19
$\begingroup$

Suppose you have the function: $$f(x) = e^x \cos{x} $$ and you need to find the 3rd degree Taylor Series representation. The way I have been taught to do this is to express each separate function as a power series and multiply as necessary for the 3rd degree. For example for $$ \cos x =\sum_{n=0}^\infty (-1)^n\frac{ x^{2n}}{(2n)!} = 1-\frac{x^2}{2!}+\frac{x^4}{4!}+\cdots \text{ and } e^x =\sum_{n=0}^\infty \frac{x^n}{n!} = 1 + x + \frac{x^2}{2!} + \frac{x^3}{3!}+\cdots $$ multiply the terms on the right of each until you get the 3rd degree.

Logically, I am happy. However, I have not seen a theorem or any rule that says you can just multiply series in this way. Doing it this way, is there a guarantee that I will always get the power series representation of $f(x)$?

Additionally, if instead of multiplying, functions were being added? Would the above hold true - take the series of each function and add up the necessary terms?

$\endgroup$
  • 3
    $\begingroup$ Hint: The Taylor series converge uniformly and absolutely on any interval $[-a,a]$ where $a<R$. Then, you have tons of theorems you can use. $\endgroup$ – Michael Burr Apr 3 '18 at 16:35
  • 1
    $\begingroup$ The product of two series, one of which is absolutely convergent, converges. Look up Cauchy Product. For example, en.wikipedia.org/wiki/Cauchy_product gets you started. Since the Taylor series are both absolutely convergent... $\endgroup$ – James S. Cook Apr 3 '18 at 23:45
23
$\begingroup$

Taylor's theorem allows you to use the Big O notation: $$\cos(x)= 1-\frac{x^2}{2!}+O(x^4)\quad\mbox{and}\quad e^x=1 + x + \frac{x^2}{2!} + \frac{x^3}{3!}+O(x^4).$$ Therefore $$e^x\cos(x)=1 + x + \frac{x^2}{2!} + \frac{x^3}{3!}+O(x^4)-\frac{x^2}{2!}(1+x+O(x^2))=1+x-\frac{x^3}{3}+O(x^4).$$

$\endgroup$
1
$\begingroup$

Alternatively, you can directly expand the given function $f(x)=e^x\cos x$ to the Taylor series: $$\begin{align}f(0)=&1 \\ f'(x)=e^x\cos x-e^x\sin x \Rightarrow f'(0)=&1;\\ f''(x)=-2e^x\sin x \Rightarrow f''(0)=&0; \\ f'''(x)=-2e^x\sin x-2e^x\cos x \Rightarrow f'''(0)=&-2;\\ f^{(4)}=-4e^x\cos x \Rightarrow f''(0)=&-4 \end{align}$$ Hence: $$\begin{align}e^x\cos x=&1+\frac{1}{1!}x+\frac{0}{2!}x^2+\frac{-2}{3!}x^3+\frac{-4}{4!}x^4+O(x^5)=\\ =&1+x-\frac{x^3}{3}-\frac{x^4}{6}+O(x^5).\end{align}$$

$\endgroup$
1
$\begingroup$

You can use the real part of $e^{(1+i)x}$. As the powers of $1+i$ are $$1,1+i,2i,-2+2i,-4,-4-4i,-8i\cdots$$ you write immediately $$e^x\cos x= 1+x-\frac{x^3}3-\frac{x^4}6-\frac{x^5}{30}+\cdots$$


Using the polar form, the general term is

$$\frac{(1+i)^n}{n!}x^n=\frac{\sqrt2^n\cos\frac{n\pi}{4}}{n!}x^n.$$

$\endgroup$
  • $\begingroup$ Straight version of Marty Cohen's answer. This method is very efficient. $\endgroup$ – Yves Daoust Apr 5 '18 at 18:58
  • 1
    $\begingroup$ And, of course, also gives you $e^x \sin x$. $\endgroup$ – marty cohen Apr 5 '18 at 21:36
0
$\begingroup$

$\begin{array}\\ f(x) &= e^x \cos{x}\\ &= e^x Re(e^{ix})\\ &= Re(e^{x+ix})\\ &= Re(e^{x(1+i)})\\ &= Re(\sum_{n=0}^{\infty} \dfrac{x^n(1+i)^n}{n!})\\ &= Re(\sum_{n=0}^{\infty} \dfrac{2^{n/2}x^ne^{i\pi n/4}}{n!})\\ &= Re(\sum_{n=0}^{\infty} \dfrac{2^{n/2}x^n(\cos(\pi n/4)+i\sin(\pi n/4))}{n!})\\ &= \sum_{n=0}^{\infty} \dfrac{2^{n/2}x^n\cos(\pi n/4)}{n!}\\ &= \sum_{n=0}^{\infty} \dfrac{2^{n/2}x^n[1, 1/\sqrt{2}, 0, -1/\sqrt{2}, -1, -1/\sqrt{2}, 0, 1/\sqrt{2}]}{n!}\\ &= \sum_{n=0}^{\infty} \dfrac{2^{\lfloor n/2 \rfloor}x^n[1, 1, 0, -1, -1, -1, 0, 1]}{n!}\\ &=1+x+0x^2-2x^3/6-4x^4/24-4x^5/120+...\\ &=1+x-x^3/3-x^4/6-x^5/30+...\\ \end{array} $

$\endgroup$
  • $\begingroup$ Sorry, I can't understand your notation. $\endgroup$ – Yves Daoust Apr 5 '18 at 18:46
  • $\begingroup$ $[1, 1, 0, -1, -1, -1, 0, 1]$ means $1*x^0+1*x^1-1*x^3-1*x^4-1*x^5+1*x^7$ with the coefficients repeated for subsequent exponents. $\endgroup$ – marty cohen Apr 5 '18 at 20:19
  • $\begingroup$ I get it now. You take the $n\bmod8^{th}$ component of the tuple. Actually this approach is both faster and more general than the "winning" answer. $\endgroup$ – Yves Daoust Apr 5 '18 at 20:45
0
$\begingroup$

About the product of Taylor series.

In the development of the product

$$f(x)g(x)=\left(\sum_n\frac{f^{(n)}(0)}{n!}x^n\right)\left(\sum_m\frac{g^{(m)}(0)}{m!}x^m\right)=\sum_n\sum_m\frac{f^{(n)}(0)}{n!}\frac{g^{(m)}(0)}{m!}x^{n+m},$$

the terms of degree $n+m=d$ can be grouped as

$$\sum_{n+m=d}\frac{f^{(n)}(0)}{n!}\frac{g^{(m)}(0)}{m!}x^{d} =\sum_{n}\frac{f^{(n)}(0)}{n!}\frac{g^{(d-n)}(0)}{(d-n)!}x^{d} =\sum_{n}\binom dnf^{(n)}(0)g^{(d-n)}(0)\frac{x^{d}}{d!}.$$

Compare this to the Leibniz product rule:

$$(f(x)g(x))^{(d)}=\sum_n\binom dnf^{(n)}(x)g^{(d-n)}(x)$$

and you obtain

$$\left.\left(f(x)g(x)\right)^{(d)}\right|_{x=0}\frac{x^{d}}{d!}.$$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.