2

I would like to solve $x - \ln(x) > 1$ without studying the function $f(x) = x - \ln(x) - 1$ and computing its derivative.

I wonder if solving for $x$ in $x - \ln(x) = 1$ is possible.

It could be possible to consider the Lambert $W$-function, since we can rewrite the expression such that we have

\begin{align*} \ln(x) - x &= 1 \\ \ln(xe^{-x}) &= 1 \\ xe^{-x} &= e \\ -xe^{-x} &= -e \end{align*}

Yet it seems impossible and non-elementary to have a "nice" analytical solution to this.

emandret
  • 926
  • 2
    The unique solution is $x=1$.... – Greg Martin Jan 16 '21 at 19:03
  • 2
    It is equivalent to $e^x > 1+x$ for $x \ne 0$, for which many proofs are known: https://math.stackexchange.com/q/504663/42969. – Martin R Jan 16 '21 at 19:05
  • 1
    What is your definition of the logarithm? What is a “nice analytical proof” without using derivatives? – Martin R Jan 16 '21 at 19:10
  • $x-ln x=ln e^x-ln x=ln \frac{e^x} x>1= ln e$ , $\rightarrow$, $\frac {e^x}x>e $, which gives: $e^x > e\cdot x$ which is clear for any $x ∈ \mathbb N$ . – sirous Jan 16 '21 at 19:19
  • You said you wanted to solve $x - \ln(x) = 1$ (which I think is a good idea) but when you tried to solve it you immediately wrote $\ln(x)-x = 1$, which has no solution. – David K Jan 16 '21 at 20:33

2 Answers2

3

I think you made a mistake in your solution \begin{align*} x-\ln(x) &>1 \\ \ln\left(\frac{e^x}{x}\right) &> 1 \\ \frac{e^x}{x} &= e \\ e^x &= xe \end{align*} So, it has only one solution which is $x=1.$ I hope that helps

Sebastiano
  • 7,649
00GB
  • 2,401
3

Going on what Martin R said in the comments to the question, using the fact that $\ln(x)$ is defined only for $x>0$ and that $e^x$ is a strictly increasing function,

$x-\ln(x)>1,\ x>0 \implies e^{x-\ln(x)} > e^1,\ x>0 \implies \frac{1}{x}e^x > e,\ x>0 \implies e^x > ex,\ x>0.$

Substituting $x = y+1$ gives:

$e^{y+1} > e(y+1),\ y > -1 \implies e^y> y+1,\ y > -1.$

Various proofs of the final equation for all $y \in \mathbb{R}$ can be found here. In fact, you only need a proof that works for all $y > -1, $ which other people are welcome to do if they so desire (I'm not going to...).

Adam Rubinson
  • 20,052