Wikipedia:Reference desk/Archives/Mathematics/2007 July 18

From Wikipedia, the free encyclopedia

Mathematics desk
< July 17 << Jun | July | Aug >> July 19 >
Welcome to the Wikipedia Mathematics Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.


Contents

[edit] July 18

[edit] Question

Are there any other solutions for x^y = y^x other than y=x? --frotht 19:44, 18 July 2007 (UTC)

Yes. For example 2^4 = 4^2 = 16. Try drawing the graph. It looks like one lobe of an hyperbola (but it isn't). --Trovatore 19:47, 18 July 2007 (UTC)


Here's how you can find more solutions :
x^y=y^x \Leftrightarrow y \ln(x) = x \ln(y) \Leftrightarrow \frac{\ln(y)}{y} = \frac{\ln(x)}{x}
(With x and y both real)
This leads to the study of the function f(x) = ln(x)/x
For each value of a such that the line of equation f(x) = a cuts the curve f(x) = ln(x)/x in two points, there is a solution to the equation x^y = y^x, as this gives two different values of x for which ln(x)/x is the same.
--Xedi 20:35, 18 July 2007 (UTC)
It's worth noting that graphing xy = yx can confuse some numerical methods a bit (the OS X Grapher application makes the curve a bit jaggy as it moves out, but panning/zooming eliminates the weird jagginess). I think, but I haven't proved, that the curved part of the graph intersects the straight line at (e,e). I think one way to show this is to calculate the derivatives of the graph and show that it is undefined at (e,e) but nowhere else on the line y = x. Donald Hosek 22:34, 18 July 2007 (UTC)
Image:Ln(X)divX.png -- SGBailey 09:56, 19 July 2007 (UTC)
If you define the function u on the real numbers except 0 by:
u(z) = \exp\left(\frac{z}{e^z-1}\right)\,,
all real-valued solutions in nonnegative reals of xy = yx with x ≠ y are given parametrically by:
(x, y) = (u(z), u(-z)),~z\ne 0\,.
The solution (2, 4) corresponds to z = 0.69314718... . The point 0 is a removable singularity; just define u(0) = e.
In the rationals, (x, y) = (–2, –4) is also a solution.  --Lambiam 23:49, 18 July 2007 (UTC)
Where does this solution come from ? --Xedi 17:44, 19 July 2007 (UTC)
From fiddling around with the equation xy = yx to get a parametric solution, and then some more fiddling to make it symmetric. I'm sorry, I have to confess it is OR, but readily verified.  --Lambiam 20:03, 19 July 2007 (UTC)
There is a Hungarian language article about this equation: Lóczy Lajos, Mikor kommutatív illetve asszociatív a hatványozás, Középiskolai Matematikai és Fizikai Lapok 2000/1 p. 7–16. However, that issue is not in the online archive, and it has no English abstract nor bibliography, so you can probably only use it if you speak Hungarian. – b_jonas 21:27, 23 July 2007 (UTC)
I think I've found an English translation: http://www.komal.hu/cikkek/loczy/powers/commpower.e.shtml. The article gives a parametric solution that is equivalent to mine, but not as symmetric. On the other hand, it considers the rational solutions and finds a whole family (for example, x = 117649/46656, y = 823543/279936), and also considers solutions of the "associative case", being the equation (xy)z = x(yz).  --Lambiam 03:52, 24 July 2007 (UTC)
Oh, lucky you found it. I didn't look at the webpage because new articles almost never appear there nowdays and didn't think that this old article might be on. – b_jonas 10:44, 24 July 2007 (UTC)