You should be able to solve this. What's the end result?

if y is the above, then y=sqrt(xy), and hence y=x

You got it right.
If you look closely, the same expression can be rewritten as pic related.
Notice that the exponents form a geometric progression, and their sum will eventually reach 1.

I think it's lim a->inf x^(1/2^a) = 1

Call the whole equation y. y=sqrt(xsqrt(xsqrt(x.....)
Since y is the infinite equation you can rewrite it as y=(xy)^1/2. Solve for x.

If Y is the value of that term, then you have Y = sqrt(xY). Thus Y^2 = xY. Which has two solutions: either Y = 0, or Y = x.

this is true.

between x=0 and x=1 it's too small.

x = 0.60, y = 1.6e-13
x = 0.93, y = 0.0152
x = 1.00, y = 1.00

when x > 1, y becomes very fast infinity.

x = 1.06 , y = 28
x = 1.25, y = 387542
x = 1.28, y = 1500000
x = 2.00, y = 2.2e+17

shut up

the correct root is x, it's going to converge to it. see

Going further, you find the infinite series (excuse my phoneposting)

sum((1/(2^n)),1,inf) for the exponents

Which can be represented as a gemetric series with starting point 1/2 and ratio 1/2. Using the geometric series formula

((first term - first missing term)/(1 - ratio))

you see that the exponents converge to one

>the correct root is x
ONE of the correct solutions is x. If you start at zero, it will stay there, which makes 0 another solution. This is a fixpoint problem, there can be more than one solution.

the only correct solution is x. if x = 0, this means that x, that is, 0, is the only correct solution.