xx3 = 3
Here's a problem. Let f(x) = xx3. Obviously, f(1) = 1 and f grows with x without bound. It follows that the equation
has a solution greater than 1. Find at least one such solution.

Copyright © 1996-2008 Alexander Bogomolny
We are only concerned with positive x. Introduce y = x3, then xy = 3. Take the logarithm of both sides of both equations:
| |
ln(y) = 3·ln(x)
y·ln(x) = ln(3),
|
which simplifies to
The latter equation is trivially solved by y = 3. (We may observe in passing that the equation has no other real solutions. Indeed, the function g(y) = y·ln(y) is negative for y < 1 and, since it's derivative g'(y) = ln(y) + 1 is positive, is increasing for y > 1.)
Therefore, 3 = x3, and x = 31/3.
Well, we got one solution to the equation xx3 = 3 and may consider the problem solved and the case closed. However, I was quite surprised to arrive at that solution. For, note that the value x = 31/3 solves, by definition, another equation, namely x3 = 3. The situations begs for a generalization. x = 31/3 solves
Will it also solve xxx3 = 3? Yes, of course. Since 3 = x3 and xx3 = 3 we can replace in the latter the exponent 3 with x3, which then asserts that indeed xxx3 = 3.
Having made the first step, why not to continue? We can substitute x3 for the exponent 3 in the new equation, to obtain xxxx3 = 3. Let fn(x) = xx...x3 with n occurrences of x, so that f1(x) = x3 and f2(x) = xx3. Then, by an obvious induction, fn(31/3) = 3. On the right, you can see the graphs of the first few functions fn.
(While many graphics program are capable of generating such graphs, the diagram above was produced by DPGraph 2000, a program written and distributed by David Parker. DPGraph 2000 shines in 3D. It generates surfaces with the same ease as curves. All it takes is to write the surface equation. The program is available to download for the nominal price of $1.99. My 6 months old son is absolutely enchanted with those 3D shapes and the speed with which the program manipulates them.)
The next logical step is to replace 3 with an arbitrary parameter a. For positive a, define fn(x, a) = xx...xa with n x's and consider the equation fn(x, a) = a. Had we started with say xxxa = a, would have it been immediately obvious that the equation has a solution x = a1/a?
We see that, for all n, fn(a1/a, a) = a. But that's not all. We may also consider an equation with an "infinite" exponent: xxx... = a. If the expression xxx... does at all have any sense, the equation will again imply xa = a, so that this equation, too, has the same solution a1/a. But how can we define such an infinite exponent? The answer is, in a standard way we should check whether the sequence
| |
s1 = x, s2 = xx, s2 = xxx, ...
|
has a limit. If it does, the infinite exponent xxx... is declared to equal this limit. For 1 < x < e1/e the limit in fact exists, and this is how we prove that.
The sequence {sn} is monotone increasing and is bounded from above. (It's a well known fact that bounded monotone sequences have a limit.) Indeed, since x > 1, s1 > 1. We then also have
and more generally, by induction,
The sequence {sn} is also bounded from above. To see that, first of all observe that if
then x and y are either both greater or both less than 1. If they are greater than 1 then x < y. This starts an induction: s1 < y. Assume sn < y. Then
We conclude that sequence {sn} has a limit, say, A which, by necessity, satisfies
It is important to note that A may be different from y! In fact the function f(z) = z1/z is defined for all positive z, has a maximum at z = e, is monotone increasing for z < e and is monotone decreasing for z > e.

This means that, for 1 < x < e1/e, the equation y1/y = x has two roots: one is less, the other is greater, than e. For example, for x = √2, (1) has two solutions: y = 2 and y = 4. (This is because 41/4 = 21/2.)
The sequence {sn} results if one attempts to solve (1) by an iterative procedure.
| (3) |
s1 = 1, sn+1 = g(sn),
|
where g(s) = xs. The derivative g'(s) = dg(s)/ds is found to be
| |
g'(s) = ln(x)·xs = ln(x)·g(s).
|
If x and s are related via (1), then g'(s) = (ln s)/s·s = ln s, which is greater or less than 1 depending as to whether s is greater or less than e. If |g'(s)| > 1, point s is a repeller for the function g(s), while for |g'(s)| < 1, point s is an attractor.
References
- P. Winkler, Mathematical Puzzles: A Connoisseur's Collection, A K Peters, 2004

Copyright © 1996-2008 Alexander Bogomolny
|