1
$\begingroup$

I seem to have found a method to compute the solution $(x,y)$ to the equation, $x^y=y^x=a$ $a\geq 1$ where $a$ is real, by using limits. But I don't know if this is something new. Does there already exist any method of finding such a solution? If yes, what is it?

$\endgroup$
3
  • 2
    $\begingroup$ Maybe duplicate $x^y=y^x$ $\endgroup$
    – GAVD
    Commented Oct 2, 2015 at 3:00
  • $\begingroup$ Take logarithm of both sides, and contemplate the result. $\endgroup$
    – Lubin
    Commented Oct 2, 2015 at 3:08
  • $\begingroup$ @lubin ya we get x/y=logx/logy ..... how do we proceed after this in this method>? $\endgroup$ Commented Oct 7, 2015 at 18:52

1 Answer 1

0
$\begingroup$

This is the same solution as Michael Lugo has given in the earlier question, as pointed out by GAVD.

From your equation $x_1/x_2=\log x_1/\log x_2$, you get $$ \frac{\log x_1}{x_1}=\frac{\log x_2}{x_2}\,. $$ In other words, assuming that $x_1\ne x_2$, you’ll have two points of the same height on the graph $y=f(x)=(\log x)/x$. This function $f$ has derivative $f'(x)=(1-\log x)/x^2$, zero when $x=e$, nowhere else. So there’ll be a pair $(x_1,x_2)$ solving your equation with one of them greater than $e$, the other less. Since the maximum value of $f$ is $1/e$, all you need to do is choose any number $\alpha<1/e$ and find the two solutions of $f(x)=\alpha$. My primitive way of doing this would be numerical, using Newton-Raphson, I suppose. This method may be less elegant than some of the others given to the prior question, but it does show immediately that $(2,4)$ is the only solution with $x_1$ and $x_2$ integers and $x_1<x_2$.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .