MathOverflow will be down for maintenance for approximately 3 hours, starting Monday evening (06/24/2013) at approximately 9:00 PM Eastern time (UTC-4).
3

1

The Kumaraswamy distribution has cdf $F(x;a,b) = 1-(1-x^a)^b$. Does anyone know any formulas or properties relating to iterations of this on itself, meaning $$ F_i(x;a,b) = 1-(1-F_{i-1}^a)^b$$

If something similar exists for the Beta distribution, I would be interested in that as well.

flag
Could you add some information about what you want specifically about the iterations? I've run into these before and could answer you better if I knew what you were looking for. – BSteinhurst Oct 11 2011 at 18:22
@BSteinhurst: any information at all would be great. But specifically, I am interested in calculating/comparing the mean and variance of the distributions at successive iterations. Ideally, I would love to be able to derive $a(i,a,b) = \mathbb{E}X_i$, $X_i\sim F_i$ but more realistically looking for formal proofs that the mean/variance of $F$ decreases/increases with $i$ for constraints on $a$ and $b$. – Jennifer Oct 11 2011 at 19:29
note: I probably shouldn't have used $a(i,a,b)$ as $a()$ is unrelated to $a$. Pretend I wrote $g(i,a,b)$ or some other unrelated letter. – Jennifer Oct 11 2011 at 19:33
1 
Ahh, I think then what I did may be tangential at best. A student and I came across not these distributions but the equations $x = 1-(1-x^{a})^{b}$ in a percolation problem. So we have some experience with the fixed points of the iteration on numbers. That all can be nicely done with a few pages of calculus. – BSteinhurst Oct 11 2011 at 21:41
1 
Only for the dependence of the stable point upon a and b, not the location of the stable point itself. I've not added an answer to this question because what I know doesn't really answer the question you asked. If you are interested in this side line feel free to contact me by e-mail though. – BSteinhurst Oct 13 2011 at 3:34
show 1 more comment

Your Answer

Get an OpenID
or

Browse other questions tagged or ask your own question.