Scribe logoScribeNotes & practice
Sign inStart free
  1. Notes
  2. ›AP Calculus BC
  3. ›10 Infinite Sequences and Series
  4. ›10.12 Lagrange error bound

AP Calculus BC · Unit 10 · Infinite Sequences and Series

10.12 Lagrange error bound

Apply the Lagrange error bound to limit the error in a Taylor polynomial and interpret what it says about accuracy.

  • 4 min read
  • 5 parts
  • 10 questions
  • Draft
  • BC only
Start a Quick 5 on this topic→Jump to the 10 questions
1/5Every Taylor approximation leaves a remainder we can limit
  1. 1Every Taylor approximation leaves a remainder we can limit
  2. 2The Lagrange error bound limits the size of the remainder
  3. 3Finding the bound takes four short steps
  4. 4The bound shows how large the degree must be
  5. 5The bound guarantees accuracy without giving the exact error
  6. Check your recall
  7. Practice: 10 questions
1

Part 1 of 5: Every Taylor approximation leaves a remainder we can limit

A Taylor polynomial PnP_nPn​ of degree nnn centered at aaa matches fff and its first nnn derivatives at aaa. It approximates f(x)f(x)f(x), but away from aaa the match is rarely exact.

The gap between the true value and the approximation is the remainder. We write it as Rn(x)=f(x)−Pn(x)R_n(x)=f(x)-P_n(x)Rn​(x)=f(x)−Pn​(x). It can be positive or negative, so we usually care about its size, ∣Rn(x)∣|R_n(x)|∣Rn​(x)∣.

We cannot find the remainder exactly without knowing f(x)f(x)f(x). If we knew f(x)f(x)f(x), we would not need an approximation. So we need a limit that uses only information we have.

Key term

remainder

The remainder Rn(x)R_n(x)Rn​(x) is the difference f(x)−Pn(x)f(x)-P_n(x)f(x)−Pn​(x) between a function and its Taylor polynomial of degree nnn at xxx.

Part 1 done

2

Part 2 of 5: The Lagrange error bound limits the size of the remainder

Now we can state that limit. The Lagrange error bound applies when ∣f(n+1)(t)∣≤M|f^{(n+1)}(t)|\le M∣f(n+1)(t)∣≤M for every ttt in the interval between aaa and xxx.

∣Rn(x)∣≤M(n+1)!∣x−a∣n+1|R_n(x)|\le\frac{M}{(n+1)!}|x-a|^{n+1}∣Rn​(x)∣≤(n+1)!M​∣x−a∣n+1

  • MMM is an upper limit for the size of the (n+1)(n+1)(n+1)th derivative on the interval.
  • (n+1)!(n+1)!(n+1)! is a factorial, which grows very quickly.
  • ∣x−a∣n+1|x-a|^{n+1}∣x−a∣n+1 measures how far xxx is from the center.

The bound looks like the next Taylor term, with MMM standing in for f(n+1)(a)f^{(n+1)}(a)f(n+1)(a). That is why it uses one derivative beyond the degree.

Key terms · tap one for its definition

Lagrange error bound

The Lagrange error bound states that ∣Rn(x)∣≤M(n+1)!∣x−a∣n+1|R_n(x)|\le\frac{M}{(n+1)!}|x-a|^{n+1}∣Rn​(x)∣≤(n+1)!M​∣x−a∣n+1, where MMM limits ∣f(n+1)∣|f^{(n+1)}|∣f(n+1)∣ between aaa and xxx.

interval

An interval specifies the input values from a lower endpoint to an upper endpoint.

Part 2 done

Quick check

One question on the part you just read

Easy

Identify the remainder Rn(x)R_n(x)Rn​(x) for a Taylor polynomial PnP_nPn​ of fff.
Show answer and explanation

Answer: B · Worked explanation

The remainder is the true value minus the approximation, Rn(x)=f(x)−Pn(x)R_n(x)=f(x)-P_n(x)Rn​(x)=f(x)−Pn​(x). Adding the two would not measure a gap between them. The expression Pn(x)P_n(x)Pn​(x) alone is the approximation, and f(n+1)(a)f^{(n+1)}(a)f(n+1)(a) is a derivative value that appears in the bound but is not itself the error.
3

Part 3 of 5: Finding the bound takes four short steps

Applying the bound follows the same pattern each time.

  • Identify nnn, aaa, and xxx.
  • Find the (n+1)(n+1)(n+1)th derivative of fff.
  • Choose MMM, a value at least as large as ∣f(n+1)∣|f^{(n+1)}|∣f(n+1)∣ on the interval.
  • Substitute into M(n+1)!∣x−a∣n+1\frac{M}{(n+1)!}|x-a|^{n+1}(n+1)!M​∣x−a∣n+1.

The value MMM need not be the exact maximum. Any value that stays above ∣f(n+1)∣|f^{(n+1)}|∣f(n+1)∣ on the interval is valid. A tighter MMM simply gives a tighter bound. Sketch the interval first, so you can see where f(n+1)f^{(n+1)}f(n+1) is largest.

Worked example

Bounding an error for the exponential function

The polynomial P3(x)=1+x+x22+x36P_3(x)=1+x+\frac{x^2}{2}+\frac{x^3}{6}P3​(x)=1+x+2x2​+6x3​ approximates exe^xex near 0. Find a Lagrange error bound for P3(0.5)P_3(0.5)P3​(0.5).
  1. 1
    Here n=3n=3n=3, a=0a=0a=0, and x=0.5x=0.5x=0.5. The needed derivative is f(4)(t)=etf^{(4)}(t)=e^tf(4)(t)=et.
  2. 2
    On the interval from 0 to 0.5, ete^tet is increasing. So et≤e0.5<2e^t\le e^{0.5}<2et≤e0.5<2, since e<4e<4e<4.
  3. 3
    Take M=2M=2M=2.
  4. 4
    Substitute: ∣R3(0.5)∣≤24!(0.5)4=224⋅0.0625|R_3(0.5)|\le\frac{2}{4!}(0.5)^4=\frac{2}{24}\cdot0.0625∣R3​(0.5)∣≤4!2​(0.5)4=242​⋅0.0625.
  5. 5
    Simplify: 0.12524=1192≈0.005\frac{0.125}{24}=\frac1{192}\approx0.005240.125​=1921​≈0.005.

Answer: ∣e0.5−P3(0.5)∣≤1192≈0.005|e^{0.5}-P_3(0.5)|\le\frac1{192}\approx0.005∣e0.5−P3​(0.5)∣≤1921​≈0.005.

Common mistake

Use the (n+1)(n+1)(n+1)th derivative, one beyond the degree. Here n=3n=3n=3, so the fourth derivative is the one to limit.

Part 3 done

4

Part 4 of 5: The bound shows how large the degree must be

Sometimes the problem runs the other way. You are given a target accuracy and must choose the degree.

Set the bound at or below the target, then test increasing values of nnn. The factorial in the denominator makes the bound shrink quickly. The table shows this when M=1M=1M=1 and ∣x−a∣=1|x-a|=1∣x−a∣=1. Only nnn changes in it, so every change in the bound comes from the factorial.

The bound 1(n+1)!\frac{1}{(n+1)!}(n+1)!1​ for M=1M=1M=1 and ∣x−a∣=1|x-a|=1∣x−a∣=1

Degree nnnBound
3124≈0.0417\frac1{24}\approx0.0417241​≈0.0417
51720≈0.00139\frac1{720}\approx0.001397201​≈0.00139
7140,320≈0.0000248\frac1{40{,}320}\approx0.000024840,3201​≈0.0000248
Degree $n$
3
Bound
124≈0.0417\frac1{24}\approx0.0417241​≈0.0417
Degree $n$
5
Bound
1720≈0.00139\frac1{720}\approx0.001397201​≈0.00139
Degree $n$
7
Bound
140,320≈0.0000248\frac1{40{,}320}\approx0.000024840,3201​≈0.0000248

Worked example

Choosing the degree for sine

Find the smallest nnn so that the Lagrange error bound guarantees ∣sin⁡1−Pn(1)∣≤0.001|\sin 1-P_n(1)|\le0.001∣sin1−Pn​(1)∣≤0.001, using center a=0a=0a=0.
  1. 1
    Every derivative of sin⁡x\sin xsinx is ±sin⁡x\pm\sin x±sinx or ±cos⁡x\pm\cos x±cosx, so ∣f(n+1)(t)∣≤1|f^{(n+1)}(t)|\le1∣f(n+1)(t)∣≤1. Take M=1M=1M=1.
  2. 2
    The bound is 1(n+1)!⋅1n+1=1(n+1)!\frac{1}{(n+1)!}\cdot1^{n+1}=\frac{1}{(n+1)!}(n+1)!1​⋅1n+1=(n+1)!1​.
  3. 3
    Require 1(n+1)!≤0.001\frac{1}{(n+1)!}\le0.001(n+1)!1​≤0.001, so (n+1)!≥1000(n+1)!\ge1000(n+1)!≥1000.
  4. 4
    Since 6!6!6! is 720 and 7!7!7! is 5,040, we need n+1≥7n+1\ge7n+1≥7.
  5. 5
    So n≥6n\ge6n≥6.

Answer: The smallest degree the bound guarantees is n=6n=6n=6.

Exam tip

Show the bound for the largest nnn that fails and the smallest nnn that works. That comparison justifies your choice.

Part 4 done

Quick check

One question on the part you just read

Easy

Identify which derivative of fff is limited by MMM for a Taylor polynomial of degree 4.
Show answer and explanation

Answer: C · Worked explanation

The bound for degree nnn uses the (n+1)(n+1)(n+1)th derivative, so for n=4n=4n=4 it is the fifth. The fourth derivative is the most tempting wrong answer because it matches the degree. The polynomial already uses that derivative at the center, so the error depends on the next derivative. The third and sixth derivatives have no role in the bound.
5

Part 5 of 5: The bound guarantees accuracy without giving the exact error

The bound is a guarantee about distance. If it equals 0.005, then f(x)f(x)f(x) lies between Pn(x)−0.005P_n(x)-0.005Pn​(x)−0.005 and Pn(x)+0.005P_n(x)+0.005Pn​(x)+0.005. A smaller bound means a narrower interval and a more precise approximation.

Three things make the bound small.

  • A small distance ∣x−a∣|x-a|∣x−a∣ from the center.
  • A large degree nnn.
  • A small derivative limit MMM.

Raising nnn helps because (n+1)!(n+1)!(n+1)! eventually grows faster than ∣x−a∣n+1|x-a|^{n+1}∣x−a∣n+1. For sin⁡x\sin xsinx with P3(x)=x−x36P_3(x)=x-\frac{x^3}{6}P3​(x)=x−6x3​ and M=1M=1M=1, the bound is x424\frac{x^4}{24}24x4​. The graph compares it with the actual error. At x=1x=1x=1 the bound is 124≈0.042\frac1{24}\approx0.042241​≈0.042, while the actual error is about 0.008.

Figure 1. The bound stays above the actual error for every input shown, and both grow with distance from the center.

Common mistake

The bound is not the error. The actual error is usually smaller than the bound.

Go deeperWhere the bound comes from

Lagrange’s form of the remainder says there is a number ccc between aaa and xxx with Rn(x)=f(n+1)(c)(n+1)!(x−a)n+1R_n(x)=\frac{f^{(n+1)}(c)}{(n+1)!}(x-a)^{n+1}Rn​(x)=(n+1)!f(n+1)(c)​(x−a)n+1.

We cannot find ccc. So we replace ∣f(n+1)(c)∣|f^{(n+1)}(c)|∣f(n+1)(c)∣ by the limit MMM, which is at least as large, and the bound follows.

Note read

Check your recall[14 marks]

6 written questions. Answer each one in your head or on paper, then open the mark scheme.

  1. 1.Identify[2 marks]

    Identify which derivative of fff is limited by MMM in the Lagrange error bound for a Taylor polynomial of degree nnn, and where the limit must hold.

    Show mark schemeHide mark scheme

    Model answer

    The (n+1)(n+1)(n+1)th derivative is limited, so ∣f(n+1)(t)∣≤M|f^{(n+1)}(t)|\le M∣f(n+1)(t)∣≤M for every ttt in the interval between aaa and xxx.

    • 1 mark for identifying the (n+1)(n+1)(n+1)th derivative
    • 1 mark for stating that the limit applies on the interval between aaa and xxx
  2. 2.Calculate[3 marks]

    Calculate the Lagrange error bound for approximating cos⁡0.6\cos 0.6cos0.6 with P2(x)=1−x22P_2(x)=1-\frac{x^2}{2}P2​(x)=1−2x2​ centered at 0. Use M=1M=1M=1, since ∣f′′′(t)∣=∣sin⁡t∣≤1|f'''(t)|=|\sin t|\le1∣f′′′(t)∣=∣sint∣≤1.

    Show mark schemeHide mark scheme

    Model answer

    Here n=2n=2n=2, so the bound is 13!(0.6)3=0.2166=0.036\frac{1}{3!}(0.6)^3=\frac{0.216}{6}=0.0363!1​(0.6)3=60.216​=0.036.

    • 1 mark for using n=2n=2n=2 and (n+1)!=3!(n+1)!=3!(n+1)!=3!
    • 1 mark for substituting M=1M=1M=1 and ∣x−a∣=0.6|x-a|=0.6∣x−a∣=0.6
    • 1 mark for obtaining 0.036
  3. 3.Interpret[2 marks]

    Interpret a Lagrange error bound of 0.036 for the approximation P2(0.6)P_2(0.6)P2​(0.6) of cos⁡0.6\cos 0.6cos0.6.

    Show mark schemeHide mark scheme

    Model answer

    The actual error has size at most 0.036, so cos⁡0.6\cos 0.6cos0.6 lies within 0.036 of P2(0.6)P_2(0.6)P2​(0.6). The actual error may be smaller.

    • 1 mark for stating that the size of the error is at most 0.036
    • 1 mark for stating that cos⁡0.6\cos 0.6cos0.6 lies within 0.036 of P2(0.6)P_2(0.6)P2​(0.6)
  4. 4.Determine[3 marks]

    Determine the smallest nnn for which the Lagrange error bound guarantees ∣sin⁡2−Pn(2)∣<0.01|\sin 2-P_n(2)|<0.01∣sin2−Pn​(2)∣<0.01, using center 0 and M=1M=1M=1.

    Show mark schemeHide mark scheme

    Model answer

    The bound is 2n+1(n+1)!\frac{2^{n+1}}{(n+1)!}(n+1)!2n+1​. For n=6n=6n=6 it is 1285040≈0.0254\frac{128}{5040}\approx0.02545040128​≈0.0254, which is too large. For n=7n=7n=7 it is 25640,320≈0.00635\frac{256}{40{,}320}\approx0.0063540,320256​≈0.00635, which is below 0.01. So n=7n=7n=7.

    • 1 mark for writing the bound as 2n+1(n+1)!\frac{2^{n+1}}{(n+1)!}(n+1)!2n+1​
    • 1 mark for showing the bound is too large for n=6n=6n=6
    • 1 mark for showing the bound is below 0.01 for n=7n=7n=7
  5. 5.Explain[2 marks]

    Explain why increasing the degree nnn usually reduces the Lagrange error bound at a fixed input.

    Show mark schemeHide mark scheme

    Model answer

    The factorial (n+1)!(n+1)!(n+1)! in the denominator grows faster than ∣x−a∣n+1|x-a|^{n+1}∣x−a∣n+1 in the numerator. So the fraction becomes smaller.

    • 1 mark for stating that (n+1)!(n+1)!(n+1)! grows very quickly
    • 1 mark for concluding that the bound becomes smaller
  6. 6.Verify[2 marks]

    Verify that the actual error in estimating sin⁡0.5\sin 0.5sin0.5 by P3(0.5)=0.5−0.1256P_3(0.5)=0.5-\frac{0.125}{6}P3​(0.5)=0.5−60.125​ is within the bound 0.0026. Use sin⁡0.5≈0.479426\sin 0.5\approx0.479426sin0.5≈0.479426.

    Show mark schemeHide mark scheme

    Model answer

    P3(0.5)≈0.479167P_3(0.5)\approx0.479167P3​(0.5)≈0.479167. The error is 0.479426−0.479167≈0.000260.479426-0.479167\approx0.000260.479426−0.479167≈0.00026, which is less than 0.0026.

    • 1 mark for computing P3(0.5)≈0.479167P_3(0.5)\approx0.479167P3​(0.5)≈0.479167
    • 1 mark for finding an error of about 0.00026 and comparing it with 0.0026

Practice this topic

10 quick-fire multiple-choice questions, easy to hard. Tap an answer to see if you're right and the worked explanation. (The written questions with mark schemes are under “Check your recall” above.)

0 right0 of 10 answered

Includes the 2 quick checks in the note above.

3.

Easy·Identify[1 mark]

A degree 3 polynomial is centered at 0, and M=2M=2M=2 in each case. Identify the input that gives the smallest Lagrange error bound.
Show answer and explanation

Answer: D · Worked explanation

The bound is 24!∣x∣4\frac{2}{4!}|x|^44!2​∣x∣4, which increases with ∣x∣|x|∣x∣. The input x=0.3x=0.3x=0.3 is closest to the center, so it gives the smallest bound of the four inputs. The input x=−0.5x=-0.5x=−0.5 is negative, but only ∣x∣|x|∣x∣ matters, and ∣−0.5∣=0.5|-0.5|=0.5∣−0.5∣=0.5 is larger than 0.3.
4.

Medium·Calculate[2 marks]

A function has ∣f′′′(t)∣≤6|f'''(t)|\le6∣f′′′(t)∣≤6 for every ttt between 0 and 0.5. Calculate the Lagrange error bound for P2(0.5)P_2(0.5)P2​(0.5) centered at a=0a=0a=0.
Show answer and explanation

Answer: B · Worked explanation

With n=2n=2n=2, the bound is 63!(0.5)3=1×0.125=0.125\frac{6}{3!}(0.5)^3=1\times0.125=0.1253!6​(0.5)3=1×0.125=0.125. The value 0.250 uses the square of 0.5 instead of the cube, which is the wrong power of the distance. The value 0.375 divides by 2!2!2! instead of 3!3!3!, and 0.750 leaves out the factorial.
5.

Medium·Calculate[2 marks]

A degree 3 polynomial is centered at a=2a=2a=2, and ∣f(4)(t)∣≤5|f^{(4)}(t)|\le5∣f(4)(t)∣≤5 near 2. Calculate the Lagrange error bound at x=2.5x=2.5x=2.5.
Show answer and explanation

Answer: C · Worked explanation

The bound is 54!(0.5)4=524(0.0625)≈0.013\frac{5}{4!}(0.5)^4=\frac{5}{24}(0.0625)\approx0.0134!5​(0.5)4=245​(0.0625)≈0.013. The value 0.052 divides by 3!3!3! instead of 4!4!4!, which is the most tempting slip. The value 0.313 leaves out the factorial entirely, and 8.138 uses x=2.5x=2.5x=2.5 instead of the distance ∣x−a∣=0.5|x-a|=0.5∣x−a∣=0.5 from the center. A bound built on the wrong distance is far too large.
6.

Medium·Interpret[2 marks]

The Lagrange error bound for P4(1.2)P_4(1.2)P4​(1.2) is 0.0008. Which statement is guaranteed?
Show answer and explanation

Answer: B · Worked explanation

A bound limits the size of the error, so ∣f(1.2)−P4(1.2)∣≤0.0008|f(1.2)-P_4(1.2)|\le0.0008∣f(1.2)−P4​(1.2)∣≤0.0008. It does not say the error equals 0.0008, and it does not give a lower limit on the error. The bound also tells us nothing about P5P_5P5​, which has its own separate bound, so the last statement is not guaranteed.
7.

Medium·Find[2 marks]

For f(x)=exf(x)=e^xf(x)=ex on the interval from 0 to 1, find the smallest valid MMM for a degree 2 polynomial centered at 0.
Show answer and explanation

Answer: C · Worked explanation

Here f′′′(t)=etf'''(t)=e^tf′′′(t)=et, which is increasing, so its largest value on [0,1][0,1][0,1] is at t=1t=1t=1, giving e≈2.718e\approx2.718e≈2.718. The value 1 is the value at the center, not the maximum. The value e6\frac e66e​ already includes the division by 3!3!3!, which belongs in the bound formula. The value e2e^2e2 would be a valid limit but it is not the smallest one, so it fails the question.
8.

Hard·Determine[2 marks]

Determine the smallest nnn for which the Lagrange error bound guarantees ∣cos⁡1−Pn(1)∣<0.0001|\cos 1-P_n(1)|<0.0001∣cos1−Pn​(1)∣<0.0001, using center 0 and M=1M=1M=1.
Show answer and explanation

Answer: D · Worked explanation

The bound is 1(n+1)!\frac{1}{(n+1)!}(n+1)!1​, so we need (n+1)!>10,000(n+1)!>10{,}000(n+1)!>10,000. Since 7!=50407!=50407!=5040 is too small and 8!=40,3208!=40{,}3208!=40,320 works, n+1=8n+1=8n+1=8 and n=7n=7n=7. The value 8 comes from using n!n!n! instead of (n+1)!(n+1)!(n+1)!, which shifts the answer up by one and gives a degree that is one too large.
9.

Hard·Calculate[2 marks]

For f(x)=ln⁡xf(x)=\ln xf(x)=lnx and P2P_2P2​ centered at a=1a=1a=1, calculate the Lagrange error bound at x=1.5x=1.5x=1.5, using the smallest valid MMM.
Show answer and explanation

Answer: C · Worked explanation

Here f′′′(t)=2t3f'''(t)=\frac{2}{t^3}f′′′(t)=t32​, which is largest at t=1t=1t=1, so M=2M=2M=2. The bound is 23!(0.5)3=13(0.125)≈0.042\frac{2}{3!}(0.5)^3=\frac13(0.125)\approx0.0423!2​(0.5)3=31​(0.125)≈0.042. The value 0.012 uses MMM at t=1.5t=1.5t=1.5, which is not the maximum. The value 0.083 squares 0.5 instead of cubing it, and 0.125 divides by 2!2!2! instead of 3!3!3!.
10.

Hard·Predict[2 marks]

At the same input, the Lagrange error bound for P3(x)P_3(x)P3​(x) is 0.04 and for P4(x)P_4(x)P4​(x) is 0.005. Which conclusion is guaranteed?
Show answer and explanation

Answer: C · Worked explanation

The bound guarantees only a limit, so the error of P4(x)P_4(x)P4​(x) is at most 0.005. A smaller bound does not prove that P4P_4P4​ is closer than P3P_3P3​, because actual errors may fall below their bounds by different amounts. A bound is never the exact error.

Keep going

Next topic10.13 Radius and interval of convergencePrevious topic10.11 Taylor polynomial approximations

More in Unit 10 · Infinite Sequences and Series

  1. 10.7Alternating series testNotes · 10 questions· BC only
  2. 10.8Ratio testNotes · 10 questions· BC only
  3. 10.9Absolute and conditional convergenceNotes · 10 questions· BC only
  4. 10.10Alternating series error boundNotes · 10 questions· BC only
  5. 10.11Taylor polynomial approximationsNotes · 10 questions· BC only
  6. 10.12Lagrange error boundNotes · 10 questions· BC only
  7. 10.13Radius and interval of convergenceNotes · 10 questions· BC only
  8. 10.14Taylor and Maclaurin seriesNotes · 10 questions· BC only
  9. 10.15Representing functions as power seriesNotes · 10 questions· BC only

All AP Calculus BC topics →

Draft: written by Scribe and checked automatically, not yet by a subject reviewer.

Practice questions are original, written by Scribe and checked automatically for correctness.

Topic codes follow the official numbering. Not endorsed by College Board or any exam board.

On this page

0/5 read

  1. 1Every Taylor approximation leaves a remainder we can limit
  2. 2The Lagrange error bound limits the size of the remainder
  3. 3Finding the bound takes four short steps
  4. 4The bound shows how large the degree must be
  5. 5The bound guarantees accuracy without giving the exact error
  6. Check your recall
  7. Practice: 10 questions
All coursesScribeHow it worksPricingPrivacyTerms

Notes and questions are written by Scribe in our own words. Topic codes follow the official course numbering so you can find your place. Not produced or endorsed by the IB, College Board, AQA, OCR, Pearson Edexcel or any exam board.

© 2026 Scribe

Actual error and the Lagrange bound

Size of error-2-101200.20.40.60.8x (input)
  • Actual error
  • Lagrange bound