2

When proving $e^x.e^{-x}=1$ by using Taylor series, there are infinite many terms of $e^x$ and $e^{-x}$.

Is there any fancy way to combine terms by terms to show that eventually it is equal to $1$?

miracle173
  • 11,049
jessie
  • 1,107

3 Answers3

8

$$\begin{align} e^x\,e^{-x}&=\sum_{m=0}^\infty\frac{x^m}{m!}\sum_{n=0}^\infty\frac{(-1)^nx^n}{n!}\\\\ &=\sum_{m=0}^\infty \sum_{n=0}^\infty\frac{(-1)^nx^{m+n}}{m!\,n!}\end{align}$$

Now, let $p=m+n$. Then, we can write

$$\begin{align} e^x\,e^{-x}&=\sum_{m=0}^\infty \sum_{p=m}^\infty\frac{(-1)^{p-m}x^{p}}{m!\,(p-m)!}\\\\ &=\sum_{p=0}^\infty \sum_{m=0}^p\frac{(-1)^{p-m}x^{p}}{m!\,(p-m)!}\\\\ &=\sum_{p=0}^\infty (-1)^px^{p}\sum_{m=0}^p\frac{(-1)^{m}}{m!\,(p-m)!}\\\\ &=\sum_{p=0}^\infty \frac{(-1)^px^{p}}{p!}\sum_{m=0}^p\binom{p}{m}\,(-1)^m\\\\ &=1+\sum_{p=1}^\infty \frac{(-1)^px^{p}}{p!}\left(\sum_{m=0}^p\binom{p}{m}\,(-1)^m\right) \tag 1\\\\ \end{align}$$

Noting that the binomial expansion of

$$\begin{align} 0&=(1-1)^p\\\\ &=\sum_{m=0}^p\binom{p}{m}(-1)^m \tag 2 \end{align}$$

Using $(2)$ in $(1)$ reveals $e^x\,e^{-x}=1$. And we are done!

Mark Viola
  • 179,405
  • 1
    Using exactly the same approach we can prove via binomial theorem that $e^{x}e^{y} = e^{x + y}$. +1 – Paramanand Singh Dec 03 '15 at 06:23
  • @paramanandsingh Hello my friend. And yes, I read your excellent posts on your site where you used various approaches to develop properties of the log and exponential functions. +1 for your comment. – Mark Viola Dec 03 '15 at 15:02
3

Set $e^x=\sum_{n\ge 0}a_nx^n$, where $a_n=\frac{1}{n!}$. Similarly, set $e^{-x}=\sum_{n\ge 0} b_n x^n$, where $b_n=\frac{(-1)^n}{n!}$.

By the usual formula for multiplying power series, we have $$e^x e^{-x}=\sum_{i\ge 0}c_i x^i$$ where $$c_i=\sum_{j=0}^i a_jb_{i-j}=\sum_{j=0}^i \frac{1}{j!}\frac{(-1)^{i-j}}{(i-j)!}=\frac{(-1)^i}{i!}\sum_{j=0}^i(-1)^j{i\choose j}=\frac{(-1)^i}{i!}(1-1)^i=\begin{cases}0&i\ge 1\\1 & i=0\end{cases}$$ Note that $(1-1)^i=\sum_{j=0}^i(-1)^j{i\choose j}$ by the binomial theorem. In the special case $i=0$, there is just one summand, and that is $1$.

vadim123
  • 82,796
2

EDIT: As per the comments discussion my answer is perhaps not in the spirit of the exercise and my argument can be made cleaner, but I'll keep it as is.

Here is a method which avoids the headache of formulas for multiplying infinte sums.

You could naively compute the Taylor series by computing the derivatives $$ \frac{d^k}{dx^k} e^{x} e^{-x} $$ so for $k=1$ you'd get $$ e^x e^{-x} - e^x e^{-x} =0 $$ for all $x$ and so all the derivatives vanish. Thus the taylor series for the function has only a constant term and thus $e^x e^{-x}$ is constant. If you were then to multiply the Taylor series for each product you'd quickly see that the only way for their product to have a constant term, i.e. a term with no $x^n$ is if you multiplied the constant terms of the two taylor series. Given that both taylor series have constant term $1$ you get the desired result.

Yeldarbskich
  • 1,394
  • This is nice, but it presupposes knowing the derivatives of $e^x$ and $e^{-x}$. If we are trying to prove their product is $1$, we are likely just developing basic properties and haven't determined derivatives yet. – vadim123 Dec 03 '15 at 05:06
  • @vadim123 I suppose that's fair. In the context of an intro calculus course though I'd say this is okay. Maybe we could start from the assumption that $e^x$ is just notation for a solution to the differential equation $\frac{df}{dx} = f$ and then want to determine the property above? That way we are starting with knowledge of derivatives. – Yeldarbskich Dec 03 '15 at 05:11
  • 1
    @vadim123 You can determine the derivative of $e^x$ via the Taylor series . The derivative of $e^{-x}$ follows by the chain rule. – Thomas Andrews Dec 03 '15 at 05:13
  • I think derivatives are fair game. – copper.hat Dec 03 '15 at 05:15
  • @ThomasAndrews, I'm not sure that was what he meant. I think he was intending to say that we only accept the Maclaurin representation of $e^x$ to start with. – Yeldarbskich Dec 03 '15 at 05:15
  • 1
    Right, but the Maclaurin representation of $e^x$ lets us figure out the derivative of $e^x$ is $e^x$. @Yeldarbskich – Thomas Andrews Dec 03 '15 at 05:17
  • @Yeldarbskich BTW, if $e^xe^{-x}$ is constant, that constant is $e^0e^{-0}=1$. You don't need any further arguments. – Thomas Andrews Dec 03 '15 at 05:20
  • Sure, so we are replacing the headache of series multiplication with the headache of finding derivatives of series. – vadim123 Dec 03 '15 at 05:22