Let $P$ be a poset with a least element $\bot$ ($\forall x \in P.\ \bot \le x$). Let $M$ be a set of monotone maps $P \to P$. Call $x \in P$ reachable if $x = f_1(f_2(...f_n(\bot)...))$ for some sequence $f_i \in M$. Call $x$ fixed if $x = f(x)$ for all $f \in M$.

If $x$ reachable, $y$ fixed, then $x \le y$. For then $x = F(\bot)$ for $F$ some composition of functions in $M$. Since $F$ is monotone and $\bot \le y$, we have $x = F(\bot) \le F(y) = y$. Thus there is at most one reachable fixed point.

Question #1: If there is a reachable fixed point $x$, will applying functions in $M$ uniformly at random (starting from $\bot$) eventually find $x$ with probability 1?

This requires a uniform probability distribution over $M$, so is ill-defined if $M$ is infinite. So:

Question #2: If there is a reachable fixed point $x$, will any schedule which eventually includes every sequence of $M$s infinitely often eventually find $x$ (starting from $\bot$)?

For our purposes, $M$ is at most countable. I'd even be interested in a solution to the case where $M$ is finite.


I think yes, to Question #2, which implies Question #1.

Does this work?

Suppose $x$ is reachable and fixed. Pick $F$ with $x = F(\bot)$ to witness $x$'s reachability. Let $s$ be an arbitrary reachable element, with $F' \in M^*$ witnessing its reachability, i.e. $F'(\bot) = s$.

Observe that $F(s)$ is reachable, since $F(s) = F(F'(\bot))$. So $F(s) \le x$ because $x$ is fixed, by your own reasoning above. But $\bot \le s$ so $x = F(\bot) \le F(s)$ by monotonicity. Hence $x = F(s)$.

That is, if we start from $\bot$ and hit it with some functions $F'$ from $M$, $x$ will always remain reachable, even from that point $F'(\bot)$, by exactly the sequence of functions $F$.


Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.