatives there.
Key test
Am I building an infinite polynomial whose successive derivatives at one center match the function's?
Formula
f(x)=∑n=0∞f(n)(a)n!(x−a)n=f(a)+f′(a)(x−a)+f′′(a)2!(x−a)2+⋯
Example
Find the Maclaurin series (Taylor at a=0) of f(x)=ex.

Power series (general)

Meaning
Any series ∑an(x−c)n; a Taylor series is the one whose coefficients are f(n)(a)n!.
Key test
Use 'power series' when coefficients aren't tied to a function's derivatives.
Formula
∑an(x−c)n
Example
an arbitrary coefficient series

Linear approximation / tangent line

Meaning
Just the FIRST two Taylor terms: f(a)+f′(a)(x−a).
Key test
Use when one term of accuracy near a suffices.
Formula
L(x)=f(a)+f′(a)(x−a)
Example
4.1≈2.025

Infinite geometric series

Meaning
A specific closed-form sum a1−r; 11−x's Taylor series IS one, but most aren't.
Key test
Use when each term is a constant ratio times the one before; no derivatives needed.
Formula
a1−r
Example
1+x+x2+⋯=11−x

Formula & Notation

f(x)=∑n=0∞f(n)(a)n!(x−a)n=f(a)+f′(a)(x−a)+f′′(a)2!(x−a)2+⋯
f(x)=∑n=0∞f(n)(a)n!(x−a)n for ∣x−a∣<R (radius of convergence). Taylor's theorem with remainder: f(x)=Tn(x)+Rn(x) where Rn(x)=f(n+1)(c)(n+1)!(x−a)n+1 for some c between a and x (Lagrange form).

How to read it: Tn(x) = Taylor polynomial of degree n. Rn(x) = remainder (error) term.

Worked Examples

Example 1 — Maclaurin series of e^x

Easy

Find the Maclaurin series (Taylor at a=0) of f(x)=ex.
Show the steps and the answer
  1. Differentiate repeatedly: f(x)=ex, f′(x)=ex, f′′(x)=ex, and so on, so f(n)(0)=e0=1 for every n.

    The coefficients need the derivatives at the center, and since ex is its own derivative they are all equal.

  2. Divide by n!: the coefficients are 10!,11!,12!,13!,14!, which is 1,1,12,16,124.

    Differentiating xn n times produces n!, so dividing by n! makes the nth derivative of the series exactly 1 at x=0.

  3. Assemble the series: ex=1+x+x22+x36+x424+⋯=∑n=0∞xnn!.

    Each term fixes one more derivative at 0; its own lower derivatives are zero there, so it leaves the earlier matches alone.

  4. Test at x=1: 1+1+0.5+0.1667+0.0417=2.7083, against e≈2.7183.

    Five terms already agree to within 0.01, because the factorials make later terms shrink fast.

Answer: ex=∑n=0∞xnn!

Takeaway: When every derivative at the center is easy to find, the coefficients are those values divided by n!. The series for ex converges for every x.

Example 2 — Geometric, no derivatives needed

Standard

Do you need derivatives to find the sum of ∑n=0∞2nxn, and what is the sum?
Show the steps and the answer
  1. Write out the terms: 1+2x+4x2+8x3+⋯. Each term is 2x times the one before.

    A fixed ratio between neighbouring terms means a geometric series, here with first term 1 and ratio r=2x.

  2. Use the geometric sum 11−r: the series equals 11−2x whenever ∣r∣<1, that is ∣2x∣<1, or ∣x∣<12.

    A geometric series has a closed form, so no derivatives are needed, and the condition ∣r∣<1 gives the interval.

  3. Check at x=0.25: the series is 1+0.5+0.25+0.125+⋯=2, and 11−0.5=2.

    A value inside the interval confirms the closed form.

  4. Compare with the derivative route: for f(x)=(1−2x)−1, f(0)=1, f′(0)=2, f′′(0)=8, f′′′(0)=48, and dividing by 0!,1!,2!,3! gives 1,2,4,8.

    The coefficients match 2n, so this is the Taylor series of 11−2x, but the derivative route took four differentiations to get four terms.

  5. Look outside the interval at x=1: the formula gives −1, while 1+2+4+⋯ diverges.

    The closed form has values everywhere except x=12, but the series equals it only for ∣x∣<12.

Answer: 11−2x for ∣x∣<12 (a geometric series with ratio 2x)

Takeaway: When neighbouring terms have a constant ratio, sum the geometric series directly. Taylor's derivatives give the same coefficients, only more slowly.

Example 3 — Find the mistake: the missing factorials

Application

Find the degree-3 Maclaurin polynomial of sin⁡x. Ravi writes: "f(0)=0, f′(0)=1, f′′(0)=0, f′′′(0)=−1, so sin⁡x≈x−x3." Find the mistake and correct it.
Show the steps and the answer
  1. Check the derivatives: sin⁡x→cos⁡x→−sin⁡x→−cos⁡x, which give 0,1,0,−1 at x=0.

    Ravi's derivative values are right, so the error is in turning them into coefficients.

  2. Divide each by n!: 00!,11!,02!,−13!, which is 0,1,0,−16.

    The third derivative of c x3 is 6c, so c must be −16 for that derivative to equal −1.

  3. Write the corrected polynomial: sin⁡x≈x−x36.

    Only the cubic coefficient changes, because 0! and 1! both equal 1.

  4. Test at x=0.5: sin⁡0.5≈0.4794, the corrected polynomial gives 0.5−0.1256≈0.4792, and Ravi's gives 0.5−0.125=0.375.

    The corrected version is off by less than 0.0003, while Ravi's is off by more than 0.1, which points straight at a wrong coefficient.

Answer: sin⁡x≈x−x36

Takeaway: Each coefficient is f(n)(a)n!. Leaving out the factorial keeps the constant and linear terms right, so the error only shows from the x2 term on.

Where You'll See It

  • A pendulum clock's period

    The pendulum equation contains sin⁡θ, which makes it hard to solve exactly. The Taylor series sin⁡θ=θ−θ36+⋯ says sin⁡θ≈θ for small swings: at 10° (0.1745 rad), sin⁡θ=0.1736, only 0.5% smaller. That shortcut gives the period T=2πL/g, about 2.0 s for a 1 m pendulum.

  • The rule of 72 for doubling money

    Money growing at rate r per year doubles in ln⁡2ln⁡(1+r) years. The Taylor series ln⁡(1+r)=r−r22+⋯ turns this into roughly 0.693r(1+r2). At 8% that is 69.3×1.048≈728=9 years, and the exact answer is 9.01 years.

  • GPS satellite clocks

    Relativity makes a moving clock tick slow by the factor 11−v2/c2. Its Taylor series 1+v22c2+⋯ is far easier to use: a GPS satellite moving at about 3.9 km/s gives v22c2≈8×10−11, which adds up to about 7 microseconds per day. The satellites' clocks are corrected for this, and for gravity's opposite effect.

Common Mistakes

  1. The slip: Forgetting the n! in the denominator

    Instead: each coefficient is f(n)(a)n!, not just f(n)(a).

  2. The slip: Assuming convergence everywhere

    Instead: the series only represents f within its interval of convergence; check it.

  3. The slip: Centering at the wrong point

    Instead: use (x−a) powers about the center a; Maclaurin specifically means a=0.


Mini Practice

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

  1. Find the first three nonzero terms of the Maclaurin series of sin⁡x.

    Hint: The derivatives of sin⁡x at 0 cycle through 0,1,0,−1. Divide each by n!.

  2. Use the first three terms of the Maclaurin series of ex to estimate e0.1.

    Hint: The series starts 1+x+x22.

  3. Find the degree-2 Taylor polynomial of x centred at a=4, and use it to estimate 4.2.

    Hint: You need f(4), f′(4) and f′′(4). Remember to divide f′′(4) by 2!.

  4. Use the Maclaurin series of ex to write the first four terms of the series for e−x2.

    Hint: Substitute −x2 for x in 1+x+x22+x36.

  5. You need cos⁡(0.5) correct to 3 decimal places. Is the tangent line at x=0 enough, or do you need more Taylor terms?

    Hint: Find the tangent line to cos⁡x at 0 and see what it gives at 0.5.

  6. A student writes the Maclaurin series of e2x as "1+2x+4x2+8x3+⋯." Correct the work.

    Hint: Each coefficient is f(n)(0)n!. Was anything divided?

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

Can I know how accurate a Taylor polynomial is before checking with a calculator?

Often, yes. For sin⁡x\sin x the series alternates with shrinking terms, so the error is smaller than the first term you leave out. Using x−x36x-\frac{x^3}{6} for sin⁡0.5\sin0.5 gives 0.4791670.479167, and the next term, 0.55120≈0.00026\frac{0.5^5}{120}\approx0.00026, bounds the error; the true value is 0.4794260.479426. For other functions, the Lagrange remainder bounds the error using the next derivative.

How can derivatives at one point know what the function does far away?

For exe^x, sin⁡x\sin x and cos⁡x\cos x they do: these Maclaurin series converge to the function for every real xx, so the derivatives at 0 determine the whole graph. That is not true of every smooth function. f(x)=e−1/x2f(x)=e^{-1/x^2}, with f(0)=0f(0)=0, has every derivative equal to 0 at the origin, so its Maclaurin series is 0, yet the function is positive for every x≠0x\ne0.

Why does the series for 11+x2 fail beyond ∣x∣=1 when the function is smooth everywhere?

On the real line nothing goes wrong at x=±1x=\pm1, yet the series 1−x2+x4−⋯1-x^2+x^4-\cdots diverges there and beyond. The reason lives in the complex numbers: 1+x2=01+x^2=0 at x=±ix=\pm i, a distance 1 from the center 0. A Taylor series converges on a disc that reaches only as far as the nearest such trouble point, real or complex.

Where do Taylor series show up in physics?

In the pendulum, the equation contains sin⁡θ\sin\theta, and keeping only the first term, sin⁡θ≈θ\sin\theta\approx\theta, gives the simple period formula T=2πL/gT=2\pi\sqrt{L/g} for small swings. Einstein's energy mc21−v2/c2\frac{mc^2}{\sqrt{1-v^2/c^2}} expands as mc2+12mv2+⋯mc^2+\frac12mv^2+\cdots, so the everyday kinetic energy is just the second term of a Taylor series.

Learning Path

Taylor Series

You are here

Next →

Power Series
Build on Derivative and Differentiation Rules (to compute f(n)(a) quickly), Infinite Geometric Series (the model case 11−x=1+x+x2+⋯) and Convergence and Divergence (to know when an infinite sum has a value). Taylor series add a recipe for turning a function into a polynomial from its derivatives at one point. Power Series comes next: the general study of series in powers of (x−c) and where they converge.

See Also