普林斯顿微积分读本 VIII

Jun 19, 2021    泰勒级数   幂级数  

§泰勒多项式、泰勒级数与幂级数

24.1 近似值和泰勒多项式

Here’s a nice fact: for any real number $x$, we have $$ e^x \cong 1 + x + {x^2\over2} + {x^3\over6} $$ Also, the closer x is to 0, the better the approximation.

24.1.1 线性化的回顾

平滑的函数可经过多次求导(见13.2),曲线在点 $(a,f(a))$ 处的切线的方程为: $$ y=f(a)+f’(a)(x-a) $$

称为函数 $f$ 在 $x=a$ 处的线性化(linearization)

线性近似,即在切点处附近,切线与曲线差别很小

24.1.2 二次方程近似

二次近似(quadratic approximates) $$ y=f(a)+f’(a)(x-a)+{f’’(a)\over2}(x-a)^2 $$

关于 $x$ 的二次函数

  1. $P_2(x)=y$,代入 $x=a$ 时,$P_2(a)=f(a)$
  2. $P_2’(x)=f’(a)+f’’(a)(x-a)$,代入 $x=a$ 时,$P_2’(a)=f’(a)$
  3. $P_2’’(x)=f’’(a)$,代入 $x=a$ 时,$P_2’’(a)=f’(a)$
  4. 因为 $f’’(a)$ 是一个常数,$P_2’’’(x)=0$

The second-order correction term helps us get even closer to the curve, at least for $x$ near $a$.

在 $P_2(x)=y$ 中,${f’’(a)\over2}(x-a)^2$ 是所谓的二阶修正项(second-order correction term)

24.1.3 高阶近似

泰勒近似定理(A Taylor approximation theorem):

如果 $f$ 在 $x=a$ 处平滑,在所有次数为 $N$ 或更低的多项式中, 当 $x$ 在 $a$ 附近时,最近似于 $f(x)$ 的是 $$ \begin{aligned} P_N(x)&={f^{(0)}(a)\over 0!}(x-a)^0+{f^{(1)}(a)\over 1!}(x-a)+{f^{(2)}(a)\over 2!}(x-a)^2\&+{f^{(3)}(a)\over 3!}(x-a)^3+\dots+{f^{(N)}(a)\over N!}(x-a)^N \end{aligned} $$ 用求和号表示为: $$ P_N(x)=\sum_{n=0}^N{f^{(n)}(a) \over n!}(x-a)^n $$ 称为 $f(x)$ 在 $x=a$ 处的 $N$ 阶泰勒多项式($N$ th-order Taylor polynomial of $f(x)$)

$P_N$ 的一个重要性质:

对所有 $n=0,1,\dots,N$,$P_N$ 在 $x=a$ 处导数与 $f$ 对应,即 $$ P_N^{(n)}(a)=f^{(n)}(a) $$

24.1.4 泰勒定理

How good is this approximation? One way to measure this is to consider the difference between the true quantity and the approximation . We’ll call this quantity the error in the approximation, since it shows how wrong we are to use our approximation instead of the true value.

衡量近似:误差(error) $$ R_N(x)=f(x)-P_N(x) $$ $R_N(x)$ 称为 $N$ 阶误差项($N$ th-order error term)或 $N$ 阶余项($N$ th-order remainder term)

泰勒定理(Taylor’s Theorem:):

关于 $x=a$ 的 $N$ 阶余项 $R_N(x)$ 为 $$ R_N(x)=\frac{f^{(N+1)}(c)}{(N+1)!}(x-a)^{N+1} $$ 其中,$c$ 是介于 $x$ 与 $a$ 之间的数(见11.3,中值定理) $$ f’(c)=\frac{f(b)-f(a)}{b-a} $$ 令 $x=b$,可得 $f(x)=f(a)+f’(c)(x-a)$

(注意:不限制 $x$ 与 $a$ 的大小)

Note that the number $c$ depends on what $x$ and $N$ are, and cannot be determined in general!

$$ f(x)=\sum_{n=0}^N{f^{(n)}(a) \over n!}(x-a)^n+\frac{f^{(N+1)}(c)}{(N+1)!}(x-a)^{N+1} $$

Taylor’s Theorem is basically then Mean Value Theorem(MVT) on steroids.

泰勒定理是中值定理($N=0$)的扩展

24.2 幂级数和泰勒级数

$$ e^x=1+x+\frac{x^2}{2!}+\frac{x^3}{3!}+\frac{x^4}{4!}+\frac{x^5}{5!}+\dots $$

注意:等式右边不是多项式(无最高次项)

In fact, it’s an example of a power series. If you replace $x$ by any particular value, you get a regular old series.

右边的幂级数收敛于 $e^x$ (证明见24.2.3

24.2.1 一般幂级数

关于 $x=0$ 的幂级数的为 $$ a_0+a_1x+a_2x^2+a_3x^3+a_4x^4+\dots $$ 用求和号表示为 $$ \sum_{n=0}^{\infty}a_nx^n $$ 其中 $a_n$ 是确定的常数(在前一节中 $a_n=\frac{1}{n!}$)

当系数 $a_n=1$ 时 $\sum\limits_{n=0}^{\infty}x^n$,幂级数为首项为 $1$,公比为 $x$ 的几何级数

当 $-1<x<1$ $$ {1\over1-x}=1+x+x^2+x^3+x^4+\dots $$ 当 $x=0$ 时,幂级数自动收敛于 $a_0$

将这一关于 $0$ 的性质推广至其他实数 $a$ (幂级数的中心

Let’s transfer this special property over to some other number $a$. All we have to do is replace $x$ by $(x-a)$. So here is the general expression for a power series about $x = a$:

$$ a_0+a_1(x-a)+a_2(x-a)^2+a_3(x-a)^3+a_4(x-a)^4+\dots $$

用求和号表示为 $$ \sum_{n=0}^{\infty}a_n(x-a)^n $$ 当 $x=a$ 时,幂级数自动收敛于 $a_0$

The number $a$ is called the center of the power series.

24.2.2 泰勒级数和麦克劳林级数

假设从一个光滑的函数 $f$ 开始,用其所有导数定义一个在 $x=a$ 处的特定幂级数(系数为 $a_n=f^{(n)}(a)/n!$) $$ \sum_{n=0}^{\infty}{f^{(n)}(a) \over n!}(x-a)^n $$ 称为 $f$ 关于 $x=a$ 的泰勒级数

注意:与泰勒多项式的区别——一直持续到 $\infty$,而没有终止于 $n=N$

麦克劳林级数(Maclaurin Series):$f$ 关于 $x=0$ 的泰勒级数 $$ \sum_{n=0}^{\infty}{f^{(n)}(0) \over n!}(x)^n $$

24.2.3 泰勒级数的收敛性

泰勒级数是一个中心为 $a$ 的幂级数

we let $N$ get larger and larger. This should hopefully make the approximation $P_N(x)$ get closer and closer to the true value $f(x)$. This is the same thing as saying that hopefully the error $R_N(x)$ gets smaller and smaller.

对等式 $f(x)=P_N(x)+R_N(x)$ 取 $N\to\infty$ 的极限 $$ \lim_{N\to\infty}f(x)=\lim_{N\to\infty}P_N(x)+\lim_{N\to\infty}R_N(x)=\lim_{N\to\infty}P_N(x) $$ 因为 $f(x)$ 不依赖于 $N$, $$ f(x)=\lim_{N\to\infty}P_N(x)=\sum_{n=0}^{\infty}{f^{(n)}(a) \over n!}(x-a)^n $$

So $f(x)$ equals its Taylor series! In other words, if you want to prove that a function equals its Taylor series at some number x, try to show that $R_N (x) \to 0$ as $N \to \infty$.

讨论 $f(x)=e^x$ 在 $a=0$ 时的情况(即 $f$ 在 $x=0$ 处平滑,$N$ 阶导数 $f^{(n)}(0)=e^0=1$)的近似(麦克劳林级数), $$ f(x)=P_N(x)=\sum_{n=0}^{\infty}\frac{e^0}{n!}{x^n} $$ 只须证明: $$ \lim_{N\to\infty}R_N(x)=\lim_{N\to\infty}e^c\frac{x^{N+1}}{(N+1)!}=0 $$ 注意:$e^c$ 依赖于 $N$,当 $x$ 为一个固定的常数时,无论 $N$ 为何值,有 $0{\leq}e^c{\leq}C$

故有 $0{\leq}e^c\frac{x^{N+1}}{(N+1)!}{\leq}C\frac{x^{N+1}}{(N+1)!}$,当 $N\to\infty$ 时左右极限为 $0$ $$ e^x=1+x+{x^2\over2!}+{x^3\over3!}+{x^4\over4!}+{x^5\over5!}+\dots $$

可自控的例子

$$ f(x)=\cos(x)=1-{x^2\over2!}+{x^4\over4!}-{x^6\over6!}+\dots $$

在这个例子中,麦克劳林级数没有奇数次项;

其相应的泰勒多项式中,奇数阶的多项式的阶数(order)与次数(degree)不一致,例如 $P_5(x)$ 的次数为 $4$,泰勒多项式 $P_5(x)=P_4(x)$

通过取绝对值的方式($\cos(x)$ 和 $\sin(x)$ 的绝对值不大于 $1$),证明 $\lim\limits_{N\to\infty}R_N(x)=0$ $$ \cos(x)=\sum_{n=0}^{\infty}(-1)^n{x^{2n}\over(2n)!} $$

24.3 一个有用的极限

$$ \lim_{N\to\infty}\frac{x^{N+1}}{(N+1)!}=0 $$

证明略

§如何解决近似问题

25.1 泰勒多项式与泰勒级数的总结

  1. 关于 $x=a$ 的 $N$ 阶泰勒多项式
  2. 多项式 $P_N(x)$ 与 $f$ 在 $x=a$ 点直到 $N$ 阶的导数相同
  3. $N$ 阶余项(误差项)
  4. $f(x)$ 的完整表达式
  5. 关于 $x=a$ 的泰勒级数
  6. 关于 $x=a=0$ 的麦克劳林级数

25.2 求泰勒多项式与泰勒级数

例 1

$f(x)=e^x$ 关于 $x=-2$ 的泰勒级数

例 2

$\sin(x)$ 关于 $x=\pi/6$ 的泰勒级数

例 3

$(1+x)^{1/2}$ 的麦克劳林级数

当 $-1<x<1$ 时,余项趋于 $0$ $$ (1+x)^{1\over2}=1+{x\over2}-{x^2\over8}+{x^3\over16}-{x^4\over128}+\dots $$ 二项定理(binomial theorem)的一个特殊情形 $$ (1+x)^a=1+ax+{a(a-1)\over2!}x^2+{a(a-1)(a-2)\over3!}x^3+\dots $$

for $-1 < x < 1$. The series on the right-hand side diverges when $x > 1$ or $x < -1$ unless a happens to be a nonnegative integer.

25.3 用误差项进行估算

  1. 相关函数 $f$ 的选择
  2. 选择一个 $x$ 附近的数 $a$
  3. 导数表($N+1$ 阶导)
  4. 考虑误差, $c$ 的取值范围
  5. 确定泰勒多项式的阶,$N$ 越大,误差越小,求 $|R_N(x)|$ 的最大值,根据 $c$ 的合适区间对 $f^{N+1}(c)$ 最大化处理
  6. 泰勒多项式进行实际近似 ${f(x)}\cong{P_N(x)}$
  7. 若 $R_N(x)$ 为正,近似结果低估(underestimate);若 $R_N(x)$ 为负,近似结果高估(overestimate)

25.3.1 第一个例子

问题:二阶泰勒多项式估算 $e^{1/3}$

最大化处理,确定估算误差:$e^c<e^{1/3}<8^{1/3}=2$

25.3.2 第二个例子

问题:估算 $e^{1/3}$,使误差小于 $1/10000$

使用四阶泰勒多项式

25.3.3 第三个例子

问题:估算 $\sqrt{27}$ ,使误差不大于 $1/250$

  1. $f(x)=\sqrt{x}$
  2. $x=27,a=25$
  3. $N+1=3$ 导数表
  4. $25<c<27$
  5. $|R_1(27)|={c^{-2/3}\over2}\leq{1\over125}\times{1\over2}={1\over250}$
  6. $P_1(27)$
  7. $R_1(27)={f’’(c)\over2!}(27-25)^2=-{1\over4}\frac{c^{-3/2}}{2!}\times2^2=-\frac{c^{-3/2}}{2}<0$

25.3.4 第四个例子

问题:估算 $\sqrt{23}$ ,使误差不大于 $1/250$

  1. $f(x)=\sqrt{x}$
  2. $x=23,a=25$
  3. $N+1=3$ 导数表
  4. $23<c<25$
  5. $|R_2(23)|={c^{-2/5}\over2}\leq{23^{-5/2}\over2}\leq{16^{-5/2}\over2}={1\over2048}$
  6. $P_2(23)$
  7. $R_2(23)={f’’(c)\over2!}(23-25)^2={3\over8}\cdot\frac{c^{-5/2}}{3!}\cdot(-2)^3=-\frac{c^{-5/2}}{2}<0$

25.3.5 第五个例子

问题:三级泰勒级数估算 $\cos({\pi\over3}-0.01)$

25.3.6 误差项估算的一般方法

根据 $c$ 的合适区间来估算 $f^{(N+1)}(c)$

  1. $|\sin(c)|\leq1$ 和 $|\cos(c)|\leq1$
  2. 根据函数 $f^{(N+1)}(c)$ 的单调性判断
  3. 可能还要求函数 $f^{(N+1)}(c)$ 的临界点

25.4 误差估算近似问题的另一种方法

各项绝对值递减且趋近于 $0$ 交错级数,则误差小于下一项的绝对值 $$ |R_N(x)|\leq|\frac{f^{(N+1)}(a)}{(N+1)!}(x-a)^{N+1}| $$ 注意:函数的输入不是 $c$

麦克劳林级数求解定积分

问题:估算 $\int_0^1\frac{1-\cos(t)}{t^2}\mathrm{d}t$,误差不大于 $1/3000$

由洛必达法则可得 $\lim\limits_{t \to 0}\frac{1-\cos(t)}{t^2}=\lim\limits_{t \to 0}\frac{\sin(t)}{2t}=\frac{1}{2}$

解:构造函数 $$ f(x)=\int_0^x\frac{1-\cos(t)}{t^2}\mathrm{d}t $$ 估算 $f(1)$ $$ \begin{aligned} f(x)&=\int_0^1\frac{1-(1-\frac{t^2}{2!}+\frac{t^4}{4!}-\frac{t^6}{6!}-\cdots)}{t^2}\mathrm{d}t\&=\int_0^1{\frac{1}{2!}+\frac{t^2}{4!}-\frac{t^4}{6!}-\frac{t^6}{8!}\cdots}\mathrm{d}t \end{aligned} $$ 求积分并计算端点处的值 $$ \begin{aligned} f(x)&=(\frac{t}{2!}-\frac{t^3}{3\times4!}+\frac{t^5}{5\times6!}-\frac{t^7}{7\times8!}+\cdots)|_0^x\&=\frac{x}{2!}-\frac{x^3}{3\times4!}+\frac{x^5}{5\times6!}-\frac{x^7}{\times8!}+\cdots \end{aligned} $$ 所以 $f(1)={1\over2!}-{1\over3\times4}+{1\over5\times6}-{1\over7\times8}+\dots$

$f(1){\cong}{1\over2!}-{1\over3\times4}={35\over72}$,那么,$|error|\leq{1\over5\times6!}={1\over3600}$

§泰勒级数、幂级数:如何解题

26.1 幂级数的收敛性

幂级数:$\sum\limits_{n=0}^{\infty}a_n(x-a)^n$

Furthermore, if the series converges for a specific x, it would be nice to know whether the convergence is absolute or merely conditional.

26.1.1 收敛半径

几何级数 $\sum\limits_{n=0}^{\infty}x^n$ 在 $-1<x<1$ 收敛

无穷级数 $\sum\limits_{n=0}^{\infty}\frac{x^n}{n!}$ 对所有 $x$ 均收敛于 $e^x$

幂级数 $\sum\limits_{n=0}^{\infty}a_n(x-a)^n$ 在 $x=a$ 处收敛于 $a_0$,考虑其他可能的情况:

  1. There is some number $R > 0$, called the radius of convergence of the power series,
  2. The power series might converge absolutely for all $x$,
  3. The power series might converge absolutely only for $x = a$ and diverge for all other $x$. In this case, the radius of convergence is $0$.

26.1.2 寻找收敛半径和收敛区域

比式判别法(ratio test) $$ \lim_{n\to\infty}|\frac{a_{n+1}(x-a)^{n+1}}{a_n(x-a)^n}|=\lim_{n\to\infty}|\frac{a_{n+1}}{a_n}||x-a| $$ 或根式判别法(root test) $$ \lim_{n\to\infty}|a_n(x-a)^n|^{1/n}=\lim_{n\to\infty}|a_n|^{1/n}|x-a| $$ 计算极限**(注意并非 $x\to\infty$)**,形如 $L|x-a|$,其中 $L$ 可能为有限值、$0$ 或 $\infty$

根据判别法的,对极限值与 $1$ 的大小进行比较:

假设 $L>0$ ,收敛半径为 $1/L$

假设 $L=0$,极限小于 $1$,故幂级数对所有 $x$ 绝对收敛,收敛半径为 $\infty$

假设 $L=\infty$,仅当 $x=a$ 时,幂级数收敛,即收敛半径为 $0$

例 1

$\sum\limits_{n=2}^{\infty}\frac{x^n}{n\ln(n)}$,应用比式判别法计算极限 $L=|x|$,收敛半径为 $1$,检验端点 $x=-1$ 交错收敛,$x=1$ 发散

例 2

$\sum\limits_{n=2}^{\infty}\frac{x^n}{n(\ln(n))^2}$,应用比式判别法计算极限 $L=|x|$,收敛半径为 $1$,检验端点收敛

例 3

$\sum\limits_{n=2}^{\infty}n!x^n$,应用根式判别法计算极限 $L=\infty,x\neq0$,收敛半径为 $0$

例 4

$\sum\limits_{n=2}^{\infty}\frac{(-2)^n}{\sqrt{n}}(x-7)^n$,应用比式判别法计算极限 $L=2|x-7|$,收敛半径为 $1/2$,检验端点 $x=7{1\over2}$ 收敛,端点 $x=6{1\over2}$ 发散

例 5

$\sum\limits_{n=2}^{\infty}\frac{3^n}{2^{n^2}}(x+2)^n$,应用根式判别法计算极限 $L=0$,幂级数绝对收敛

26.2 寻找新的泰勒级数

常用的麦克劳林级数:

  1. $e^x=\sum\limits_{n=0}^{\infty}{\frac{x^n}{n!}}=1+x+{x^2\over2!}+{x^3\over3!}+\dots$

  2. $\sin(x)=\sum\limits_{n=0}^{\infty}{\frac{(-1)^nx^{2n+1}}{(2n+1)!}}=x-{x^3\over3!}+{x^5\over5!}-{x^7\over7!}+\dots$

    $\cos(x)\sum\limits_{n=0}^{\infty}{\frac{(-1)^nx^{2n}}{(2n)!}}=x-{x^2\over2!}+{x^4\over4!}-{x^6\over6!}+\dots$

  3. ${1\over{1-x}}=\sum\limits_{n=0}^{\infty}{x^n}=1+x+x^2+x^3+\dots,-1<x<1$

  4. $\ln(1\pm{x})$

$$ \begin{aligned} \ln(1+x)=\sum\limits_{n=0}^{\infty}{-\frac{(-1)^n{x^n}}{n}}=x-{x^2\over2}+{x^3\over3}-{x^4\over4}+\dots,-1<x{\leq}1 \ \ln(1-x)=\sum\limits_{n=0}^{\infty}{-\frac{x^n}{n}}=-x-{x^2\over2}-{x^3\over3}-{x^4\over4}-\dots,-1{\leq}x<1 \end{aligned} $$

26.2.1 代换和泰勒级数

将上述的无穷级数 1 中的 $x$ 代换为 $x^2$,求 $f(x)=e^{x^2}$ 的麦克劳林级数 $$ e^{x^2}=1+x^2+{x^4\over2!}+{x^6\over3!}+{x^8\over4!}+\dots $$ $6$ 阶泰勒多项式为 $$ P_6(x)=1+x^2+{x^4\over2!}+{x^6\over3!} $$

This is actually the sixth-order Taylor polynomial of $g$ about $0$, so the left hand side should say $P_6(x)$ instead of $P_3(x)$. To get the correct formula for $P_3(x)$, just drop all the terms of degree greater than $3$. This means that $P_3(x) = 1 + x^2$. Of course, this is also $P_2(x)$ as well! Be careful with your degrees. That’s an order.

关于 $e^{x^2}$ 的泰勒多项式的奇数项为 $0$,故其阶数(order)应该根据 $x$ 的次数(degree)来判断,而非阶乘数

26.2.2 泰勒级数求导

If a power series converges to a differentiable function $f$ on an open interval $(a, b)$, then it turns out that you can differentiate the series term-by-term to get a new series which converges to $f’(x)$ on the same interval.

若幂级数收敛于开区间上的函数 $f$,可通过求导获得一个新的幂级数,且收敛于这一开区间上导函数$f’(x)$

例如,通过对 $\cos(x)$ 的泰勒级数求导,取负数,可以获得 $\sin(x)$ 的泰勒级数

注意:求导后的新级数需要检验端点处是否收敛,若在端点处收敛,则原级数在端点处也收敛

Once again, you can apply these ideas to Taylor polynomials; you just have to be careful with orders, once again. Since differentiating a polynomial knocks the degree down by one, the differentiated Taylor polynomial is order one less than the original polynomial.

判断泰勒多项式的阶数时,注意其次数因为求导而减少了

26.2.3 泰勒级数的积分

You can also integrate a power series term-by-term. The new series converges in the same interval as the old one (except perhaps at the endpoints of the interval of convergence).

注意:不定积分的结果中包含常数 $C$

例如,通过对 ${1}\over{1-x}$ 求积分,取负数,来证明 $\ln(1-x)$ 的泰勒级数($C=0$)

例 1

求 $\arctan(x)$ 的泰勒级数

当 $-1<x<1$,对泰勒级数 ${1\over1+x^2}=1-x^2+x^4-x^6+\dots$ 求不定积分 $$ \begin{aligned} \int{1\over1+x^2}\mathrm{d}x&=\int(1-x^2+x^4-x^6+\dots)\mathrm{d}x \ \implies\tan^{-1}(x)&=C+x-{x^3\over3}+{x^5\over5}-{x^7\over7}+\dots \&=\sum_{n=0}^{\infty}{(-1)^nx^{(2n+1)}\over{2n+1}} \end{aligned} $$ 取 $x=0$,解得 $C=\tan^{-1}(0)=0$

In fact, the series for $\tan^{-1}(x)$ also converges when $x = 1$ (or $x = -1$) by the alternating series test, eventually leading to the cute formula $$ 1-{1\over3}+{1\over5}-{1\over7}+\dots=\tan^{-1}(1)={\pi\over4} $$

例 2

不定积分

求函数 $f(x)=\int_0^x{\sin(t^3)}\mathrm{d}t$ 的麦克劳林级数

$$ \begin{aligned} f(x)&=\int_0^x{\sin(t^3)}\mathrm{d}t=\int_0^x(t^3-{t^9\over3!}+{t^{15}\over5!}-{t^{21}\over7!}=\dots)\mathrm{d}t \ &=({t^4\over4}-{t^{10}\over10\cdot3!}+{t^{16}\over16\cdot5!}-{t^{22}\over22\cdot7!}+\dots)|_0^x \ &={x^4\over4}-{x^{10}\over10\cdot3!}+{x^{16}\over16\cdot5!}-{x^{22}\over22\cdot7!}+\dots \end{aligned} $$

You can also apply the above integration techniques to Taylor polynomials; this time the order of the Taylor polynomial increases by $1$.

26.2.4 泰勒级数的加减

If you know the Taylor series about $x = a$ for two functions $f$ and $g$, then the Taylor series for the sum $f(x) + g(x)$ is of course the sum of the two respective Taylor series, at least in the overlap of the regions where the Taylor series converge. The same goes for the difference $f(x) - g(x)$.

合并同类项

If you want to deal with Taylor polynomials, you have to be careful to take the order to be the lesser of the two orders.

泰勒多项式的加减应该考虑一致的阶数(不同阶数则取低者)

26.2.5 泰勒级数的乘法

You can also multiply two Taylor series to get a new one which converges to the product of the two relevant functions, at least in the intersection of the regions where the Taylor series converge.

注意:略去阶数大于原函数级数的阶的项

例如

$e^x$ 的二阶泰勒多项式 $1+x+x^2$ 与 $e^{-x}$ 的二阶泰勒多项式 $1-x+x^2$ 的乘积,只需要考虑展开项 $1$ 即可。

因为 $x^4\over4$ 的次数 $4$ 比原函数的级数 $2$ 大,显然 ${e^x}\cdot{e^{-x}}=1$

26.2.6 泰勒级数的相除

使用长除法

例如 求 $\sec(x)$ 和 $\tan(x)$ 的泰勒级数

……

26.3 利用幂级数和泰勒级数求导

$f(x)$ 关于 $x=a$ 的泰勒级数的第 $n$ 项系数公式 $a_n=\frac{f^{(n)}(x)}{n!}$,两边乘上 $n!$ 可得 $$ f^{(n)}(a)=a_n\times{n!} $$ 故可获得 $x=a$ 处的导数(对第 $n$ 阶导数只需要知道其 $n$ 阶或更高的泰勒多项式)

例 1

$f(x)=e^{x^2}$,计算 $f^{(100)}(0)$

根据代换,求得麦克劳林级数 $$ e^{x^2}=1+x^2+\frac{x^4}{2!}+\frac{x^6}{3!}+\frac{x^8}{4!}+\dots $$ $f^{(100)}(0)=(b){\times}100!$,$(b)$ 为 $x^{100}$ 的系数,本例中为 $1/(50!)$, $$ f^{(100)}(0)=\frac{100!}{50!} $$

例 2

$f(x)=\int_0^x{\sin(t^3)}\mathrm{d}t$,计算 $f^{(50)}(0)$ 和 $f^{(52)}(0)$ $$ f(x)=\sum_{n=0}^{\infty}\frac{(-1)^nx^{6n+4}}{(6n+4)(2n+1)!} $$ 求解 $6n+4=50$,得 $n=23/3$ 不为整数($\Bbb{Z}$),所以 $x^{50}$ 的系数为 $0$

求解 $6n+4=52$,得 $n=8$,所以计算 $n=8$ 的项为 $x^{52}\over{52}\times{17!}$

$f^{(50)}(0)=0$

$f^{(52)}(0)=52!{\times}{1\over52{\times}17!}={51!\over17!}$

关于 $x=a$ 的幂级数

26.4 利用麦克劳林级数求极限

形如 $\lim\limits_{x\to0}\frac{f(x)}{g(x)}$ 分子分母都为 $0$ 时,则可以用洛必达法则

一个需要多次求导的例子 $$ \lim_{x\to0}\frac{e^{-x^2}+x^2\cos(x)-1}{1-\cos(2x^3)} $$ 可以将 $e^{-x^2}$、$x^2\cos(x)$ 和 $\cos(2x^3)$ 展开到第 $8$ 阶,最后上下同除以最低次项 $x^6$

若 $f$ 有最低次项为 $a_Nx^N$ 的麦克劳林级数,则当 $x\to0$ 时, $f(x){\sim}a_NX^N$

($f$ 的 $N+1$ 阶导在 $0$ 附近有界)

证明: $$ \lim_{x\to0}\frac{e^x-1-\sin(x)}{\sin(x)(e^x-1)}=1/2 $$ $e^x-1-\sin(x)$ 的麦克劳林级数的首项(最低次)为 $x^2/2$,所以当 $x\to0$ 时,$e^x-1-\sin(x){\sim}x^2/2$

由极限比较判别法,当 $x\to0$ 时,$e^x-1{\sim}x$,$\sin(x){\sim}x$

所以, $$ \frac{e^x-1-\sin(x)}{\sin(x)(e^x-1)}\sim\frac{x^2/2}{(x)(x)} $$