## Taylor’s Theorem with Remainder and Convergence

### Learning Outcomes

• Explain the meaning and significance of Taylor’s theorem with remainder
• Estimate the remainder for a Taylor series approximation of a given function

## Taylor’s Theorem with Remainder

Recall that the $n$th Taylor polynomial for a function $f$ at $a$ is the $n$th partial sum of the Taylor series for $f$ at $a$. Therefore, to determine if the Taylor series converges, we need to determine whether the sequence of Taylor polynomials $\left\{{p}_{n}\right\}$ converges. However, not only do we want to know if the sequence of Taylor polynomials converges, we want to know if it converges to $f$. To answer this question, we define the remainder ${R}_{n}\left(x\right)$ as

${R}_{n}\left(x\right)=f\left(x\right)-{p}_{n}\left(x\right)$.

For the sequence of Taylor polynomials to converge to $f$, we need the remainder $R_{n}$ to converge to zero. To determine if $R_{n}$ converges to zero, we introduce Taylor’s theorem with remainder. Not only is this theorem useful in proving that a Taylor series converges to its related function, but it will also allow us to quantify how well the $n$th Taylor polynomial approximates the function.

Here we look for a bound on $|{R}_{n}|$. Consider the simplest case: $n=0$. Let $p_{0}$ be the 0th Taylor polynomial at $a$ for a function $f$. The remainder $R_{0}$ satisfies

$\begin{array}{cc}\hfill {R}_{0}\left(x\right)& =f\left(x\right)-{p}_{0}\left(x\right)\hfill \\ & =f\left(x\right)-f\left(a\right).\hfill \end{array}$

If $f$ is differentiable on an interval $I$ containing $a$ and $x$, then by the Mean Value Theorem there exists a real number $c$ between $a$ and $x$ such that $f\left(x\right)-f\left(a\right)={f}^{\prime }\left(c\right)\left(x-a\right)$. Therefore,

${R}_{0}\left(x\right)={f}^{\prime }\left(c\right)\left(x-a\right)$.

Using the Mean Value Theorem in a similar argument, we can show that if $f$ is $n$ times differentiable on an interval $I$ containing $a$ and $x$, then the $n$th remainder ${R}_{n}$ satisfies

${R}_{n}\left(x\right)=\frac{{f}^{\left(n+1\right)}\left(c\right)}{\left(n+1\right)\text{!}}{\left(x-a\right)}^{n+1}$

for some real number $c$ between $a$ and $x$. It is important to note that the value $c$ in the numerator above is not the center $a$, but rather an unknown value $c$ between $a$ and $x$. This formula allows us to get a bound on the remainder ${R}_{n}$. If we happen to know that $|{f}^{\left(n+1\right)}\left(x\right)|$ is bounded by some real number $M$ on this interval $I$, then

$|{R}_{n}\left(x\right)|\le \frac{M}{\left(n+1\right)\text{!}}{|x-a|}^{n+1}$

for all $x$ in the interval $I$.

We now state Taylor’s theorem, which provides the formal relationship between a function $f$ and its $n$th degree Taylor polynomial ${p}_{n}\left(x\right)$. This theorem allows us to bound the error when using a Taylor polynomial to approximate a function value, and will be important in proving that a Taylor series for $f$ converges to $f$.

### theorem: Taylor’s Theorem with Remainder

Let $f$ be a function that can be differentiated $n+1$ times on an interval $I$ containing the real number $a$. Let $p_{n}$ be the $n$th Taylor polynomial of $f$ at $a$ and let

${R}_{n}\left(x\right)=f\left(x\right)-{p}_{n}\left(x\right)$

be the $n$th remainder. Then for each $x$ in the interval $I$, there exists a real number $c$ between $a$ and $x$ such that

${R}_{n}\left(x\right)=\frac{{f}^{\left(n+1\right)}\left(c\right)}{\left(n+1\right)\text{!}}{\left(x-a\right)}^{n+1}$.

If there exists a real number $M$ such that $|{f}^{\left(n+1\right)}\left(x\right)|\le M$ for all $x\in I$, then

$|{R}_{n}\left(x\right)|\le \frac{M}{\left(n+1\right)\text{!}}{|x-a|}^{n+1}$

for all $x$ in $I$.

#### Proof

Fix a point $x\in I$ and introduce the function g such that

$g\left(t\right)=f\left(x\right)-f\left(t\right)-{f}^{\prime }\left(t\right)\left(x-t\right)-\frac{f^{\prime\prime}\left(t\right)}{2\text{!}}{\left(x-t\right)}^{2}-\cdots -\frac{{f}^{\left(n\right)}\left(t\right)}{n\text{!}}{\left(x-t\right)}^{n}-{R}_{n}\left(x\right)\frac{{\left(x-t\right)}^{n+1}}{{\left(x-a\right)}^{n+1}}$.

We claim that g satisfies the criteria of Rolle’s theorem. Since g is a polynomial function (in t), it is a differentiable function. Also, g is zero at $t=a$ and $t=x$ because

$\begin{array}{ccc}\hfill g\left(a\right)& =\hfill & f\left(x\right)-f\left(a\right)-{f}^{\prime }\left(a\right)\left(x-a\right)-\frac{f^{\prime\prime}\left(a\right)}{2\text{!}}{\left(x-a\right)}^{2}+\cdots +\frac{{f}^{\left(n\right)}\left(a\right)}{n\text{!}}{\left(x-a\right)}^{n}-{R}_{n}\left(x\right)\hfill \\ & =\hfill & f\left(x\right)-{p}_{n}\left(x\right)-{R}_{n}\left(x\right)\hfill \\ & =\hfill & 0,\hfill \\ g\left(x\right)\hfill & =\hfill & f\left(x\right)-f\left(x\right)-0-\cdots -0\hfill \\ & =\hfill & 0.\hfill \end{array}$

Therefore, g satisfies Rolle’s theorem, and consequently, there exists c between a and x such that ${g}^{\prime }\left(c\right)=0$. We now calculate ${g}^{\prime }$. Using the product rule, we note that

$\frac{d}{dt}\left[\frac{{f}^{\left(n\right)}\left(t\right)}{n\text{!}}{\left(x-t\right)}^{n}\right]=\frac{-{f}^{\left(n\right)}\left(t\right)}{\left(n - 1\right)\text{!}}{\left(x-t\right)}^{n - 1}+\frac{{f}^{\left(n+1\right)}\left(t\right)}{n\text{!}}{\left(x-t\right)}^{n}$.

Consequently,

$\begin{array}{cc}{g}^{\prime }\left(t\right)\hfill & = -{f}^{\prime }\left(t\right)+\left[{f}^{\prime }\left(t\right)-f^{\prime\prime}\left(t\right)\left(x-t\right)\right]+\left[f^{\prime\prime}\left(t\right)\left(x-t\right)-\frac{f^{\prime\prime\prime}\left(t\right)}{2\text{!}}{\left(x-t\right)}^{2}\right]+\cdots \hfill \\ & +\left[\frac{{f}^{\left(n\right)}\left(t\right)}{\left(n - 1\right)\text{!}}{\left(x-t\right)}^{n - 1}-\frac{{f}^{\left(n+1\right)}\left(t\right)}{n\text{!}}{\left(x-t\right)}^{n}\right]+\left(n+1\right){R}_{n}\left(x\right)\frac{{\left(x-t\right)}^{n}}{{\left(x-a\right)}^{n+1}}.\hfill \end{array}$

Notice that there is a telescoping effect. Therefore,

${g}^{\prime }\left(t\right)=-\frac{{f}^{\left(n+1\right)}\left(t\right)}{n\text{!}}{\left(x-t\right)}^{n}+\left(n+1\right){R}_{n}\left(x\right)\frac{{\left(x-t\right)}^{n}}{{\left(x-a\right)}^{n+1}}$.

By Rolle’s theorem, we conclude that there exists a number c between a and x such that ${g}^{\prime }\left(c\right)=0$. Since

${g}^{\prime }\left(c\right)=-\frac{{f}^{\left(n+1\right)}\left(c\right)}{n\text{!}}{\left(x-c\right)}^{n}+\left(n+1\right){R}_{n}\left(x\right)\frac{{\left(x-c\right)}^{n}}{{\left(x-a\right)}^{n+1}}$

we conclude that

$-\frac{{f}^{\left(n+1\right)}\left(c\right)}{n\text{!}}{\left(x-c\right)}^{n}+\left(n+1\right){R}_{n}\left(x\right)\frac{{\left(x-c\right)}^{n}}{{\left(x-a\right)}^{n+1}}=0$.

Adding the first term on the left-hand side to both sides of the equation and dividing both sides of the equation by $n+1$, we conclude that

${R}_{n}\left(x\right)=\frac{{f}^{\left(n+1\right)}\left(c\right)}{\left(n+1\right)\text{!}}{\left(x-a\right)}^{n+1}$

as desired. From this fact, it follows that if there exists M such that $|{f}^{\left(n+1\right)}\left(x\right)|\le M$ for all x in I, then

$|{R}_{n}\left(x\right)|\le \frac{M}{\left(n+1\right)\text{!}}{|x-a|}^{n+1}$.

$_\blacksquare$

Not only does Taylor’s theorem allow us to prove that a Taylor series converges to a function, but it also allows us to estimate the accuracy of Taylor polynomials in approximating function values. We begin by looking at linear and quadratic approximations of $f\left(x\right)=\sqrt{x}$ at $x=8$ and determine how accurate these approximations are at estimating $\sqrt{11}$.

### Example: Using Linear and Quadratic Approximations to Estimate Function Values

Consider the function $f\left(x\right)=\sqrt{x}$.

1. Find the first and second Taylor polynomials for $f$ at $x=8$. Use a graphing utility to compare these polynomials with $f$ near $x=8$.
2. Use these two polynomials to estimate $\sqrt{11}$.
3. Use Taylor’s theorem to bound the error.

Watch the following video to see the worked solution to Example: Using Linear and Quadratic Approximations to Estimate Function Values.

For closed captioning, open the video on its original page by clicking the Youtube logo in the lower right-hand corner of the video display. In YouTube, the video will begin at the same starting point as this clip, but will continue playing until the very end.

### try it

Find the first and second Taylor polynomials for $f\left(x\right)=\sqrt{x}$ at $x=4$. Use these polynomials to estimate $\sqrt{6}$. Use Taylor’s theorem to bound the error.

### Example: Approximating sinx using maclaurin polynomials

From the Example: Finding Maclaurin Polynomials, the Maclaurin polynomials for $\sin{x}$ are given by

$\begin{array}{cc}\hfill {p}_{2m+1}\left(x\right)& ={p}_{2m+2}\left(x\right)\hfill \\ & =x-\frac{{x}^{3}}{3\text{!}}+\frac{{x}^{5}}{5\text{!}}-\frac{{x}^{7}}{7\text{!}}+\cdots +{\left(-1\right)}^{m}\frac{{x}^{2m+1}}{\left(2m+1\right)\text{!}}\hfill \end{array}$

for $m=0,1,2,\dots$.

1. Use the fifth Maclaurin polynomial for $\sin{x}$ to approximate $\sin\left(\frac{\pi }{18}\right)$ and bound the error.
2. For what values of $x$ does the fifth Maclaurin polynomial approximate $\sin{x}$ to within 0.0001?

Watch the following video to see the worked solution to Example: Approximating sin x Using Maclaurin Polynomials.

For closed captioning, open the video on its original page by clicking the Youtube logo in the lower right-hand corner of the video display. In YouTube, the video will begin at the same starting point as this clip, but will continue playing until the very end.

### try it

Use the fourth Maclaurin polynomial for $\cos{x}$ to approximate $\cos\left(\frac{\pi }{12}\right)$.

Now that we are able to bound the remainder ${R}_{n}\left(x\right)$, we can use this bound to prove that a Taylor series for $f$ at $a$ converges to $f$.

## Representing Functions with Taylor and Maclaurin Series

We now discuss issues of convergence for Taylor series. We begin by showing how to find a Taylor series for a function, and how to find its interval of convergence.

### Example: Finding a Taylor Series

Find the Taylor series for $f\left(x\right)=\frac{1}{x}$ at $x=1$. Determine the interval of convergence.

Watch the following video to see the worked solution to Example: Finding a Taylor Series.

For closed captioning, open the video on its original page by clicking the Youtube logo in the lower right-hand corner of the video display. In YouTube, the video will begin at the same starting point as this clip, but will continue playing until the very end.

### try it

Find the Taylor series for $f\left(x\right)=\frac{1}{2x}$ at $x=2$ and determine its interval of convergence.

We know that the Taylor series found in this example converges on the interval $\left(0,2\right)$, but how do we know it actually converges to $f?$ We consider this question in more generality in a moment, but for this example, we can answer this question by writing

$f\left(x\right)=\frac{1}{x}=\frac{1}{1-\left(1-x\right)}$.

That is, $f$ can be represented by the geometric series $\displaystyle\sum _{n=0}^{\infty }{\left(1-x\right)}^{n}$. Since this is a geometric series, it converges to $\frac{1}{x}$ as long as $|1-x|<1$. Therefore, the Taylor series found in the previous example does converge to $f\left(x\right)=\frac{1}{x}$ on $\left(0,2\right)$.

We now consider the more general question: if a Taylor series for a function $f$ converges on some interval, how can we determine if it actually converges to $f?$ To answer this question, recall that a series converges to a particular value if and only if its sequence of partial sums converges to that value. Given a Taylor series for $f$ at a, the nth partial sum is given by the nth Taylor polynomial pn. Therefore, to determine if the Taylor series converges to $f$, we need to determine whether

$\underset{n\to \infty }{\text{lim}}{p}_{n}\left(x\right)=f\left(x\right)$.

Since the remainder ${R}_{n}\left(x\right)=f\left(x\right)-{p}_{n}\left(x\right)$, the Taylor series converges to $f$ if and only if

$\underset{n\to \infty }{\text{lim}}{R}_{n}\left(x\right)=0$.

We now state this theorem formally.

### Theorem: Convergence of Taylor Series

Suppose that $f$ has derivatives of all orders on an interval $I$ containing $a$. Then the Taylor series

$\displaystyle\sum _{n=0}^{\infty }\frac{{f}^{\left(n\right)}\left(a\right)}{n\text{!}}{\left(x-a\right)}^{n}$

converges to $f\left(x\right)$ for all $x$ in $I$ if and only if

$\underset{n\to\infty}\lim {R}_{n}\left(x\right)=0$

for all $x$ in $I$.

With this theorem, we can prove that a Taylor series for $f$ at a converges to $f$ if we can prove that the remainder ${R}_{n}\left(x\right)\to 0$. To prove that ${R}_{n}\left(x\right)\to 0$, we typically use the bound

$|{R}_{n}\left(x\right)|\le \frac{M}{\left(n+1\right)\text{!}}{|x-a|}^{n+1}$

from Taylor’s theorem with remainder.

In the next example, we find the Maclaurin series for ex and $\sin{x}$ and show that these series converge to the corresponding functions for all real numbers by proving that the remainders ${R}_{n}\left(x\right)\to 0$ for all real numbers x.

### Example: Finding Maclaurin Series

For each of the following functions, find the Maclaurin series and its interval of convergence. Use Taylor’s Theorem with Remainder to prove that the Maclaurin series for $f$ converges to $f$ on that interval.

1. $e^{x}$
2. $\sin{x}$

Watch the following video to see the worked solution to Example: Finding Maclaurin Series.

For closed captioning, open the video on its original page by clicking the Youtube logo in the lower right-hand corner of the video display. In YouTube, the video will begin at the same starting point as this clip, but will continue playing until the very end.

### try it

Find the Maclaurin series for $f\left(x\right)=\cos{x}$. Use the ratio test to show that the interval of convergence is $\left(-\infty ,\infty \right)$. Show that the Maclaurin series converges to $\cos{x}$ for all real numbers $x$.

### Activity: Proving that $e$ is Irrational

In this project, we use the Maclaurin polynomials for $e^{x}$ to prove that e is irrational. The proof relies on supposing that e is rational and arriving at a contradiction. Therefore, in the following steps, we suppose $e=\frac{r}{s}$ for some integers $r$ and $s$ where $s\ne 0$.

1. Write the Maclaurin polynomials ${p}_{0}\left(x\right),{p}_{1}\left(x\right),{p}_{2}\left(x\right),{p}_{3}\left(x\right),{p}_{4}\left(x\right)$ for ex. Evaluate ${p}_{0}\left(1\right),{p}_{1}\left(1\right),{p}_{2}\left(1\right),{p}_{3}\left(1\right),{p}_{4}\left(1\right)$ to estimate e.
2. Let ${R}_{n}\left(x\right)$ denote the remainder when using ${p}_{n}\left(x\right)$ to estimate $e^{x}$. Therefore, ${R}_{n}\left(x\right)={e}^{x}-{p}_{n}\left(x\right)$, and ${R}_{n}\left(1\right)=e-{p}_{n}\left(1\right)$. Assuming that $e=\frac{r}{s}$ for integers r and s, evaluate ${R}_{0}\left(1\right),{R}_{1}\left(1\right),{R}_{2}\left(1\right),{R}_{3}\left(1\right),{R}_{4}\left(1\right)$.
3. Using the results from part 2, show that for each remainder ${R}_{0}\left(1\right),{R}_{1}\left(1\right),{R}_{2}\left(1\right),{R}_{3}\left(1\right),{R}_{4}\left(1\right)$, we can find an integer $n$ such that $k{R}_{n}\left(1\right)$ is an integer for $n=0,1,2,3,4$.
4. Write down the formula for the nth Maclaurin polynomial ${p}_{n}\left(x\right)$ for $e^{x}$ and the corresponding remainder ${R}_{n}\left(x\right)$. Show that $sn\text{!}{R}_{n}\left(1\right)$ is an integer.
5. Use Taylor’s theorem to write down an explicit formula for ${R}_{n}\left(1\right)$. Conclude that ${R}_{n}\left(1\right)\ne 0$, and therefore, $sn\text{!}{R}_{n}\left(1\right)\ne 0$.
6. Use Taylor’s theorem to find an estimate on ${R}_{n}\left(1\right)$. Use this estimate combined with the result from part 5 to show that $|sn\text{!}{R}_{n}\left(1\right)|<\frac{se}{n+1}$. Conclude that if $n$ is large enough, then $|sn\text{!}{R}_{n}\left(1\right)|<1$. Therefore, $sn\text{!}{R}_{n}\left(1\right)$ is an integer with magnitude less than 1. Thus, $sn\text{!}{R}_{n}\left(1\right)=0$. But from part 5, we know that $sn\text{!}{R}_{n}\left(1\right)\ne 0$. We have arrived at a contradiction, and consequently, the original supposition that e is rational must be false.