# Derivative of $x^x$ at $x=1$ from first principles

Find the derivative of $x^x$ at $x=1$ by definition (i.e. using the limit of the incremental ratio).

The only trick I know is $x^x = e^{x \ln x}$ but it doesn't work.

-
What do you mean "it doen't works"? – Alex Becker Aug 1 '12 at 21:05
And "by definition", using the limit of the incremental ratio? – enzotib Aug 1 '12 at 21:06
In fact, that "trick" does work. – Michael Hardy Aug 1 '12 at 21:07
Yes @enzotib, using th limit of the incremental ratio. – jon jones Aug 1 '12 at 21:14
Oh, I see: "by definition" in the title means you want to find the derivative by using the definition. – Michael Hardy Aug 1 '12 at 21:20

Using the definition: \begin{align} f'(1)&=\lim_{x\rightarrow1}\frac{x^x-1}{x-1}\\ &=\lim_{x\rightarrow1}\frac{e^{x\log{x}}-1}{x-1}\\ &=\lim_{y\rightarrow0}\frac{e^{(1+y)\log(1+y)}-1}{y}\\ &=\lim_{y\rightarrow0}\frac{e^{(1+y)\log(1+y)}-1}{(1+y)\log(1+y)}\frac{(1+y)\log(1+y)}{y}\\ &=\lim_{t\rightarrow0}\frac{e^{t}-1}{t}\lim_{y\rightarrow0}(1+y)\frac{\log(1+y)}{y}=1 \end{align} where $y=x-1$, and $t=(1+y)\log(1+y).$

-
@Neat answer. I like it (+1). – S. Snape Sep 17 '12 at 14:20

The trick you mentioned $\frac{d}{dx}[x^{x}] = \frac{d}{dx} e^{x \ln{x}}$ still works. :)

Apply the chain rule: $e^{x \ln{x}}\frac{d}{dx}[x \ln{x}]$

And then the product rule: $e^{x \ln{x}}(\ln{x}+x\frac{1}{x})$

Simplify: $x^x(1+\ln{x})$

Edit: You wanted the value of the derivative evaluated at $x = 1$, so just substitute in and you get 1.

-
Thanks @user758556 . I know to do in this way. What I wanted to do is find this limit using the limit of the incremental ratio. – jon jones Aug 1 '12 at 21:19
Not sure what you mean by limit of the incremental ratio (I'm really a physicist, so apologies for sloppy nonrigorous math). But just using the definition of derivative: $\displaystyle \lim_{h \rightarrow 0} \frac{(x+h)^{(x+h)} - x^x}{h} = \displaystyle \lim_{h \rightarrow 0} x^x \frac{[(x+h)^h - 1]}{h}$. Since you want this for $x=1$, you have 0/0, so evaluate the limit using L'Hospital's rule at $x = 1$. Apologies if using L'Hospital's rule is assuming too much for whatever you're trying to do? – user758556 Aug 1 '12 at 21:33
@user758556: I guess to use L'Hôpital's you would have to know $\frac{\mathrm{d}}{\mathrm{d}h} (x+h)^h$, which sorts of defeats the purpose of using the limit definition in the first place. – Javier Aug 1 '12 at 22:20