Branch iterated Galois groups with positive fixed-point proportion and positive Hausdorff dimension Thanks: The author is supported by Grigorchuk’s Simons Foundation Grant MP-TSM-00002045 and the department of Mathematics of Texas A&M University. Santiago Radi Abstract . In this article we prove that the arithmetic profinite iterated monodromy group of a post-critically infinite unicritical polynomial is regular branch (and so of positive Hausdorff dimension), and has positive fixed-point proportion when the degree is odd. The examples are instances of a bigger family of regular branch groups constructed in this article, whose fixed-point proportion can be computed explicitly and is positive in many cases. This gives the first examples outside the binary rooted tree where a level-transitive group has positive Hausdorff dimension and positive fixed-point proportion, answering in the negative a question of Jones (2008). 2020 Mathematics Subject Classification Primary: 20E08, 11R32; Secondary: 60G42, 37F10 1. Introduction Given a d -regular rooted tree T , the group of automorphisms of T , denoted \Aut(T) , is the set of bijective functions g:T\rightarrow T that preserve adjacency between the vertices. Since elements in \Aut(T) preserve adjacency, the group \Aut(T) acts on \mathcal{L}_{n} , the n -th level of the tree for each n\geq 0 . The kernel of each of these actions is denoted \St(n) and then one has natural projections \pi_{n}:\Aut(T)\rightarrow\Aut(T)/\St(n) , which corresponds to the action of the elements in \Aut(T) over the first n levels. If G is a subgroup of \Aut(T) , then \St_{G}(n):=G\cap\St(n) is the stabilizer of level n of G and we have the restrictions of the maps \pi_{n}:G\rightarrow G/\St_{G}(n) . The fixed-point proportion of G is calculated as \FPP(G)=\lim_{n\rightarrow\infty}\frac{\#{\left\{g\in\pi_{n}(G):\text{$g$ fixes a vertex in $\mathcal{L}_{n}$}\right\}}}{\#\pi_{n}(G)}. Odoni in Odoni1985 was the first one to realize that this quantity had applications to problems in arithmetic dynamics. Using the fixed-point proportion, Odoni proved that the Dirichlet density of primes dividing at least one term in the Sylvester sequence is zero, where the Sylvester sequence is given by the initial term w_{1}=2 and the recursion w_{n}=1+w_{1}\cdots w_{n-1} . Odoni’s ideas were later generalized to number fields and rational functions. Given K a number field and f a rational function with degree at least 2 and coefficients in K , the extension K(\cup_{n\geq 0}f^{-n}(0))/K is Galois and its Galois group acts faithfully on the d -regular rooted tree of preimages of 0 (see Jones2014 ). Consequently, it embeds in \Aut(T) as a closed subgroup. This group is denoted G_{\infty}(K,f,0) and called the iterated Galois group. Then, if a_{0} is an element of K , we have that the Dirichlet density of the set P_{f}(a_{0}):={\left\{\mathfrak{p}\in M_{K}^{0}:v_{\mathfrak{p}}(f^{n}(a_{0}))>0\text{ for some $n$ such that $f^{n}(a_{0})\neq 0,\infty$}\right\}}, is bounded above by \FPP(G_{\infty}(K,f,0)) (see ( JonesManes2012 , Theorem 6.1) ). Here, M_{K}^{0} denotes the set of prime ideals of the ring of integers of K . Later, more applications of the fixed-point proportion of iterated Galois groups to problems in arithmetic dynamics were discovered (see for example BridyJones2022 ; Juul2014 ). Replacing K by any field and 0 by an element t transcendental over K , the extension K(\cup_{n\geq 0}f^{-n}(t))/K(t) is still Galois and one denotes the Galois group of the extension as G_{\infty}(K,f,t) . This group also acts faithfully in the tree of preimages of t , it embeds as a closed subgroup in \Aut(T) and it represents the generic iterated Galois group. In a loose sense, G_{\infty}(K,f,t) is the Galois group one expects when we specialize t to a value in K . The group G_{\infty}(K,f,t) is level-transitive, namely, the action of G_{\infty}(K,f,t) over \mathcal{L}_{n} is transitive for every n\in\mathbb{N} . In the case studied by Odoni, the iterated Galois group was \Aut(T) and it is known that the fixed-point proportion of \Aut(T) is zero (see for example ( AbertVirag2004 , Corollary 2.6) ). Even more, if the iterated Galois group has finite index with respect to \Aut(T) , the fixed-point proportion is also zero (see 2.10 .(1)). However, it is not completely clear yet in which cases the iterated Galois group has finite index and many cases where the index is not finite are known (see ( Jones2014 , Section 3) for a discussion about this and ( Jones2014 , Conjecture 3.11) for a conjecture in the quadratic case). Although there are examples with infinite index, there is a tendency for the iterated Galois groups to have positive Hausdorff dimension. Given G\leq\Aut(T) , the Hausdorff dimension of G is defined as \mathcal{H}(G):=\liminf_{n\rightarrow+\infty}\frac{\log(\left|{\pi_{n}(G)}\right|)}{\log(\left|{\pi_{n}(\Aut(T))}\right|)}. The Hausdorff dimension is another way to measure how large a group inside \Aut(T) is. In the context of iterated Galois groups, Pink proved in ( Pink2013 , Theorem 4.4.2) that if f is a quadratic rational function with critical points c_{1} and c_{2} whose post-critical orbits are infinite, and r is the smallest natural number such that f^{r+1}(c_{1})=f^{r+1}(c_{2}) , then \mathcal{H}(G_{\infty}(K,f,t))=1-\frac{1}{2^{r}} , where K is any field of characteristic different to 2 and t is transcendental over K . In case of degree greater than 2 , authors proved in ( Benedetto2017 , Theorem 1.1) that the iterated Galois group of the polynomial 3x^{2}-2x^{3} has Hausdorff dimension \approx 0.871 . However, both cases are known to have null fixed-point proportion. On the other hand, Jones proved in Jones2012 that if T_{d} corresponds to the Chebyshev polynomial of degree d and t is transcendental over \mathbb{C} , then \displaystyle\FPP(G_{\infty}(\mathbb{C},\pm T_{d},t))=\left\{\begin{matrix}1/2&\mbox{if $d$ is odd,}\\ 1/4&\mbox{if $d$ is even.}\end{matrix}\right. However, the Hausdorff dimension of G_{\infty}(\mathbb{C},\pm T_{d},t) is zero (the group is isomorphic to the infinite dihedral group and therefore satisfies the group law [x^{2},y^{2}]=1 . Then the Hausdorff dimension is zero by ( Fari~na2025 , Corollary 5) ). This led Jones to pose the following conjecture: Question 1.1 ( ( Jones2008survey , Conjecture 17) ) . If G is a level-transitive group in \Aut(T) with \mathcal{H}(G)>0 , then \FPP(G)=0 . 1.1 was addressed by Boston in Boston2010 , who constructed a group with Hausdorff dimension 1/2 and fixed-point proportion of at least 1/4 acting on the binary rooted tree. Nevertheless, the exact value of the fixed-point proportion is not provided nor is it proved that the group correspond to an iterated Galois group. For larger trees, there are no examples constructed. In light of Boston’s counterexample, one can modify 1.1 by asking if the result holds when we restrict to iterated Galois groups. Even more, one can restrict further to the case where the iterated Galois groups are branch. A group G\leq\Aut(T) is branch if it is level-transitive and for every level n , the elements of G that act non-trivially below only one subtree at level n generate a subgroup of finite index. From the point of view of number theory, branchness is related to a rapid growth of the field extensions of any preimage of t when we take its iterations by f . An important subfamily of closed branch groups are regular branch groups, which can be given in terms of a finite set of allowed actions (called set of patterns) and have recently been studied in RadiFiniteType2025 . By ( Bartholdi2006 , Proposition 2.7) , closed regular branch groups have positive Hausdorff dimension and its dimension is computable in terms of its set of patterns. The existence of regular branch iterated Galois groups is known. For example, combining the results of Pink (see ( Self_similar_groups , Theorem 6.4.2) ) and the authors in ( GrigSavchukSunic2007 , Theorem 2.5) , one has that G_{\infty}(\mathbb{C},z^{2}+i,t) is regular branch. However, combining the results of the author in ( RadiFiniteType2025 , Section 8.3) and ( Fari~naRadi2025 , Theorem 1) , one obtains that the fixed-point proportion of G_{\infty}(\mathbb{C},z^{2}+i,t) is zero. Question 1.2 . Can we find regular branch (and therefore with positive Hausdorff dimension) iterated Galois groups with positive fixed-point proportion? In this article, we answer 1.2 in the positive by providing examples of regular branch iterated Galois groups whose fixed-point proportion and Hausdorff dimension are calculated explicitly. The main result is as follows: Theorem 1 . Let K be a field and K^{sep} a separable closure of K . Let t be a transcendental element over K , d\geq 2 , c\in K and \xi\in K^{sep} a primitive d th root of unity. Suppose that f(x)=x^{d}+c\in K[x] is a post-critically infinite unicritical polynomial. Then, the group G_{\infty}(K,x^{d}+c,t) is regular branch, its Hausdorff dimension is \mathcal{H}(G_{\infty}(K,x^{d}+c,t))=\frac{\log(d)}{\log(d!)} and its fixed-point proportion is \FPP(G_{\infty}(K,x^{d}+c,t))=\frac{\#{\left\{a\in I:a-1\in(\mathbb{Z}/d\mathbb{Z})^{\times}\right\}}}{\Phi(d)}, where I is the image of the map \Gal(K(\xi)/K)\rightarrow(\mathbb{Z}/d\mathbb{Z})^{\times} given by \sigma\mapsto a with \sigma(\xi)=\xi^{a} , and \Phi is the Euler’s totient function. In the particular case where K=\mathbb{Q} , we have \FPP(G_{\infty}(\mathbb{Q},x^{d}+c,t))=\prod_{p\mid d}\frac{p-2}{p-1}, where the product runs over the prime divisors of d . So, if d is odd, the fixed-point proportion is positive. The iterated Galois group of x^{d}+c has recently being studied by Adams and Hyde in AdamsHyde2025 . 1 allow us to have information about its fixed-point proportion. Using ( Juul2018 , Proposition 3.3) , 1 improves the result of Juul in ( Juul2018 , Theorem 1.4(c)) , since we obtain a sharper bound for the limit inferior of the proportion of periodic points in the reduction by prime ideals of the map x^{d}+c . The proof of 1 is based on the construction of a bigger family of regular branch groups whose fixed-point proportion can be calculated explicitly. Chosen d\geq 2 and two subgroups \mathcal{Q},\mathcal{P}\leq\Sym(d) such that \mathcal{Q}\lhd\mathcal{P} , we define in the d -regular rooted tree T , the subgroup G_{\mathcal{Q}}^{\mathcal{P}}\leq\Aut(T) consisting of all the elements in \Aut(T) whose action over the children of any vertex is in the same coset of \mathcal{P}/\mathcal{Q} . Using another instance of this family, we provide more counterexamples to 1.1 in the case of d -regular rooted trees with d\neq 2\pmod{4} : Theorem 2 . Let d\geq 2 and write d=2^{n}r with r odd. Take \mathcal{Q}=C_{2}^{n}\times C_{r} and \mathcal{P}=\mathcal{Q}\rtimes\Aut(\mathcal{Q}) . Then \mathcal{Q} and \mathcal{P} can be embedded in \Sym(d) such that \mathcal{Q} acts transitively on {\left\{1,\dots,d\right\}} and consequently G_{\mathcal{Q}}^{\mathcal{P}} is regular branch, its Hausdorff dimension equals \mathcal{H}(G_{\mathcal{Q}}^{\mathcal{P}})=\frac{\log(d)}{\log(d!)} and the fixed-point proportion is \FPP(G_{\mathcal{Q}}^{\mathcal{P}})=\frac{\#{\left\{A\in\GL_{n}(\mathbb{F}_{2}):A-1\in\GL_{n}(\mathbb{F}_{2})\right\}}}{\left|{\GL_{n}(\mathbb{F}_{2})}\right|}\,\prod_{p\in\mathbb{P},p\mid r}\left(\frac{p-2}{p-1}\right). When d\neq 2\pmod{4} , the fixed-point proportion is positive. The key fact that allows one to calculate explicitly the fixed-point proportion of the groups G_{\mathcal{Q}}^{\mathcal{P}} comes from a generalization of iterated wreath products that we do in this article. If S is a subset of \Sym(d) , the iterated wreath product W_{S} is the subset of elements in \Aut(T) whose action on the children of any vertex of T is an element in S . In the case that S=H is a subgroup of \Sym(d) , then W_{H} is a closed subgroup of \Aut(T) . In ( AbertVirag2004 , Corollary 2.6) , authors proved that when H is transitive over {\left\{1,\dots,d\right\}} , then \FPP(W_{H})=0 . Extending the notion of fixed-point proportion to sets of \Aut(T) (see Equation 6 for a definition), we obtain an explicit way to calculate the fixed-point proportion of any iterated wreath product W_{S} : Theorem 3 . Let d\geq 2 and S any set in \Sym(d) . Given k\in{\left\{0,\dots,d\right\}} , let D_{S}(k) be the number of permutations in S that fix exactly k points of {\left\{1,\dots,d\right\}} and define the polynomial f_{S}(x):=\sum_{k=1}^{d}\frac{D_{S}(k)}{\#S}(1-(1-x)^{k}) . Then (1) \FPP(W_{S}) equals the largest fixed point of f_{S} in [0,1] . (2) \FPP(W_{S}) is the solution of a polynomial of degree at most d-1 with coefficients in the ring \mathbb{Z}\left[\frac{1}{\#S}\right] . In particular, it is an algebraic integer. (3) \FPP(W_{S})=0 if and only if f_{S}^{\prime}(0)\leq 1 and f_{S} is not the identity function. (4) \FPP(W_{S})=1 if and only if every element in S fixes at least one element in {\left\{1,\dots,d\right\}} . In the particular case where S=H is a subgroup of \Sym(d) , 3 extends the result in ( AbertVirag2004 , Corollary 2.6) for any subgroup of \Sym(d) , either transitive or not: Corollary 4 . If d\geq 2 , H is a subgroup of \Sym(d) and X={\left\{1,\dots,d\right\}} . Then, the fixed point proportion of W_{H} is: \displaystyle\FPP(W_{H})=\left\{\begin{matrix}0&\mbox{if $\mathcal{P}$ is transitive over $X$,}\\ \alpha\in(0,1)&\mbox{if $\mathcal{P}$ not transitive and $\exists\,\sigma$ with no fixed points}\\ 1&\mbox{if $\mathcal{P}$ not transitive and every element fixes a point,}\end{matrix}\right. where \alpha is an algebraic integer. Organization This article is organized into seven sections. Section 2 introduces the background necessary to follow this article. Section 3 focuses on proving 3 and 4 . In Section 4 , we define the groups G_{\mathcal{Q}}^{\mathcal{P}} and characterize when they are level-transitive, calculate their Hausdorff dimensions and their fixed-point proportions. Section 5 includes two examples, with the second one resulting in 2 . In Section 6 , it is proved that the first example in Section 5 corresponds to the iterated Galois group of x^{d}+c , proving 1 . Finally, in Section 7 , the author presents open questions considered relevant to the theory of fixed-point proportions. 2. Preliminaries 2.1. About general notation Given a finite set S , the notation \#S will refer to its number of elements. In the case where S is a subgroup, we will use \left|{S}\right| instead. If H is a subgroup of G , we will use H\leq G and if H has finite index in G , we will denote it by H\leq_{f}G . The subgroup G^{\prime} will denote the commutator subgroup of G and N_{G}(H) the normalizer of H in G . 2.2. Groups acting on rooted trees A d -regular rooted tree T is a tree with infinitely many vertices and a root \emptyset , where all the vertices have the same number of descendants d\geq 2 . The set of vertices at a distance exactly n\geq 0 from the root form the n -th level of T and it will be denoted \mathcal{L}_{n} . The vertices whose distance is at most n from the root form the n -th truncated tree of T , denoted T^{n} . The group of automorphisms of T , denoted \Aut(T) , is the group of bijective functions from T to T that send the root to the root and preserve adjacency between the vertices. This in particular implies that \Aut(T)\curvearrowright\mathcal{L}_{n} for all n\in\mathbb{N} . For any vertex v\in T , the subtree rooted at v , which is again a d -regular rooted tree, is denoted T_{v} . In this article, the action of \Aut(T) on T will be on the left, so if g,h\in\Aut(T) and v\in T , then (gh)(v)=g(h(v)) . Given a vertex v\in T we write \st(v) for the stabilizer of the vertex v , namely, the subgroup of the elements g\in\Aut(T) such that g(v)=v . Given n\in\mathbb{N} , we write \St(n)=\bigcap_{v\in\mathcal{L}_{n}}\st(v) , and we call it the stabilizer of level n . Then \St(n) is a normal subgroup of finite index and \Aut(T)/\St(n) is isomorphic to \Aut(T^{n}) . We can make \Aut(T) a topological group by declaring \St(n) to be a base of neighborhoods of the identity. Then \Aut(T) is homeomorphic and isomorphic to \varprojlim\Aut(T)/\St(n) , making \Aut(T) a profinite group. The transition maps will be \pi_{n}:\Aut(T)\rightarrow\Aut(T^{n}) corresponding to the restricting of the action of an element on the whole tree to only the first n levels. Let v\in T and g\in\Aut(T) . By preservation of adjacency, we have g(T_{v})=T_{g(v)} . Calling g|_{v} the restriction of the map g to T_{v} , for all w\in T_{v} we have g(vw)=g(v)g|_{v}(w). This map is called the section of g at vertex v . To ease notation, we will write g|_{v}^{n} for \pi_{n}(g|_{v}) . A way to describe an element g in \Aut(T) is by decorating each vertex v on the tree with the permutation g|_{v}^{1} . Notice that g|_{v}^{1} is a permutation belonging to \Sym(d) . We will call it label and the collection of all the labels of an element g completely determines g and is called the portrait of g . The sections satisfy the following two properties: (1) \displaystyle(gh)|_{v}^{n}=g|_{h(v)}^{n}h|_{v}^{n}\text{ and }(g^{-1})|_{v}^{n}=(g|_{g^{-1}(v)}^{n})^{-1}. Let us fix now a subgroup G\leq\Aut(T) . Then G also has actions on T and \mathcal{L}_{n} . We define vertex stabilizers and level stabilizers by restricting the previous ones to the group, i.e, \st_{G}(v):=\st(v)\cap G and \St_{G}(n):=\St(n)\cap G for v\in T and n\in\mathbb{N} . Definition 2.1 . We say that a group G\leq\Aut(T) is level-transitive if the actions G\curvearrowright\mathcal{L}_{n} are transitive for all n\in\mathbb{N} . Definition 2.2 . Consider H\leq G\leq\Aut(T) . The relative Hausdorff dimension of H on G is defined as the number \mathcal{H}_{G}(H):=\liminf_{n\rightarrow+\infty}\frac{\log(\left|{\pi_{n}(H)}\right|)}{\log(\left|{\pi_{n}(G)}\right|)}. Lemma 2.3 . Let T be a d -regular rooted tree and K\leq_{f}H\leq G\leq\Aut(T) . Then (1) \mathcal{H}_{G}(H)=\mathcal{H}_{G}(\overline{H}) , (2) if G has infinitely many elements, then \mathcal{H}_{G}(H)=\mathcal{H}_{G}(K) . Proof. (1) follows from the fact that \pi_{n}(H)=\pi_{n}(\overline{H}) for all n\in\mathbb{N} and (2) because for n big enough we have \left|{\pi_{n}(H)}\right|=[H:K]\left|{\pi_{n}(K)}\right| . ∎ Definition 2.4 . If G\leq\Aut(T) , the Hausdorff dimension of G is defined as \mathcal{H}(G):=\mathcal{H}_{\Aut(T)}(G). 2.3. Groups of finite type We say that a group G\leq\Aut(T) is self-similar if g|_{v}\in G for all v\in T and g\in G . If K is a subgroup of \Aut(T) and n\in\mathbb{N} , we define the geometric product of K on level n as K_{n}:={\left\{g\in\St(n):g|_{v}\in K\text{ for all $v\in\mathcal{L}_{n}$}\right\}}. We say that G\leq\Aut(T) is regular branch over a subgroup K if G is level-transitive and K_{1}\leq_{f}K\leq_{f}G . Lemma 2.5 ( ( Sunic2006 , Lemma 10) ) . Let T be a d -regular rooted tree and G\leq\Aut(T) be a self-similar regular branch group, branching over a subgroup K containing \St_{G}(m) . Then, for all n\geq m , \St_{G}(n)=(\St_{G}(m))_{n-m}, where the right-hand side corresponds to the geometric product of \St_{G}(m) . Let D be a positive natural number, T a d -regular rooted tree and \mathcal{P} a subgroup of \Aut(T^{D}) . The group of finite type of depth D and set of patterns \mathcal{P} is defined as (2) G_{\mathcal{P}}:={\left\{g\in\Aut(T):g|_{v}^{D}\in\mathcal{P}\text{ for all $v\in T$}\right\}}. Thus, the elements of G_{\mathcal{P}} must act at every vertex according to a finite family of allowed actions on the D -truncated tree. The smallest D necessary to define the group G_{\mathcal{P}} is called the depth . Proposition 2.6 ( ( Bondarenko2014 , page 3) and ( Grigorchuk2005 , Proposition 7.5) ) . If G is a group of finite type of depth D then G is closed in \Aut(T) , self-similar and regular branch over \St_{G}(D-1) . The following result characterizes groups of finite type and their importance in the theory of groups acting on trees: Theorem 2.7 ( ( Sunic2010 , Theorem 3) ) . Groups of finite type are the closure of regular branch groups. Furthermore, if G is a regular branch group branching over a subgroup K containing \St_{G}(D-1) , then \overline{G} is a group of finite type of depth D . If for example D=1 , we are forcing the labels of each element in G_{\mathcal{P}} to be in a certain subgroup \mathcal{P} of \Sym(d) . In this case the group is isomorphic to the inverse limit of the wreath products \mathcal{P}\wr\mathcal{P}\wr\dots\wr\mathcal{P} . We will denote these group by W_{\mathcal{P}} and call it the iterated wreath product . In the case of iterated wreath products, it is straightforward to calculate its Hausdorff dimension: Lemma 2.8 . Let T be a d -regular tree and H\leq\Sym(d) . Then \mathcal{H}(W_{H})=\frac{\log(\left|{H}\right|)}{\log(d!)} Proof. Since at any vertex of the truncated tree T^{n-1} we are allowed to put as label any element of H , we have \left|{\pi_{n}(W_{H})}\right|=\left|{\mathcal{H}}\right|^{\sum_{i=0}^{n-1}\#\mathcal{L}_{i}}=\left|{H}\right|^{\frac{d^{n}-1}{d-1}}, since \#\mathcal{L}_{i}=d^{i} . This in particular applies when H=\Sym(d) that corresponds to the whole groups of automorphisms as W_{\Sym(d)}=\Aut(T) . Therefore, \displaystyle\mathcal{H}(W_{H})=\liminf_{n\rightarrow+\infty}\frac{\log(\left|{\pi_{n}(W_{H})}\right|)}{\log(\left|{\pi_{n}(\Aut(T))}\right|)}=\frac{\log(\left|{H}\right|)}{\log(\left|{\Sym(d)}\right|)}=\frac{\log(\left|{H}\right|)}{\log(d!)}. ∎ 2.4. Fixed-point proportion We define the fixed-point proportion of a subgroup G of \Aut(T) acting on a spherically homogeneous tree T as (3) \FPP(G)=\lim_{n\rightarrow\infty}\frac{\#{\left\{g\in\pi_{n}(G):\text{$g$ fixes a vertex in $\mathcal{L}_{n}$}\right\}}}{\left|{\pi_{n}(G)}\right|}. Lemma 2.9 . The fixed-point proportion of G is well-defined, namely, the limit in Equation 3 always exists. Proof. Consider the surjective map \pi:\pi_{n+1}(G)\rightarrow\pi_{n}(G) , namely, the projection of the action on level n+1 to level n . Notice that if h\in\pi_{n}(G) does not fix any vertex at level n , then all the preimages of h under \pi will not fix any vertex at level n+1 . Since \pi is a group homomorphism, the number of preimages is \left|{\pi_{n+1}(G)}\right|/\left|{\pi_{n}(G)}\right| for any point h\in\pi_{n}(G) and consequently we have the inequality \left|{\pi_{n+1}(G)}\right|-f_{n+1}\geq\frac{\left|{\pi_{n+1}(G)}\right|}{\left|{\pi_{n}(G)}\right|}\left(\left|{\pi_{n}(G)}\right|-f_{n}\right), where f_{n}=\#{\left\{g\in\pi_{n}(G):\text{$g$ fixes a vertex in $\mathcal{L}_{n}$}\right\}}. Rewriting the inequality, we obtain \frac{f_{n}}{\left|{\pi_{n}(G)}\right|}\geq\frac{f_{n+1}}{\left|{\pi_{n+1}(G)}\right|}, that is exactly the sequence considered in the limit of the fixed-point proportion. Since the sequence is bounded below by zero and decreasing, the limit must exist. ∎ Lemma 2.10 . Let G\leq\Aut(T) and H\leq_{f}G . Then (1) \FPP(G)=\FPP(\overline{G}) , (2) \FPP(H)\leq[G:H]\,\FPP(G) . In particular, if \FPP(G)=0 then \FPP(H)=0 . Proof. For (1), we just observe that \pi_{n}(G)=\pi_{n}(\overline{G}) for all n\in\mathbb{N} and the definition of the fixed-point proportion only depends on \pi_{n}(G) . For (2), we know that if the index is finite, then there exists n_{0}\in\mathbb{N} such that [\pi_{n}(G):\pi_{n}(H)]=[G:H] for all n\geq n_{0} . On the other side, if K\leq\Aut(T) and we let f_{n}(K) denote the number of elements in \pi_{n}(K) that fix a vertex in \mathcal{L}_{n} , we have that f_{n}(H)\leq f_{n}(G) since H\leq G . Therefore \displaystyle\FPP(H)=\lim_{n\rightarrow+\infty}\frac{f_{n}(H)}{\left|{\pi_{n}(H)}\right|}=[G:H]\lim_{n\rightarrow+\infty}\frac{f_{n}(H)}{\left|{\pi_{n}(G)}\right|}\leq[G:H]\lim_{n\rightarrow+\infty}\frac{f_{n}(G)}{\left|{\pi_{n}(G)}\right|}= \displaystyle[G:H]\,\FPP(G). ∎ By 2.3 , we have that if H\leq_{f}G and G is infinite, then \mathcal{H}_{G}(H)=1 . Therefore, a natural generalization for 2.10 .(2) is to consider the case of H\leq G such that \mathcal{H}_{G}(H)=1 . However, we have the following counterexample. Let G\leq\Aut(T) be an infinite group with null fixed-point proportion and take the subgroup H:=\bigcap_{n\in\mathbb{N}}\st_{G}(0^{n}) , where 0^{n} is the leftmost vertex at level n . Then, we have \pi_{n}(H)=\pi_{n}(\st_{G}(0^{n})) and therefore every element in \pi_{n}(H) fixes 0^{n} , so \FPP(H)=1 . By the orbit-stabilizer theorem, we have [\pi_{n}(G):\pi_{n}(H)]\leq\#\mathcal{L}_{n} and consequently \displaystyle\mathcal{H}_{G}(H)\geq 1-\frac{\log(\#\mathcal{L}_{n})}{\log(\left|{\pi_{n}(G)}\right|)} Then, it is not hard to find a group G such that the previous limit converges to one. For instance, G=\Aut(T) . Due to 2.10 .(1), we may assume that G is a closed subgroup of \Aut(T) . Since \Aut(T) is a profinite group and G is closed, then G is compact and has a unique probability Haar measure \mu . A remarkable property of the Haar measure is that if S is a measurable set in \overline{G} , then (4) \mu(S)=\lim_{n\to+\infty}\frac{\#\pi_{n}(S)}{\left|{\pi_{n}(\overline{G})}\right|}. Given G , a closed subgroup of \Aut(T) , we define the fixed-point process of G as the sequence of functions {\left\{X_{n}:G\rightarrow\mathbb{N}\right\}}_{n\in\mathbb{N}} such that X_{n}(g)=\#{\left\{v\in\mathcal{L}_{n}:g(v)=v\right\}}. By the definition of the Haar measure, we have (5) \displaystyle\begin{split}\FPP(G)=\lim_{n\rightarrow\infty}\frac{\#{\left\{g\in\pi_{n}(G):\text{$g$ fixes a vertex in $\mathcal{L}_{n}$}\right\}}}{\left|{\pi_{n}(G)}\right|}=\\ \mu({\left\{g\in G:\text{ $g$ fixes one vertex in $\mathcal{L}_{n}$ for all $n\in\mathbb{N}$}\right\}})=\\ \mu\left(\bigcap_{n\geq 0}{\left\{g\in G:X_{n}(g)>0\right\}}\right).\end{split} So the fixed-point process is involved in the calculation of the fixed-point proportion (see Jones2014 for more details). Lemma 2.11 . Let T be a spherically homogeneous tree and {\left\{G_{n}\right\}}_{n\in\mathbb{N}} a sequence of subgroups of \Aut(T) such that G_{n}\leq G_{n+1} and write G=\bigcup_{n\in\mathbb{N}}G_{n} . Then \FPP(G_{n})\xrightarrow[n\rightarrow+\infty]{}\FPP(G) . Proof. By 2.10 , we may assume that the subgroups G_{n} ’s and G are closed. Let \mu_{n} be the Haar measure on G_{n} and consider \hat{\mu}_{n} the extension of \mu_{n} to G defined by \hat{\mu}_{n}(A)=\mu_{n}(A\cap G_{n}). In particular, \FPP(G_{n})=\hat{\mu}_{n}\left(\bigcap_{n\geq 0}{\left\{g\in G:X_{n}(g)>0\right\}}\right) Denote by \mu the Haar measure on G . By ( Katznelson1968 , Exercise 4, Chapter VII) , the sequence \hat{\mu}_{n}\xrightarrow[n\rightarrow+\infty]{}\mu in the weak-* topology, so \displaystyle\FPP(G_{n})=\hat{\mu}_{n}\left(\bigcap_{n\geq 0}{\left\{g\in G:X_{n}(g)>0\right\}}\right)\xrightarrow[n\rightarrow+\infty]{} \displaystyle\mu\left(\bigcap_{n\geq 0}{\left\{g\in G:X_{n}(g)>0\right\}}\right)=\FPP(G). ∎ 2.5. Iterated Galois groups Let K be a field, and let f be a rational function of degree at least 2 (namely, the maximum degree of numerator and denominator is at least 2 ) and with coefficients in K . We denote the n th iteration of f by f^{n} . Let t be a transcendental element over K and fix K(t)^{sep} a separable closure of K(t) . Let us denote f^{-n}(t) the set of preimages of t by f^{n} inside K(t)^{sep} and K_{n}(f):=K(f^{-n}(t)) . Then, K_{n}(f)/K(t) is a Galois extension, and K_{n}(f)\subseteq K_{n+1}(f) for all n\geq 0 (see Jones2014 ). Define K_{\infty}(f):=\bigcup_{n\geq 0}K_{n}(f) . Then, K_{\infty}(f)/K(t) is also a Galois extension. For n\in\mathbb{N}\cup{\left\{\infty\right\}} , define also G_{n}(K,f,t):=\Gal(K_{n}(f)/K(t)). By the definition of K_{\infty}(f) we have G_{\infty}(K,f,t)=\varprojlim G_{n}(K,f,t) . So, in particular G_{\infty}(K,f,t) is a profinite group that acts naturally on f^{-n}(t) by permuting its elements. We call G_{\infty}(K,f,t) the iterated Galois group of f at t . Lemma 2.12 . Let K be a field of characteristic zero, f a rational function in K(x) of degree d and t transcendental over K . Then \#f^{-n}(t)=d^{n} . Proof. Write f^{n}(z)=p_{n}(z)/q_{n}(z) with p_{n},q_{n}\in K[x] such that \gcd(p_{n},q_{n})=1 . Then z\in f^{-n}(t) if and only if we have p_{n}(z)-tq_{n}(z)=0 . As \deg(f^{n})=d^{n} , then \deg_{x}(p_{n}(x)-tq_{n}(x))=d^{n} , so in order to prove that \#f^{-n}(t)=d^{n} , it is enough to prove that p_{n}(x)-tq_{n}(x) is irreducible as a polynomial in K(t)[x] since K is a field of characteristic zero. Write p_{n}(x)-tq_{n}(x)=A(x,t)B(x,t) . As the left hand side is a polynomial in both variables, we may assume A,B\in K[x,t] . The left hand side has degree 1 on t , therefore, without loss of generality, we may assume A\in K[x] and B(x,t)=b_{0}(x)+tb_{1}(x) . Then Ab_{0}=p_{n} and Ab_{1}=q_{n} . As \gcd(p_{n},q_{n})=1 , then A\in K . ∎ In light of 2.12 , we can create a d -regular rooted tree T by putting the n -th preimages of t by f on the n -th level of the tree and connecting a vertex v\in\mathcal{L}_{n} and a vertex w\in\mathcal{L}_{n+1} if and only if f(w)=v . Then, G_{\infty}(K,f,t) acts on T faithfully and preserving adjacency (see again Jones2014 for a proof of this fact). Therefore, the iterated Galois group can be embedded as a closed subgroup of \Aut(T) . 3. The fixed-point proportion of iterated wreath products We can extend the definition of the fixed-point proportion given in Equation 3 to any set X\subseteq\Aut(T) in the following way: (6) \FPP(X):=\lim_{n\rightarrow\infty}\frac{\#{\left\{x\in\pi_{n}(X):\text{$x$ fixes a vertex in $\mathcal{L}_{n}$}\right\}}}{\left|{\pi_{n}(X)}\right|}. Contrary to 2.9 , the existence of the limit above is not guaranteed for every set X . Let S be any subset of \Sym(d) and define W_{S}:={\left\{g\in\Aut(T):g|_{v}^{1}\in S,\forall v\in T\right\}}. Notice that if S=H is a subgroup of \Sym(d) , the set W_{H} corresponds to the definition of iterated wreath product given in Equation 2 . For this reason, we will also refer to W_{S} as iterated wreath products, even though they are not necessarily subgroups of \Aut(T) . We will devote this section to proving that \FPP(W_{S}) always exists and give a way to calculate it for any set S . Let f_{n} denote the number of elements in \pi_{n}(W_{S}) that fix at least one vertex in \mathcal{L}_{n} , write \sigma_{n}=\left|{\pi_{n}(W_{S})}\right| and p_{n} be the proportion of them, namely, p_{n}=\frac{f_{n}}{\sigma_{n}} . The idea will be to give a recurrence formula for p_{n} that allows us to calculate the fixed-point proportion of W_{S} as \FPP(W_{S})=\lim_{n\rightarrow+\infty}p_{n}. By definition of W_{S} , we have \sigma_{n+1}=\sigma_{n}^{d}\,\,(\#S) , because at each vertex of level 1 we can put any element of \pi_{n}(W_{S}) and then at the root we have all the options of S . Another way to think of \sigma_{n+1} is that we have \sigma_{n} options for the first n levels and then at each vertex of level n we can put as label any element of S , so \sigma_{n+1}=\sigma_{n}\,\,(\#S)^{\#\mathcal{L}_{n}}=\sigma_{n}\,\,(\#S)^{d^{n}} . Lemma 3.1 . Given S\subseteq\Sym(d) , then \FPP(W_{S}) always exists. Proof. To prove that \FPP(W_{S}) exists, we follow the argument given in 2.9 . Notice that if an element in \pi_{n}(W_{S}) does not fix any vertex in level n , then any extension to level n+1 of that element will not fix any vertex in level n+1 either. The number of extensions of each element is (\#S)^{\#\mathcal{L}_{n}}=(\#S)^{d^{n}} , therefore \sigma_{n+1}-f_{n+1}\geq(\#S)^{d^{n}}(\sigma_{n}-f_{n})=\frac{\sigma_{n+1}}{\sigma_{n}}(\sigma_{n}-f_{n})=\sigma_{n+1}(1-p_{n}). Reordering terms, we get p_{n+1}\leq p_{n} . Thus, {\left\{p_{n}\right\}}_{n\in\mathbb{N}} is a decreasing sequence bounded by zero and therefore \FPP(W_{S}) exists. ∎ Definition 3.2 . Let d\in\mathbb{N} and S\subseteq\Sym(d) . Define D_{S}(k) as the number of permutations in S having exactly k fixed points. This definition generalizes the concept of the number of derangements for any set (see Hassani2003 ). Clearly, \sum_{k=0}^{d}D_{S}(k)=\#S. Our next step will be to find a recurrence formula for f_{n+1} . Let s\in S be a label for the root with k fixed points. If k=0 , the resulting elements will not produce fixed points at level n+1 , so we assume that k>0 . An element g\in\pi_{n+1}(W_{S}) whose label at the root is s will have fixed points at level n+1 only in the subtrees fixed by s . In the remaining d-k subtrees, any element of \pi_{n}(W_{S}) can be placed freely, contributing a factor of \sigma_{n}^{d-k} . In the k fixed subtrees, we need to place at least one element of \pi_{n}(W_{S}) that fixes a vertex, while the rest may be arbitrary. To avoid repetition, we consider the disjoint sets F_{n}={\left\{h\in\pi_{n}(W_{S}):\text{$h$ has fixed points on $\mathcal{L}_{n}$}\right\}}\text{ and }F_{n}^{c}. We then sum in all the possible k -fold Cartesian products of F_{n} and F_{n}^{c} , excluding the product F_{n}^{c}\times\dots\times F_{n}^{c} , which does not produce elements with fixed points at level n+1 . To express this formally, let \beta\in{\left\{\pm 1\right\}} and define \displaystyle\overline{f_{n}}^{\beta}=\left\{\begin{matrix}f_{n}&\mbox{if $\beta=1$,}\\ \sigma_{n}-f_{n}&\mbox{if $\beta=-1$.}\end{matrix}\right. We conclude the following formula: (7) f_{n+1}=\sum_{k=1}^{d}D_{S}(k)\sigma_{n}^{d-k}\sum_{(\beta_{1},\dots,\beta_{k})\in{\left\{\pm 1\right\}}^{k}\setminus{\left\{(-1,\dots,-1)\right\}}}\prod_{i=1}^{k}\overline{f_{n}}^{\beta_{i}} The following lemma will help to clarify the intricate formula: Lemma 3.3 . Let d\in\mathbb{N} and S\subseteq\Sym(d) . Using the notation introduced earlier, we have (8) f_{n+1}=\sum_{k=1}^{d}D_{S}(k)\sigma_{n}^{d-k}(\sigma_{n}^{k}-(\sigma_{n}-f_{n})^{k}). Proof. First, consider the sum \sum_{(\beta_{1},\dots,\beta_{k})\in{\left\{\pm 1\right\}}^{k}}\prod_{i=1}^{k}\overline{f_{n}}^{\beta_{i}}. Fix t\in{\left\{0,\dots,k\right\}} . Notice that the number of terms of the form f_{n}^{t}(\sigma_{n}-f_{n})^{k-t} corresponds to the number of vectors (\beta_{1},\dots,\beta_{k}) having t entries equal to 1 . To count the number of such vectors, we can think of {\left\{1,\dots,k\right\}} as the set of indices, and we are selecting t of them to assign the value 1 . There are exactly \binom{k}{t} ways to do this. Hence, \sum_{(\beta_{1},\dots,\beta_{k})\in{\left\{\pm 1\right\}}^{k}}\prod_{i=1}^{k}\overline{f_{n}}^{\beta_{i}}=\sum_{t=0}^{k}\binom{k}{t}f_{n}^{t}(\sigma_{n}-f_{n})^{k-t}. By the binomial theorem, \sum_{t=0}^{k}\binom{k}{t}f_{n}^{t}(\sigma_{n}-f_{n})^{k-t}=(f_{n}+(\sigma_{n}-f_{n}))^{k}=\sigma_{n}^{k}. Now, considering the term (\beta_{1},\dots,\beta_{k})=(-1,\dots,-1) separately, we obtaining: \sum_{(\beta_{1},\dots,\beta_{k})\in{\left\{\pm 1\right\}}^{k}\setminus{\left\{(-1,\dots,-1)\right\}}}\prod_{i=1}^{k}\overline{f_{n}}^{\beta}_{i}=\sigma_{n}^{k}-(\sigma_{n}-f_{n})^{k}. Substituting this result into Equation 7 , f_{n+1}=\sum_{k=1}^{d}D_{S}(k)\sigma_{n}^{d-k}(\sigma_{n}^{k}-(\sigma_{n}-f_{n})^{k}) and the result follows. ∎ Dividing Equation 8 by \sigma_{n+1} , using its recurrence formula \sigma_{n+1}=\sigma_{n}^{d}\,\,\#S and replacing \frac{f_{n}}{\sigma_{n}} with p_{n} , we obtain (9) p_{n+1}=\frac{f_{n+1}}{\sigma_{n+1}}=\sum_{k=1}^{d}\frac{D_{S}(k)}{\#S}\left(1-\left(1-p_{n}\right)^{k}\right). Definition 3.4 . Let d\in\mathbb{N} and S\subseteq\Sym(d) . We define the characteristic polynomial of S as f_{S}(x):=\sum_{k=1}^{d}\frac{D_{S}(k)}{\#S}\left(1-\left(1-x\right)^{k}\right). Remark 3.5 . Using 3.4 and Equation 9 we conclude that p_{n+1}=f_{S}(p_{n}). Taking limit when n\rightarrow+\infty , we deduce that \FPP(W_{S}) is a fixed point of the characteristic polynomial of f_{S} . Lemma 3.6 . Let d\in\mathbb{N} and S\subseteq\Sym(d) . The characteristic polynomial f_{S} depends only on the conjugacy class of S in \Sym(d) . Proof. Let S^{\prime} be a set in \Sym(d) such that S^{\prime}=gSg^{-1} for some g\in\Sym(d) . For x\in{\left\{1\dots,d\right\}} and s\in S , observe that x is fixed by gsg^{-1} if and only if g^{-1}(x) is fixed by s . Since g is a permutation, this implies that D_{S^{\prime}}(k)=D_{S}(k) for all k=0,\dots,d . Consequently, f_{S^{\prime}}=f_{S} . ∎ Proposition 3.7 . Let d\in\mathbb{N} , S\subseteq\Sym(d) , and f_{S}:[0,1]\rightarrow\mathbb{R} be the characteristic polynomial of S . Then f_{S} has the following properties: (1) f_{S}(x)\in[0,1] for all x\in[0,1] and f_{S}(0)=0 . Moreover, if \deg(f_{S})>0 , then f_{S}(x)=0 if and only if x=0 . (2) f_{S}^{\prime}(x)\geq 0 , and if \deg(f_{S})>0 , then f_{S}^{\prime}(x)>0 for x\in[0,1) . (3) f_{S}^{\prime\prime}(x)\leq 0 for x\in[0,1] , and strictly negative in [0,1) if \deg(f_{S})>1 . (4) If f_{S} is not the identity function, the equation f_{S}(x)=x has at most two solutions in [0,1] , with x=0 being one of them. Proof. For (1), since x\in[0,1] , we have 0\leq 1-(1-x)^{k}\leq 1 , and it equals zero if and only if x=0 for all k . Furthermore, D_{S}(k)\geq 0 for all k , so 0\leq f_{S}(x)\leq\sum_{k=1}^{d}\frac{D_{S}(k)}{\#S}\leq\sum_{k=0}^{d}\frac{D_{S}(k)}{\#S}=1. The only exception occurs when \deg(f_{S})<1 , where f_{S}(x)=0 . For (2), the derivative of f_{S} is f_{S}^{\prime}(x)=\sum_{k=1}^{d}\frac{D_{S}(k)k}{\#S}(1-x)^{k-1}\geq 0 for x\in[0,1] . Moreover, if \deg(f_{S})>0 , then (1-x)^{k-1}>0 for x\in[0,1) , making f_{S}^{\prime}(x) is strictly positive in [0,1) . For (3), the second derivative of f_{S} is: f_{S}^{\prime\prime}(x)=-\sum_{k=2}^{d}\frac{D_{S}(k)}{\#S}k(k-1)(1-x)^{k-2}\leq 0 for x\in[0,1] . If \deg(f_{S})>1 , then D_{S}(k)>0 for some k\geq 2 , and f_{S}^{\prime\prime}(x)<0 for x\in[0,1) . For (4), we use the strict concavity of f_{S} established in (3). Concaveness means that for all x,y in the domain and \alpha\in[0,1] , we have f_{S}((1-\alpha)x+\alpha y)>(1-\alpha)f_{S}(x)+\alpha f_{S}(y). Suppose that f_{S} has another fixed-point different to 0 in [0,1] and call x_{1} the smallest possible. If x_{1}=1 we are done, if not, let y\in(x_{1},1] , x=0 and take \alpha\in(0,1) such that (1-\alpha)x+\alpha y=\alpha y=x_{1} . Then, f_{S}(x_{1})=x_{1}>\alpha f_{S}(y). Dividing by \alpha , we obtain that y>f_{S}(y) and this for all y>x_{1} , so we at most have two fixed points. ∎ We are now ready to conclude the main result about the fixed-point proportion of W_{S} : Theorem 3.8 . Let d\in\mathbb{N} and S\subseteq\Sym(d) . Then: (1) \FPP(W_{S}) equals the largest fixed point of f_{S} in [0,1] . (2) \FPP(W_{S}) is the solution of a polynomial of degree at most d-1 with coefficients in \mathbb{Z}\left[\frac{1}{\#S}\right] . (3) \FPP(W_{S})=0 if and only if f_{S}^{\prime}(0)\leq 1 and f_{S} is not the identity function. (4) \FPP(W_{S})=1 if and only if every element in S fixes at least one element in {\left\{1,\dots,d\right\}} . Proof. For (1), as it was observed in 3.5 , the fixed-point proportion of W_{S} is a fixed point of f_{S} in [0,1] . If \deg(f_{S})<1 , then f_{S}(x)=0 and the only fixed point is zero. If \deg(f_{S})=1 , then f_{S}^{\prime}(0)=\frac{D_{S}(1)}{\#S}\leq 1 . If f_{S}^{\prime}(0)<1 , then the only fixed point is zero. If f_{S}^{\prime}(0)=1 , this implies that D_{S}(1)=\#S or equivalently that every element in S fixes exactly one point. Therefore \FPP(W_{S})=1 , coinciding with the largest fixed point of f_{S} . If \deg(f_{S})>1 , then by 3.7 , the function f_{S} has at most two fixed points, with 0 as one of them. If 0 is the only fixed point we are done. Otherwise, let x_{0}>0 be the other fixed point. Consider g(x)=f_{S}(x)-x . Then g(0)=g(x_{0})=0 and by Rolle’s theorem there exists c\in(0,x_{0}) such that g^{\prime}(c)=0 or equivalently that f_{S}^{\prime}(c)=1 . Since f_{S}^{\prime\prime} is strictly negative, then f_{S}^{\prime} is strictly decreasing but since f_{S}^{\prime} is always positive then 01 , \FPP(W_{S}) satisfies the polynomial: \frac{f_{S}(x)}{x}-1=0, which has degree at most d-1 and coefficients in \mathbb{Z}\left[\frac{1}{\#S}\right] , as D_{S}(k)\in\mathbb{Z} . For (3), the direct was already proved in (1). For the converse, if f_{S}^{\prime}(0)\leq 1 and \deg(f_{S})>1 , since f_{S}^{\prime} is strictly decreasing, then f_{S}(x)0 and the only fixed point is 0 . For (4), \FPP(W_{S})=1 if and only if f_{S}(1)=\sum_{k=1}^{d}\frac{D_{S}(k)}{\#S}=1. Since \#S=\sum_{k=0}^{d}D_{S}(k)=1 , this implies that \FPP(W_{S})=1 if and only if D_{S}(0)=0 , meaning every element in S fixes at least one element in {\left\{1,\dots,d\right\}} . ∎ A particular case that we will be of interest in this article is when S is a coset in \Sym(d) . We first need the following definitions: Definition 3.9 . Let G be a group acting on a set Y via the action \rho:G\curvearrowright Y . Let S be a set of G , an element s\in S and y\in Y . We denote by Y^{s}:={\left\{y\in Y:sy=y\right\}} and by \Stab_{S}(y):={\left\{s\in S:sy=y\right\}} . Remark 3.10 . If we consider G=\Sym(d) with the natural action on X={\left\{1,\dots,d\right\}} and a set S\subseteq\Sym(d) , we have \displaystyle\frac{1}{\#S}\sum_{s\in S}\#X^{s}=\frac{1}{\#S}\sum_{k=0}^{d}\,\,\sum_{s\in S:\#X^{s}=k}k=\frac{1}{\#S}\sum_{k=0}^{d}D_{S}(k)k=\frac{1}{\#S}\sum_{k=1}^{d}D_{S}(k)k=f_{S}^{\prime}(0). If S is a coset, the quantity on the left-hand side can be studied as a generalization of Burnside’s lemma: Lemma 3.11 (Burnside Lemma for cosets) . Let G\curvearrowright Y be an action of a finite group G over a finite set Y . Let H\leq G and A=gH be a coset of H in G . (1) If H\curvearrowright Y acts transitively, then \frac{1}{\#A}\sum_{a\in A}\#Y^{a}=1. (2) Define Y^{*}:={\left\{y\in Y:\exists h\in H\text{ s.t. }hy=g^{-1}y\right\}} . If g\in N_{G}(H) , the normalizer of H in G , then \frac{1}{\#A}\sum_{a\in A}\#Y^{a}=\#(Y^{*}/H). Proof. Consider the set Z={\left\{(a,y)\in A\times Y:ay=y\right\}} . We can count the elements of Z in two different ways. On one hand, we have: \#Z=\sum_{a\in A}\#Y^{a}. On the other hand, we have: \#Z=\sum_{y\in Y}\#\Stab_{A}(y). Notice that a\in\Stab_{A}(y) if and only if ay=y . Since A=gH , there exists h\in H such that a=gh , so ay=y if and only if hy=g^{-1}y . If no such h exists, then \Stab_{A}(y)=\emptyset , so we can restrict the sum to Y^{*} . If y\in Y^{*} , and h_{0}\in H is an element such that h_{0}y=g^{-1}y , then we have a bijection between \Stab_{H}(y) and \Stab_{A}(y) given by h\mapsto gh_{0}h . Therefore \#Z=\sum_{y\in Y^{*}}\#\Stab_{H}(y). By the orbit-stabilizer theorem, we know that \#\Stab_{H}(y)=\frac{\left|{H}\right|}{\#\orb_{H}(y)}, so \#Z=\left|{H}\right|\sum_{y\in Y^{*}}\frac{1}{\#\orb_{H}(y)}. If H is transitive, then Y^{*}=Y , and thus \#Z=\left|{H}\right|\sum_{B\in Y/H}\sum_{y\in B}\frac{1}{\#B}=\left|{H}\right|\#(Y/H)=\left|{H}\right|=\#A. This gives the formula in the first case. If g\in N_{G}(H) , the action G\curvearrowright Y restricts to H\curvearrowright Y^{*} . Indeed, if y\in Y^{*} , there exists h_{0}\in H such that h_{0}y=g^{-1}y . Then, for h\in H , there exists h^{\prime}\in H such that g^{-1}h=h^{\prime}g^{-1} . Therefore g^{-1}(hy)=h^{\prime}g^{-1}y=h^{\prime}h_{0}y=(h^{\prime}h_{0}h^{-1})h(y). Since h^{\prime}h_{0}h^{-1}\in H , we obtain that hy\in Y^{*} . Thus, we can split the sum over the orbits Y^{*}/H , obtaining: \#Z=\left|{H}\right|\sum_{B\in Y^{*}/H}\sum_{y\in B}\frac{1}{\#B}=\left|{H}\right|\#(Y^{*}/H)=\#A\,\,\#(Y^{*}/H). This gives the formula in the second case. ∎ In the case that S is a subgroup in Theorem 3.8 , we deduce the following corollary for the fixed-point proportion of iterated wreath products: Corollary 3.12 . Let d\in\mathbb{N} with d\geq 2 , the set X={\left\{1,\dots,d\right\}} and \mathcal{P}\leq\Sym(d) . Then the fixed point proportion of W_{\mathcal{P}} is: \displaystyle\text{FPP}(W_{\mathcal{P}})=\left\{\begin{matrix}0&\mbox{if $\mathcal{P}$ is transitive over $X$,}\\ \alpha\in(0,1)&\mbox{if $\mathcal{P}$ not transitive and $\exists\,\sigma$ with no fixed points}\\ 1&\mbox{if $\mathcal{P}$ not transitive and every element fixes a point.}\end{matrix}\right. Moreover, the value \FPP(W_{\mathcal{P}}) is the solution of a polynomial of degree d-1 with coefficients in \mathbb{Z}\left[\frac{1}{\left|{\mathcal{P}}\right|}\right] . Proof. Since \mathcal{P} is a subgroup, the identity is in \mathcal{P} so D_{\mathcal{P}}(d)=1 and therefore \deg(f_{\mathcal{P}})=d\geq 2 . The case when \FPP(W_{\mathcal{P}})=1 , follows directly from Theorem 3.8 . The case when \FPP(\mathcal{P})=\alpha\in(0,1) follows from the fact that \alpha is the largest fixed point of f_{\mathcal{P}} and f_{\mathcal{P}}(0)=0 , so it is a solution of \frac{f_{\mathcal{P}}(x)}{x}-1=0 that is a polynomial with degree d-1 . Finally, by Theorem 3.8 , we have \FPP(W_{\mathcal{P}})=0 if and only if f_{\mathcal{P}}^{\prime}(0)\leq 1 and f_{\mathcal{P}} is not the identity function. Since \deg(f_{\mathcal{P}})>1 , then \FPP(W_{\mathcal{P}})=0 if and only if f_{\mathcal{P}}^{\prime}(0)\leq 1 . By 3.10 and 3.11 , we have f_{\mathcal{P}}^{\prime}(0)=\frac{1}{\left|{\mathcal{P}}\right|}\sum_{k=1}^{d}D_{\mathcal{P}}(k)k=\#(X/\mathcal{P}), so \FPP(W_{\mathcal{P}})=0 if and only if \mathcal{P} is transitive over X . ∎ Running a program in SAGE sagemath , we compute the characteristic polynomials of the subgroups of \Sym(d) for d=3 and 4 . By 3.6 , it suffices to consider the conjugacy classes of subgroups in \Sym(d) , since they have the same characteristic polynomial. For d=3 , the group \Sym(3) has four conjugacy classes. Representatives of these classes are: {\left\{1\right\}} , {\left\langle{(2,3)}\right\rangle} , {\left\langle{(1,2,3)}\right\rangle} and \Sym(3) . The corresponding fixed-point proportions are 1 , 1 , 0 and 0 , respectively. This matches with Corollary 3.12 , as the first two classes fix at least one point and the last two are transitive. Figure 1 illustrates the functions f_{\mathcal{P}} for each conjugacy class of \mathcal{P} in \Sym(3) . Notice that the intersections of f_{\mathcal{P}} with the identity function are only at 0 and 1 , as expected. In particular, there are no examples of iterated wreath products with non-trivial fixed-point proportion for d=3 . 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 Figure 1. Plot of the functions f_{\mathcal{P}} for \mathcal{P} subgroups of \Sym(3) . For d=4 , the group \Sym(4) has eleven conjugacy classes. In four of them (the ones corresponding to \Sym(3) ) their elements always fix a point, so their fixed-point proportion is 1 . There are five transitive conjugacy classes, namely, their fixed-point proportion is 0 . In addition, there are two conjugacy classes whose fixed-point proportion is not trivial. Possible representatives for these are {\left\langle{(1,2),(3,4)}\right\rangle} and {\left\langle{(3,4),(1,2)(3,4)}\right\rangle} . Notice that these classes are not transitive because their elements cannot send 1 to 3 , nor do they fix a particular point. The corresponding fixed-point proportions are approximately 0.45631\dots and 0.70440\dots respectively. Figure 2 shows the functions f_{\mathcal{P}} for each conjugacy class of \mathcal{P} in \Sym(4) . 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 Figure 2. Plot of the functions f_{\mathcal{P}} for \mathcal{P} subgroups of \Sym(4) . 4. The main construction In this section, we will use Theorem 3.8 to construct a family of regular branch groups whose fixed-point proportion and Hausdorff dimension can be computed explictly. Given d\geq 2 and T a d -regular tree, consider two subgroups \mathcal{Q},\mathcal{P}\leq\Sym(d) such that 1\neq\mathcal{Q}\lhd\mathcal{P} . Define G_{\mathcal{Q}}^{\mathcal{P}}={\left\{g\in\Aut(T):g|_{v}^{1}\in\mathcal{P},\,(g|_{v}^{1})(g|_{w}^{1})^{-1}\in\mathcal{Q}\text{ for all $v,w\in T$}\right\}}. In other words, the elements in this group lie in the iterated wreath product W_{\mathcal{P}} (see Section 2 for the definition of W_{\mathcal{P}} ) and each element has its labels in the same coset of \mathcal{P}/\mathcal{Q} . For example, if \mathcal{Q}=\mathcal{P} , then G_{\mathcal{Q}}^{\mathcal{P}}=W_{\mathcal{P}} . Lemma 4.1 . Let T be a d -regular tree with d\geq 2 and 1\neq\mathcal{Q}\lhd\mathcal{P}\leq\Sym(d) . Then, the group G_{\mathcal{Q}}^{\mathcal{P}} satisfies the following properties: (1) G_{\mathcal{Q}}^{\mathcal{P}} contains W_{\mathcal{Q}} , (2) W_{\mathcal{Q}}\lhd G_{\mathcal{Q}}^{\mathcal{P}} , (3) [G_{\mathcal{Q}}^{\mathcal{P}}:W_{\mathcal{Q}}]=[\mathcal{P}:\mathcal{Q}] . Proof. We start proving that G_{\mathcal{Q}}^{\mathcal{P}} is a group. Clearly, the identity element is in G_{\mathcal{Q}}^{\mathcal{P}} . Next, suppose g,h\in G_{\mathcal{Q}}^{\mathcal{P}} . We must verify that gh^{-1}\in G_{\mathcal{Q}}^{\mathcal{P}} . For v,w\in T , by the properties of sections (see Equation 1 ), we have: \displaystyle(gh^{-1})|_{v}^{1}\,\,((gh^{-1})|_{w}^{1}))^{-1}=(g|_{h^{-1}(v)}^{1})\,(h|_{h^{-1}(v)}^{1})^{-1}\,(h|_{h^{-1}(w)}^{1})\,(g|_{h^{-1}(w)}^{1})^{-1}. Rewriting, this becomes: \displaystyle\left[(g|_{h^{-1}(v)}^{1})\,(h|_{h^{-1}(v)}^{1})^{-1}\,(h|_{h^{-1}(w)}^{1})\,(g|_{h^{-1}(v)}^{1})^{-1}\right]\left[((g|_{h^{-1}(v)}^{1})(g|_{h^{-1}(w)}^{1})^{-1})\right]. The second term between brackets is in \mathcal{Q} since g\in G_{\mathcal{Q}}^{\mathcal{P}} and the first term in brackets is in \mathcal{Q} because h\in G_{\mathcal{Q}}^{\mathcal{P}} and \mathcal{Q}\lhd\mathcal{P} . This ensures that gh^{-1}\in G_{\mathcal{Q}}^{\mathcal{P}} . Hence, G_{\mathcal{Q}}^{\mathcal{P}} is a group. We now prove that W_{\mathcal{Q}}\lhd G_{\mathcal{Q}}^{\mathcal{P}} . The proof is similar to the previous calculation. Specifically, we need to show that if g\in G_{\mathcal{Q}}^{\mathcal{P}} and h\in W_{\mathcal{Q}} then (ghg^{-1})|_{v}^{1}\in\mathcal{Q} for all v\in T . Indeed, \displaystyle(ghg^{-1})|_{v}^{1}=g|_{hg^{-1}(v)}^{1}\,h|_{g^{-1}(v)}^{1}\,(g|_{g^{-1}(v)}^{1})^{-1}. Rewriting, this becomes: \displaystyle(ghg^{-1})|_{v}^{1}=\left[g|_{hg^{-1}(v)}^{1}\,h|_{g^{-1}(v)}^{1}\,(g|_{hg^{-1}(v)}^{1})^{-1}\right]\,\left[(g|_{hg^{-1}(v)}^{1})\,(g|_{g^{-1}(v)}^{1})^{-1}\right]. The first term in brackets belongs to \mathcal{Q} because \mathcal{Q}\lhd\mathcal{P} , and the second term in bracket belongs to \mathcal{Q} since g\in G_{\mathcal{Q}}^{\mathcal{P}} . Thus, (ghg^{-1})|_{v}^{1}\in\mathcal{Q} and consequently W_{\mathcal{Q}}\lhd G_{\mathcal{Q}}^{\mathcal{P}} . Finally, consider the map (10) \displaystyle\begin{split}G_{\mathcal{Q}}^{\mathcal{P}}/W_{\mathcal{Q}}&\rightarrow\mathcal{P}/\mathcal{Q}\\ [g]&\mapsto[g|_{\emptyset}^{1}].\end{split} The map is well-defined because if h\in W_{\mathcal{Q}} and g\in G_{\mathcal{Q}}^{\mathcal{P}} then [(gh)|_{\emptyset}^{1}]=[g|_{\emptyset}^{1}][h|_{\emptyset}^{1}]=[g|_{\emptyset}^{1}], and it is in fact a morphism since the root of the tree is fixed by any element. To show surjectivity, let \sigma\in\mathcal{P} . Define g\in\Aut(T) such that g|_{v}^{1}=\sigma for all v\in T . Then, g\in G_{\mathcal{Q}}^{\mathcal{P}} and [g|_{\emptyset}^{1}]=[\sigma] . To prove injectivity, suppose [g|_{\emptyset}^{1}]=[h|_{\emptyset}^{1}] . Then: (g|_{\emptyset}^{1})(h|_{\emptyset}^{1})^{-1}=(gh^{-1})|_{\emptyset}^{1}\in\mathcal{Q}. Since gh^{-1}\in G_{\mathcal{Q}}^{\mathcal{P}} , then (gh^{-1})|_{v}^{1}\in\mathcal{Q} for all v\in T and consequently gh^{-1}\in W_{\mathcal{Q}} . ∎ Notice that if d=2 , we can only construct one group, since \Sym(2) only has two subgroups. In this case, \mathcal{Q}=\mathcal{P}=\Sym(2) and G_{\mathcal{Q}}^{\mathcal{P}}=\Aut(T) , that by Corollary 3.12 has null fixed-point proportion. Proposition 4.2 . Let T be a d -regular tree with d\geq 3 and 1\neq\mathcal{Q}\lhd\mathcal{P}\leq\Sym(d) . Then G_{\mathcal{Q}}^{\mathcal{P}} is a group of finite type of depth 2 and Hausdorff dimension \mathcal{H}(G)=\frac{\log(\left|{\mathcal{Q}}\right|)}{\log(d!)}. In particular, its Hausdorff dimension is always positive. Proof. By Section 2.3 , the group W_{\mathcal{Q}} is regular branch, branching over \St_{W_{\mathcal{Q}}}(1) . Furthermore, \St_{G_{\mathcal{Q}}^{\mathcal{P}}}(1)=\St_{W_{\mathcal{Q}}}(1) , so (\St_{W_{\mathcal{Q}}}(1))_{1}\leq_{f}\St_{W_{\mathcal{Q}}}(1)\leq_{f}G_{\mathcal{Q}}^{\mathcal{P}}. Therefore, G_{\mathcal{Q}}^{\mathcal{P}} is regular branch. To determine the depth of G_{\mathcal{Q}}^{\mathcal{P}} , we apply Theorem 2.7 . First, G_{\mathcal{Q}}^{\mathcal{P}} is not an iterated wreath product, so its depth is greater than 1 . Since \St_{G_{\mathcal{Q}}^{\mathcal{P}}}(1)=\St_{W_{\mathcal{Q}}}(1) , it follows by 2.5 that the group G_{\mathcal{Q}}^{\mathcal{P}} is branching over \St_{G_{\mathcal{Q}}^{\mathcal{P}}}(1) . Therefore G_{\mathcal{Q}}^{\mathcal{P}} is of finite type with depth D=2 by Theorem 2.7 , as D-1=1 . Finally, we compute the Hausdorff dimension of G_{\mathcal{Q}}^{\mathcal{P}} . By 2.3 , we have \mathcal{H}(G_{\mathcal{Q}}^{\mathcal{P}})=\mathcal{H}(W_{\mathcal{Q}}) . Then, by 2.8 , the result follows. ∎ The next question concerns the level-transitivity of G_{\mathcal{Q}}^{\mathcal{P}} . Proposition 4.3 . Let T be a d -regular tree with d\geq 3 and 1\neq\mathcal{Q}\lhd\mathcal{P}\leq\Sym(d) . Then G_{\mathcal{Q}}^{\mathcal{P}} is level-transitive if and only if \mathcal{Q} is transitive on X={\left\{1,\dots,d\right\}} . Proof. For the direct, let x,y\in X . Since G_{\mathcal{Q}}^{\mathcal{P}} is level-transitive, there exists g\in G_{\mathcal{Q}}^{\mathcal{P}} such that g(xx)=xy , where xx and xy represent vertices at level 2 as words in X . Denote a_{0}=g|_{\emptyset}^{1} and a_{1}=g|_{x}^{1} . By the definition of G_{\mathcal{Q}}^{\mathcal{P}} , there exists q\in\mathcal{Q} such that a_{1}=a_{0}q . Since \mathcal{Q}\lhd\mathcal{P} , there exists q^{\prime}\in\mathcal{Q} such that a_{0}q=q^{\prime}a_{0} . Thus, a_{1}=q^{\prime}a_{0} , or equivalently, a_{1}a_{0}^{-1}\in\mathcal{Q} . Furthermore, a_{1}a_{0}^{-1}(x)=a_{1}(x)=y, implying that \mathcal{Q} is transitive on X . For the converse, if \mathcal{Q} is transitive on X , then W_{\mathcal{Q}} is level-transitive and since W_{\mathcal{Q}}\leq G_{\mathcal{Q}}^{\mathcal{P}} , then G_{\mathcal{Q}}^{\mathcal{P}} is level-transitive as well. ∎ A closed group is said to be topologically finitely generated if it contains a finitely generated dense subgroup. Proposition 4.4 . Let T be a d -regular tree with d\geq 2 and 1\neq\mathcal{Q}\lhd\mathcal{P}\leq\Sym(d) . (1) If \mathcal{Q}\neq\mathcal{Q}^{\prime} or there exists x\in X={\left\{1,\dots,d\right\}} such that q(x)=x for all q\in\mathcal{Q} , then G_{\mathcal{Q}}^{\mathcal{P}} is not topologically finitely generated. (2) If \mathcal{Q} is transitive, then G_{\mathcal{Q}}^{\mathcal{P}} is topologically finitely generated if and only if \mathcal{Q}=\mathcal{Q}^{\prime} . Proof. Since W_{\mathcal{Q}} has finite index in G_{\mathcal{Q}}^{\mathcal{P}} , the group G_{\mathcal{Q}}^{\mathcal{P}} i topologically finitely generated if and only if W_{\mathcal{Q}} is. For the first part, consider the map \displaystyle\varphi_{1}:W_{\mathcal{Q}}\rightarrow\prod_{n\in\mathbb{N}}\mathcal{Q}/\mathcal{Q}^{\prime},\quad g\mapsto\prod_{n\in\mathbb{N}}\left(\prod_{v\in\mathcal{L}_{n}}(g|_{v}^{1})\mathcal{Q}^{\prime}\right), where the outer product is the Cartesian product and the inner product is the product of the sections. The map is well-defined since we are taking the quotient by \mathcal{Q}^{\prime} . The map is clearly surjective. If \mathcal{Q}\neq\mathcal{Q}^{\prime} , the image is not topologically finitely generated, and thus W_{\mathcal{Q}} is not topologically finitely generated. Suppose there exists x\in X such that q(x)=x for all q\in\mathcal{Q} . Consider the map \displaystyle\varphi_{2}:W_{\mathcal{Q}}\rightarrow\prod_{n\in\mathbb{N}}\mathcal{Q},\quad g\mapsto\prod_{n\in\mathbb{N}}\left(g|_{x^{n}}^{1}\right), where x^{n} represents the vertex at level n reached by following the path labeled by x at each level. The map is a morphism since g(x^{n})=x^{n} for all g\in W_{\mathcal{Q}} . The map is surjective, and since \mathcal{Q}\neq 1 , the image is not topologically finitely generated. Hence, W_{\mathcal{Q}} is not topologically finitely generated. If \mathcal{Q} is transitive, the result follows directly from ( RadiFiniteType2025 , Corollary 3) or ( Bondarenko2010 , Theorem 1) ∎ Remark 4.5 . A profinite group is just-infinite if for every closed non-trivial normal subgroup N\lhd G , then [G:N]<\infty . A profinite group is strongly complete if its profinite completion coincides with the group. By a recent result proved in ( RadiFiniteType2025 , Theorem 1) , which applies to all the groups of the form G_{\mathcal{Q}}^{\mathcal{P}} , the following properties are equivalent: • G_{\mathcal{Q}}^{\mathcal{P}} is topologically finitely generated • G_{\mathcal{Q}}^{\mathcal{P}} is just-infinite • G_{\mathcal{Q}}^{\mathcal{P}} is strongly complete • \mathcal{Q}=\mathcal{Q}^{\prime} We conclude this section by showing that the fixed-point proportion of the group G_{\mathcal{Q}}^{\mathcal{P}} can be calculated explicitly in terms of its cosets: Proposition 4.6 . Let T be a d -regular tree with d\geq 2 and 1\neq\mathcal{Q}\lhd\mathcal{P}\leq\Sym(d) . Then (11) \displaystyle\FPP(G_{\mathcal{Q}}^{\mathcal{P}})=\frac{1}{[\mathcal{P}:\mathcal{Q}]}\sum_{A\in\mathcal{P}/\mathcal{Q}}\FPP(W_{A}). In particular, \FPP(G_{\mathcal{Q}}^{\mathcal{P}})>0 if and only \FPP(W_{A})>0 for one coset A . Proof. Using the map in Equation 10 , we deduce that cosets of G_{\mathcal{Q}}^{\mathcal{P}} by W_{\mathcal{Q}} are of the form W_{A}:={\left\{g\in\Aut(T):g|_{v}^{1}\in A,\forall v\in T\right\}}, where A is a coset of \mathcal{P}/\mathcal{Q} . Therefore, if A,A^{\prime} are two different cosets of \mathcal{P}/\mathcal{Q} , then \pi_{n}(W_{A})\cap\pi_{n}(W_{A^{\prime}})=\emptyset for all n\geq 1 , and \bigcup_{A\in\mathcal{P}/\mathcal{Q}}\pi_{n}(W_{A})=\pi_{n}(G_{\mathcal{Q}}^{\mathcal{P}}). This allows us to split the calculation of \FPP(G_{\mathcal{Q}}^{\mathcal{P}}) as: \displaystyle\FPP(G_{\mathcal{Q}}^{\mathcal{P}})=\lim_{n\rightarrow+\infty}\frac{\#{\left\{g\in\pi_{n}(G_{\mathcal{Q}}^{\mathcal{P}}):\text{$g$ fixes a vertex in $\mathcal{L}_{n}$}\right\}}}{\left|{\pi_{n}(G_{\mathcal{Q}}^{\mathcal{P}})}\right|}= \displaystyle\sum_{A\in\mathcal{P}/\mathcal{Q}}\lim_{n\rightarrow+\infty}\frac{1}{[\pi_{n}(G_{\mathcal{Q}}^{\mathcal{P}}):\pi_{n}(W_{\mathcal{Q}})]}\,\frac{\#{\left\{g\in\pi_{n}(W_{A}):\text{$g$ fixes a vertex in $\mathcal{L}_{n}$}\right\}}}{\left|{\pi_{n}(W_{\mathcal{Q}})}\right|}. By 4.1 , we have [\mathcal{P}:\mathcal{Q}]=[\pi_{n}(G_{\mathcal{Q}}^{\mathcal{P}}):\pi_{n}(W_{\mathcal{Q}})] and \left|{\pi_{n}(W_{\mathcal{Q}})}\right|=\left|{\pi_{n}(W_{\mathcal{A}})}\right| for any coset in \mathcal{P}/\mathcal{Q} . Therefore \displaystyle\FPP(G_{\mathcal{Q}}^{\mathcal{P}})=\frac{1}{[\mathcal{P}:\mathcal{Q}]}\sum_{A\in\mathcal{P}/\mathcal{Q}}\lim_{n\rightarrow+\infty}\frac{\#{\left\{g\in\pi_{n}(W_{A}):\text{$g$ fixes a vertex in $\mathcal{L}_{n}$}\right\}}}{\left|{\pi_{n}(W_{A})}\right|}. The last limit corresponds to our extension of the definition of the fixed-point proportion of sets made in Equation 6 . We conclude that \displaystyle\FPP(G_{\mathcal{Q}}^{\mathcal{P}})=\frac{1}{[\mathcal{P}:\mathcal{Q}]}\sum_{A\in\mathcal{P}/\mathcal{Q}}\FPP(W_{A}). ∎ 5. Examples In this subsection, we calculate the fixed-point proportion for explicit cases of level-transitive groups G_{\mathcal{Q}}^{\mathcal{P}} . By 4.6 , we have \FPP(G_{\mathcal{Q}}^{\mathcal{P}})=\frac{1}{[\mathcal{P}:\mathcal{Q}]}\sum_{A\in\mathcal{P}/\mathcal{Q}}\FPP(W_{A}), where each A is a coset of \mathcal{P}/\mathcal{Q} . By 4.3 , the subgroup \mathcal{Q} must act transitively on X={\left\{1,\dots,d\right\}} . Since \mathcal{Q} is transitive, by 3.11 .(1) applied to \mathcal{P}\curvearrowright X , we have f_{A}^{\prime}(0)=\frac{1}{\#A}\sum_{a\in A}\#X^{a}=1. Therefore, by Theorem 3.8 .(3), we have only two possibilities: \displaystyle\FPP(W_{A})=\left\{\begin{matrix}1&\mbox{if every element in $A$ fixes exactly one point in $X$,}\\ 0&\mbox{otherwise.}\end{matrix}\right. In particular \FPP(W_{\mathcal{Q}})=0 since the identity is in \mathcal{Q} . 5.1. Construction 1 Seeing X={\left\{1,\dots,d\right\}} as \mathbb{Z}/d\mathbb{Z} , the integer numbers modulo d , we define \mathcal{P}={\left\{x\mapsto ax+b:a\in I,b\in\mathbb{Z}/d\mathbb{Z}\right\}}, where I\leq(\mathbb{Z}/d\mathbb{Z})^{\times} and \mathcal{Q}={\left\{x\mapsto x+b:b\in\mathbb{Z}/d\mathbb{Z}\right\}}. Here, \mathcal{Q}\lhd\mathcal{P} , and \mathcal{Q} is transitive on X . By 4.4 , the group G_{\mathcal{Q}}^{\mathcal{P}} is not topologically finitely generated. The cosets of \mathcal{P}/\mathcal{Q} are the sets A_{a}={\left\{x\mapsto ax+b:b\in\mathbb{Z}/d\mathbb{Z}\right\}}. To determine how many fixed points an element of the form ax+b has, we use the following lemma: Lemma 5.1 . Let n\in\mathbb{N}_{>0} and \alpha,\beta\in\mathbb{Z}/n\mathbb{Z} . Then, the equation \alpha x\equiv\beta\pmod{n} has a solution if and only if \gcd(\alpha,n)\mid\beta and the number of solutions is \gcd(\alpha,n) . Applying 5.1 with (\alpha,\beta,n)=(a-1,-b,d) we conclude \displaystyle\FPP(W_{A_{a}})=\left\{\begin{matrix}1&\mbox{if $\gcd(a-1,d)=1$,}\\ 0&\mbox{$\gcd(a-1,d)>1$.}\end{matrix}\right. Thus, (12) \displaystyle\FPP(G_{\mathcal{Q}}^{\mathcal{P}})=\frac{\#{\left\{a\in I:a-1\in(\mathbb{Z}/d\mathbb{Z})^{\times}\right\}}}{\Phi(d)}, where \Phi(d) is the Euler’s totient function. In the particular case where I=(\mathbb{Z}/d\mathbb{Z})^{\times} , we can write Equation 12 in a more compact form. This case will be of particular interest in Section 6 since it corresponds to the Galois group G_{\infty}(\mathbb{Q},x^{d}+c,t) . Proposition 5.2 . Define \psi:\mathbb{N}\rightarrow\mathbb{N} the function \psi(d)=\#{\left\{a\in(\mathbb{Z}/d\mathbb{Z})^{\times}:a-1\in(\mathbb{Z}/d\mathbb{Z})^{\times}\right\}}. Then \psi(d)=d\prod_{p\mid d}\left(1-\frac{2}{p}\right). In particular, \frac{\psi(d)}{\Phi(d)}=\prod_{p\mid d}\frac{p-2}{p-1}. Proof. We start proving that \psi is multiplicative, namely, if m,n\in\mathbb{N} are coprime, then \psi(mn)=\psi(m)\psi(n) . Indeed, by Chinese remainder theorem, the map \displaystyle\varphi:\mathbb{Z}/mn\mathbb{Z} \displaystyle\rightarrow(\mathbb{Z}/m\mathbb{Z})\times(\mathbb{Z}/n\mathbb{Z}) \displaystyle x+mn\mathbb{Z} \displaystyle\mapsto(x+m\mathbb{Z},x+n\mathbb{Z}) is a ring isomorphism. Thus, 1 is sent to (1,1) and units are sent to units. So a-1 is a unit in \mathbb{Z}/mn\mathbb{Z} if and only if \varphi(a)-1=(a_{1}-1,a_{2}-1) is a unit in \mathbb{Z}/m\mathbb{Z}\times\mathbb{Z}/n\mathbb{Z} , meaning that both a_{1}-1 and a_{2}-1 are units in their respective rings. Therefore, \psi(mn)=\psi(m)\psi(n) . We now compute \psi(p^{\alpha}) for p a prime number. The elements that are not units are of the form pk with k=1,\dots,p^{\alpha-1} , so \psi(p^{\alpha})=p^{\alpha-1}(p-2) . Combining everything, if d=\prod_{i=1}^{r}p^{\alpha_{i}} , then \psi(d)=\prod_{i=1}^{r}p^{\alpha_{i}-1}(p-2)=d\prod_{i=1}^{r}\left(1-\frac{2}{p}\right). Finally, since \Phi(d)=d\prod_{i=1}^{r}\left(1-\frac{1}{p}\right) , we obtain \frac{\psi(d)}{\Phi(d)}=\prod_{p\mid d}\frac{p-2}{p-1}.\qed Thus, in the case where I=(\mathbb{Z}/d\mathbb{Z})^{\times} , we have \FPP(G_{\mathcal{Q}}^{\mathcal{P}})=0 if and only if d is even, since when d is odd we have that 2 and 2-1=1 are invertible modulo \mathbb{Z}/d\mathbb{Z} . 5.1.1. Construction 2 Write \mathcal{P}=\mathcal{Q}\rtimes\Aut(\mathcal{Q}) with the operation (q,h)(q,h^{\prime})=(q\,h(q^{\prime}),h\,h^{\prime}). In particular (q,h)^{-1}=(h^{-1}(q^{-1}),h^{-1}) . The group \mathcal{P} acts on \mathcal{P}/\Aut(\mathcal{Q}) by left translation. A set of representatives for \mathcal{P}/\Aut(\mathcal{Q}) is {\left\{(q,1)\right\}}_{q\in\mathcal{Q}} . The action has the following properties: • The action is faithful: Let (q,h)\in\mathcal{P} such that (q,h)(q^{\prime},1)\equiv(q^{\prime},1)\pmod{\Aut(\mathcal{Q})} for all q^{\prime}\in\mathcal{Q} . Since (q,h)(q^{\prime},1)=(q\,h(q^{\prime}),h) , this means that q\,h(q^{\prime})=q^{\prime} for every q^{\prime}\in\mathcal{Q} . If q^{\prime}=1 , then this implies that q=1 and therefore h(q^{\prime})=q^{\prime} for all q^{\prime} , so h is the identity. Thus, the action is faithful, and \mathcal{P} can be seen as a subgroup of \Sym(\left|{\mathcal{Q}}\right|) . • \mathcal{Q} acts transitively: Given q_{1},q_{2}\in\mathcal{Q} , then (q_{2}\,q_{1}^{-1},1)(q_{1},1)=(q_{2},1) . The cosets of \mathcal{P}/\mathcal{Q} have the form A_{h}={\left\{(q,h):q\in\mathcal{Q}\right\}}. An element (q,h) fixes exactly one point in \mathcal{P}/\Aut(\mathcal{Q}) if and only if there exists a unique q^{\prime}\in\mathcal{Q} such that (q,h)(q^{\prime},1)=(q\,h(q^{\prime}),h)\equiv(q^{\prime},1)\pmod{\Aut(\mathcal{Q})}, namely, q\,h(q^{\prime})=q^{\prime} has a unique solution. Given d\geq 2 , write d=2^{n}r with r odd and let \mathcal{Q}=C_{2}^{n}\times C_{r} , where C_{m} is the cyclic group of order m . For this choice, \Aut(\mathcal{Q})\simeq\GL_{n}(\mathbb{F}_{2})\times(\mathbb{Z}/r\mathbb{Z})^{*} , where \mathbb{F}_{2} is the field with two elements. Elements q\in\mathcal{Q} can be written as (\vec{x},z) , where \vec{x}\in\mathbb{F}_{2}^{n} and z\in\mathbb{Z}/r\mathbb{Z} , and elements h\in\Aut(\mathcal{Q}) as (A,\alpha) , where A\in\GL_{n}(\mathbb{F}_{2}) and \alpha\in(\mathbb{Z}/r\mathbb{Z})^{*} . Since the group \mathcal{Q} is abelian, we can use additive notation. Then (q,h) has a unique fixed point in \mathcal{P}/\Aut(\mathcal{Q}) if and only if there exists a unique q^{\prime}=(\vec{x^{\prime}},z^{\prime}) such that q+h(q^{\prime})=q^{\prime} , namely, \displaystyle\left\{\begin{matrix}\vec{x}+A\vec{x^{\prime}}=\vec{x^{\prime}}\\ z+\alpha z^{\prime}=z^{\prime},\end{matrix}\right. or equivalently \displaystyle\left\{\begin{matrix}(A-1)\vec{x^{\prime}}=-\vec{x}\\ (\alpha-1)z^{\prime}=-z.\end{matrix}\right. This will have a unique solution if and only if A-1 and \alpha-1 are invertible in \GL_{n}(\mathbb{F}_{2}) and (\mathbb{Z}/r\mathbb{Z})^{*} , respectively, and this is independent of q=(\vec{x},z) . Therefore \FPP(G_{\mathcal{Q}}^{\mathcal{P}})=\frac{\#{\left\{A\in\GL_{n}(\mathbb{F}_{2}):A-1\in\GL_{n}(\mathbb{F}_{2})\right\}}}{\left|{\GL_{n}(\mathbb{F}_{2})}\right|}\,\prod_{p\mid r}\frac{p-2}{p-1}. Notice that this is a generalization of construction 1 since when d is odd, the group {\left\{x\mapsto x+b:b\in\mathbb{Z}/d\mathbb{Z}\right\}} is isomorphic to C_{d} and {\left\{x\mapsto ax+b:a\in(\mathbb{Z}/d\mathbb{Z})^{*},b\in\mathbb{Z}/d\mathbb{Z}\right\}} is isomorphic to C_{d}\rtimes\Aut(C_{d})=C_{d}\rtimes(\mathbb{Z}/d\mathbb{Z})^{*} . The issue caused by 2 when d is even in construction 1 is, in this case, isolated. To prove that this construction of G_{\mathcal{Q}}^{\mathcal{P}} is an example of a group with positive fixed-point proportion, we still need to find: (1) a number in \mathbb{Z}/r\mathbb{Z} such that \alpha and \alpha-1 are invertible with r odd and (2) a matrix in \GL_{n}(\mathbb{F}_{2}) such that A and A-1 are both invertible. In the case of \alpha , since r is odd then \alpha=2 satisfies the condition. For the matrices, define A_{2}:=\left(\begin{matrix}1&1\\ 1&0\end{matrix}\right) and A_{3}:=\left(\begin{matrix}1&1&1\\ 1&1&0\\ 1&0&0\end{matrix}\right). Since the rows are linearly independent, they are invertible. Moreover, A_{2}-1=\left(\begin{matrix}0&1\\ 1&1\end{matrix}\right) and A_{3}-1=\left(\begin{matrix}0&1&1\\ 1&0&0\\ 1&0&1\end{matrix}\right), which are also invertible. Therefore, we have examples for n=2 and n=3 . If n>3 , consider two cases. If n is even, define A_{n}:=\left(\begin{matrix}A_{2}&&&\\ &A_{2}&&\\ &&\ddots&\\ &&&A_{2}\end{matrix}\right) and if n is odd, A_{n}:=\left(\begin{matrix}A_{2}&&&&\\ &A_{2}&&&\\ &&\ddots&&\\ &&&A_{2}&\\ &&&&A_{3}\end{matrix}\right) By induction and the property of the determinant of subblocks, then A_{n} and A_{n}-1 are invertible. The only case that cannot be addressed with this construction is when n=1 , or equivalently d\equiv 2\pmod{4} because \GL_{1}(\mathbb{F}_{2}) has only one invertible element. Table 1 shows the values of \#{\left\{A\in\GL_{n}(\mathbb{F}_{2}):A-1\in\GL_{n}(\mathbb{F}_{2})\right\}} and \frac{\#{\left\{A\in\GL_{n}(\mathbb{F}_{2}):A-1\in\GL_{n}(\mathbb{F}_{2})\right\}}}{\left|{\GL_{n}(\mathbb{F}_{2})}\right|} for the first cases of n . Combined with 5.2 , this allows us to calculate the fixed-point proportion for several cases of this construction. n \#{\left\{A\in\GL_{2}(\mathbb{F}_{n}):A-1\in\GL_{n}(\mathbb{F}_{2})\right\}} \left|{\GL_{n}(\mathbb{F}_{2})}\right| \frac{\#{\left\{A\in\GL_{n}(\mathbb{F}_{2}):A-1\in\GL_{n}(\mathbb{F}_{2})\right\}}}{\left|{\GL_{n}(\mathbb{F}_{2})}\right|} 1 0 1 0 2 2 6 1/3\approx 0.333 3 48 168 2/7\approx 0.2857\dots 4 5824 20160 13/45\approx 0.288\dots 5 2887680 9999360 188/651\approx 0.2887\dots Table 1. Number of matrices A\in\GL_{n}(\mathbb{F}_{2}) such that A-I is also invertible. There are other groups \mathcal{Q} similar to C_{2}^{n}\times C_{r} with r odd that can be defined, and their fixed-point proportions will have a similar formula. For example, we can consider \mathcal{Q}=\prod_{p\mid d}C_{p}^{e_{p}} where p is a prime number and e_{p} is the largest natural number such that p^{e_{p}}\mid d . 5.1.2. The problem of d\equiv 2\pmod{4} Using GAP GAP , all possible transitive groups \mathcal{Q}\leq\Sym(d) were computed for d\equiv 2\pmod{4} with d ranging from 2 to 30 . Since \mathcal{Q} must be normal in \mathcal{P} , we take \mathcal{P} the normalizer of \mathcal{Q} in \Sym(d) , since this corresponds to the most favorable scenario. However, for all tested values of d , no transitive group \mathcal{Q} as subgroup of \mathcal{P} had a coset where every element in the coset fixed exactly one element. This suggests an obstruction for these values of d to admit a subgroup \mathcal{Q} meeting the required conditions. The obstruction aligns with the special case considered in Construction 2, where \mathcal{Q} was abelian of order d . The specific structure of \mathcal{Q} in this case explains why the conditions fails for d\equiv 2\pmod{4} . 6. Application to iterated Galois groups In this subsection, we will demonstrate that Construction 1 in Section 5 corresponds to the iterated Galois group G_{\infty}(K,x^{d}+c,t) where K is any field, t is transcendental over K and c\in K^{\times} such that 0 is wandering for f (namely, f^{n}(0)\neq f^{m}(0) for all n\neq m ). Proof of 1 . We start examining the solutions of x^{d}+c=t . These solutions are of the form \xi^{x_{1}}\sqrt[d]{t-c} where \xi is a primitive d -th root of unity, and x_{1}=1,\dots,d . Denote these solutions by \alpha_{x_{1}} . On the second level of the tree, we consider the preimages of each \alpha_{x_{1}} . These preimages are given by \alpha_{x_{1}x_{2}}=\xi^{x_{2}}\sqrt[d]{\alpha_{x_{1}}-c}. Continuing in this manner, at level n , the solutions are of the form \alpha_{x_{1}\dots x_{n}} , where \alpha_{x_{1}\dots x_{n}}=\xi^{x_{n}}\alpha_{x_{1}\dots x_{n-1}d} and \alpha_{x_{1}\dots x_{n-1}d}=\sqrt[d]{\alpha_{x_{1}\dots x_{n-1}}-c}. In the d -regular tree, we place \alpha_{x_{1}\dots x_{n}} in the n -th level of the tree, ordered by lexicographical order. Notice that all preimages in the same level are different by 2.12 . To calculate the Galois group G_{\infty}(K,x^{d}+c,t) , we rely on the following theorem: Theorem 6.1 ( ( Juul2014 , Theorem 3.1) ) . Let k be a field and f\in k(x) a rational function with degree d\geq 2 such that f^{\prime}\neq 0 . Let t be transcendental over k , write L=k(t) and L_{\infty}=L\left(\bigcup_{n\in\mathbb{N}}f^{-n}(t)\right) . Let C denote the set of critical points of f , write F=L(f^{-1}(t))\cap\overline{k} and \mathcal{Q}=\Gal(L(f^{-1}(t))/F(t)) . If for any a,b\in C and m,n\in\mathbb{N} we have f^{m}(a)=f^{n}(b) if and only if m=n and a=b , then \Gal(L_{\infty}/F(t))\simeq W_{\mathcal{Q}} , where W_{\mathcal{Q}} is the iterated wreath product of \mathcal{Q} . In our case, the polynomial f has only 0 as its critical point, and the orbit of 0 is wandering by the choice of c . Thus, \Gal(K_{\infty}(f)/F(t))=W_{\mathcal{Q}} , where F=K_{1}(f)\cap\overline{K}=K(\xi) , and \mathcal{Q}=\Gal(K_{1}(f)/F(t))=\Gal(F(\alpha_{d},t)/F(t)). If \sigma\in\mathcal{Q} , there exists b\in\mathbb{Z}/d\mathbb{Z} such that \sigma(\alpha_{d})=\alpha_{b} and then \sigma(\alpha_{x})=\sigma(\xi^{x}\alpha_{d})=\xi^{x}\alpha_{b}=\alpha_{x+b}. This implies that \mathcal{Q}\simeq{\left\{x\mapsto x+b:b\in\mathbb{Z}/d\mathbb{Z}\right\}} . Now, consider the monomorphism of groups \varphi:\Gal(K(\xi)/K)\rightarrow(\mathbb{Z}/d\mathbb{Z})^{\times} given by \sigma\mapsto a such that \sigma(\xi)=\xi^{a} and write I=\IIm(\varphi) . If g\in G_{\infty}(K,x^{d}+c,t) , then there exists a\in I such that \sigma(\xi)=\xi^{a} and given x_{1},\dots,x_{n-1}\in\mathbb{Z}/d\mathbb{Z} , there exist y_{1},\dots,y_{n}\in\mathbb{Z}/d\mathbb{Z} such that \sigma(\alpha_{x_{1}\dots x_{n-1}d})=\alpha_{y_{1}\dots y_{n}} . Then, if x_{n}\in\mathbb{Z}/d\mathbb{Z} , \displaystyle g(\alpha_{x_{1}\dots x_{n-1}x_{n}})=g(\xi^{x_{n}}\alpha_{x_{1}\dots x_{n-1}d})=g(\xi)^{x_{n}}\alpha_{y_{1}\dots y_{n}}= \displaystyle\xi^{ax_{n}}\alpha_{y_{1}\dots y_{n}}=\alpha_{y_{1}\dots y_{n-1}(ax_{n}+y_{n})}. This shows that the label g|_{x_{1}\dots x_{n-1}}^{1}=x\mapsto ax+y_{n} . Notice that the slope of the linear action of the label does not depend on the vertex and it only depends on the element g . Therefore, if we call \mathcal{P}={\left\{x\mapsto ax+b:a\in I,b\in\mathbb{Z}/d\mathbb{Z}\right\}} , we obtain that G_{\infty}(K,x^{d}+c,t)\leq G_{\mathcal{Q}}^{\mathcal{P}} . If we take the quotient G_{\infty}(K,x^{d}+c,t)/W_{\mathcal{Q}} , we have \displaystyle G_{\infty}(K,x^{d}+c,t)/W_{\mathcal{Q}}=G_{\infty}(K,x^{d}+c,t)/\Gal(K_{\infty}(f)/F(t)) \displaystyle\simeq\Gal(K(\xi,t)/K(t))\simeq\Gal(K(\xi)/K(t))\simeq I and so [G_{\infty}(K,x^{d}+c,t):W_{\mathcal{Q}}]=\left|{I}\right|=\left|{\mathcal{P}/\mathcal{Q}}\right|=[G_{\mathcal{Q}}^{\mathcal{P}}:W_{\mathcal{Q}}]. Thus, G_{\infty}(\mathbb{Q},f,t)=G_{\mathcal{Q}}^{\mathcal{P}} and \FPP(G_{\infty}(\mathbb{Q},f,t)) is given by Equation 12 . ∎ 7. Open questions We conclude this article by posing the following questions, which are considered by the author to be relevant and interesting for a better understanding of the theory of the fixed-point proportion of groups acting on trees: (1) 2.11 answers the question about the continuity of the function \FPP . The next question is about the surjectivity of \FPP in [0,1] : Question 7.1 . Fix d\in\mathbb{N} and T a d -regular tree. Given \alpha\in[0,1] , can we find a group G\leq\Aut(T) such that \FPP(G)=\alpha ? Does the same hold if we restrict to topologically finitely generated groups? If we restrict to level-transitive groups, the answer to 7.1 is no. Indeed, as the fixed-point proportion has to be smaller than the proportion on the first level by the proof of 2.10 , if F_{1} denotes the set of elements in \pi_{1}(G) fixing at least one vertex in \mathcal{L}_{1} , by Burnside’s lemma, \displaystyle 1=\frac{1}{\left|{\pi_{1}(G)}\right|}\sum_{g\in\pi_{1}(G)}\#(\mathcal{L}_{1}^{g})=\frac{1}{\left|{\pi_{1}(G)}\right|}\sum_{g\in F_{1}}\#(\mathcal{L}_{1}^{g}) \displaystyle=\frac{1}{\left|{\pi_{1}(G)}\right|}\left(d+\sum_{g\in F_{1}\setminus{\left\{1\right\}}}\#(\mathcal{L}_{1}^{g})\right)\geq\frac{1}{\left|{\pi_{1}(G)}\right|}\left(d+\#F_{1}-1\right) \displaystyle=\frac{d-1}{\left|{\pi_{1}(G)}\right|}+\frac{\#F_{1}}{\left|{\pi_{1}(G)}\right|} Solving for the last term and using that \pi_{1}(G)\leq\Sym(d) , \displaystyle\FPP(G)\leq\frac{\#F_{1}}{\pi_{1}(G)}=1-\frac{d-1}{\left|{\pi_{1}(G)}\right|}\leq 1-\frac{d-1}{d!}. For example, when d=2 , we obtain that no level-transitive group can have fixed-point proportion bigger than 1/2 . Moreover, doing the same argument in the second level and inspecting the possible transitive groups \pi_{2}(G) , it can be seen that for level-transitive groups in the binary tree, the fixed-point proportion cannot be bigger than 3/8 . This leads us to rephrase 7.1 in the case of the family of level-transitive groups: Question 7.2 . Fix d\in\mathbb{N} and T a d -regular tree. What is the maximum fixed-point proportion for level-transitive groups acting on T ? Let \alpha_{d} be this maximum. Given \alpha\in[0,\alpha_{d}] , does there exist G a level-transitive group of \Aut(T) such that \FPP(G)=\alpha ? (2) Finally, we wonder about the behavior of the fixed-point proportion when we take random subgroups. Question 7.3 . Let T be a spherically homogeneous tree, n a positive integer, and G\leq\Aut(T) a closed subgroup with normalized Haar measure \mu and \FPP(G)=0 . Consider G^{n} as a probability space equipped with the product measure. Select n random elements of G and let H the subgroup they generate. Does \FPP(H)=0 with probability 1 ?