Skip to main content

Code Scholars Tutoring · AP Calculus BC

Lagrange error bounds for Taylor approximations

Unit 10 · Topics 10.12 · Detailed notes, a worked example, and a self-check with reasoning.

Understand the idea

The Lagrange remainder bounds Taylor approximation error using a derivative of one order higher than the polynomial degree. A valid bound controls that derivative throughout the interval between the center and the input being approximated.

Choose a number M\displaystyle M that bounds the absolute value of the required derivative on the whole interval. It need not be the smallest possible bound, but it must be valid. A derivative that happens to be zero at the center need not be zero elsewhere on the interval.

The error estimate uses ∣x−a∣n+1\displaystyle |x-a|^{n+1} divided by (n+1)!\displaystyle (n+1)!. Increasing the degree changes both the derivative to bound and the factorial. The resulting number is an upper bound for absolute error, not a claim that the actual error equals that number.

∣f(x)−Tn(x)∣≤M∣x−a∣n+1(n+1)!,∣f(n+1)(t)∣≤M|f(x)-T_n(x)|\le\frac{M|x-a|^{n+1}}{(n+1)!},\qquad |f^{(n+1)}(t)|\le M

A method you can apply

  1. Identify the polynomial degree n\displaystyle n and the interval from the center to the evaluation input.
  2. Bound ∣f(n+1)(t)∣\displaystyle |f^{(n+1)}(t)| by a valid constant M\displaystyle M over that interval.
  3. Substitute into the remainder inequality and compare the result with the requested tolerance.

Worked example

Bound the error when T2(x)=1+x+x2/2\displaystyle T_2(x)=1+x+x^2/2 approximates e0.1\displaystyle e^{0.1}.

  1. The center is 0\displaystyle 0 and n=2\displaystyle n=2, so the derivative to bound is the third derivative, et\displaystyle e^t.
  2. For 0≤t≤0.1\displaystyle 0\le t\le0.1, et<2\displaystyle e^t<2, so the convenient choice M=2\displaystyle M=2 is valid.
  3. The error is at most 2(0.1)3/3!=0.002/6\displaystyle 2(0.1)^3/3!=0.002/6.
The absolute error is at most 1/3000\displaystyle 1/3000. The polynomial value is 1.105\displaystyle 1.105, with the true value guaranteed to lie within that error distance.

Try it yourself

Using n=2\displaystyle n=2 and the bound M=1\displaystyle M=1, bound the error of 1−x2/2\displaystyle 1-x^2/2 as an approximation to cos⁡(0.2)\displaystyle \cos(0.2).

Show solution and reasoning

The third derivative of cos⁡x\displaystyle \cos x is sin⁡x\displaystyle \sin x, whose absolute value is at most 1\displaystyle 1. Therefore the error is at most (0.2)3/3!=1/750\displaystyle (0.2)^3/3!=1/750. This is a valid bound even though it is not the sharpest available.

AP® is a trademark registered by the College Board, which is not affiliated with, and does not endorse, this material.