In a pure maths textbook I have, they prove that $e^x$ can be expressed as $1+x+\frac{x^2}{2}+\frac{x^3}{3!}+\frac{x^4}{4!}+\ldots+\frac{x^n}{n!}+\ldots$ However, before they prove this, they say they make two assumptions:

At first we must make 2 assumptions: a) that $e^x$ can be expressed as such a series, and b) That the differential of a series is equal to the differential of all the terms in that series.

After this, they say that the proof for part a) is beyond the scope of the book. I can't find the proof using google, so I've come here. What is the proof?

  • $\begingroup$ the definition of $e^x$ is certainly wrong $\endgroup$ – Alex Jan 7 '15 at 20:11
  • $\begingroup$ @Alex I didn't quote directly from the book, it's just my memory failing me. Will edit $\endgroup$ – imulsion Jan 7 '15 at 20:11
  • $\begingroup$ @Alex Just checked on wikipedia. The definition in the question is correct(?) en.wikipedia.org/wiki/Exponential_function#Formal_definition $\endgroup$ – imulsion Jan 7 '15 at 20:12
  • $\begingroup$ Look up Taylor Expansions... theyre the general case, your question is a special case... $\endgroup$ – Asier Calbet Jan 7 '15 at 20:14
  • 4
    $\begingroup$ It heavily depends on how you defined $e^x$, and thus, in particular, $e$. $\endgroup$ – Thomas Andrews Jan 7 '15 at 20:19

Typically, $e^x$ is defined as that series, and the fact that it converges it sufficient for us to say that the definition makes sense. Showing convergence is trivial - the ratio test establishes that ratio of common terms always goes to $0$, so it must converge. So, the only thing we can really do is to show that the series definition of $e^x$ does follow our intuition of how an exponential function should operate.

In particular, exponential functions satisfy the following identity: $$e^ae^b=e^{a+b}$$ and this is the primary thing which makes them "exponential" - as it implies that $$e^n=\underbrace{e^1\cdot e^1 \cdot \ldots \cdot e^1\cdot e^1}_{n\text{ times}}.$$ If we wish to check that this property holds, we can expand the power series of both sides to get: $$\left(\sum_{i=0}^{\infty}\frac{a^i}{i!}\right)\left(\sum_{i=0}^{\infty}\frac{b^i}{i!}\right)=\sum_{i=0}^{\infty}\frac{(a+b)^i}{i!}.$$ Now, we are allowed to do a lot of manipulations to these sums given that every sum is absolutely convergent. You can prove the assertions about sums I am making here, if you desire a simple exercise in analysis. Firstly, we can distribute the product of two sums into a double sum. In particular, on the left hand, we can write: $$\sum_{i=0}^{\infty}\sum_{j=0}^{\infty}\frac{a^ib^j}{i!j!}=\sum_{i=0}^{\infty}\frac{(a+b)^i}{i!}$$ Which gives us each term $a^ib^j$ separately on the left. So, we then seek to isolate the same term on the right. In particular, notice that if we expand every $(a+b)^i$ by the binomial theorem, we get $$\sum_{i=0}^{\infty}\sum_{j=0}^{\infty}\frac{a^ib^j}{i!j!}=\sum_{i=0}^{\infty}\sum_{k=0}^{i}{i \choose k}\frac{a^kb^{i-k}}{i!}$$ or, expanding ${i \choose k}=\frac{i!}{k!(i-k)!}$ gives: $$\sum_{i=0}^{\infty}\sum_{j=0}^{\infty}\frac{a^ib^j}{i!j!}=\sum_{i=0}^{\infty}\sum_{k=0}^{i}\frac{a^kb^{i-k}}{k!(i-k)!}$$ but, since any sum over the same terms given here is absolutely convergent (you can use a sort of "two dimensional" ratio test to establish this), we can legally rewrite the left side by a change of variables with $j=i-k$ to get: $$\sum_{i=0}^{\infty}\sum_{j=0}^{\infty}\frac{a^ib^j}{i!j!}=\sum_{i=0}^{\infty}\sum_{j=0}^{\infty}\frac{a^ib^{j}}{i! j!}$$ which, of course, holds and both sums converge absolutely.

The function defined by the series, therefore, does satisfy this exponential identity. We can also check that $e^1=e$, though that depends on how we want to define $e$ - but it should be hard in any case. Also $e^x$ is clearly increasing for positive $x$, since all the terms $\frac{x^i}{i!}$ are increasing on that domain. Moreover, we can prove it is increasing everywhere, since if $e^{-n}e^x=e^{x-n}$ is increasing when $x$ is positive (if we take $n$ as a constant), which, for suitably large $n$, proves that $e^{x-n}$ is increasing for all positive $x$ and ergo $e^x$ is positive for all $x$ greater than $-n$. I can't think of anything else you would demand from an exponential function - so clearly, it must be that the series definition of $e^x$ does make sense.


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.