Profinite closures of the iterated monodromy groups associated with quadratic polynomials

In this paper we describe the profinite closure of the iterated monodromy groups arising from the arbitrary post-critically finite quadratic polynomial.

Збережено в:
Бібліографічні деталі
Дата:2017
Автор: Samoilovych, I.
Формат: Стаття
Мова:English
Опубліковано: Інститут прикладної математики і механіки НАН України 2017
Назва видання:Algebra and Discrete Mathematics
Онлайн доступ:http://dspace.nbuv.gov.ua/handle/123456789/156027
Теги: Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Цитувати:Profinite closures of the iterated monodromy groups associated with quadratic polynomials / I. Samoilovych // Algebra and Discrete Mathematics. — 2017. — Vol. 23, № 2. — С. 285-304. — Бібліогр.: 8 назв. — англ.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id irk-123456789-156027
record_format dspace
spelling irk-123456789-1560272019-06-18T01:31:19Z Profinite closures of the iterated monodromy groups associated with quadratic polynomials Samoilovych, I. In this paper we describe the profinite closure of the iterated monodromy groups arising from the arbitrary post-critically finite quadratic polynomial. 2017 Article Profinite closures of the iterated monodromy groups associated with quadratic polynomials / I. Samoilovych // Algebra and Discrete Mathematics. — 2017. — Vol. 23, № 2. — С. 285-304. — Бібліогр.: 8 назв. — англ. 1726-3255 2010 MSC:20E08, 20E18, 37F10. http://dspace.nbuv.gov.ua/handle/123456789/156027 en Algebra and Discrete Mathematics Інститут прикладної математики і механіки НАН України
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
language English
description In this paper we describe the profinite closure of the iterated monodromy groups arising from the arbitrary post-critically finite quadratic polynomial.
format Article
author Samoilovych, I.
spellingShingle Samoilovych, I.
Profinite closures of the iterated monodromy groups associated with quadratic polynomials
Algebra and Discrete Mathematics
author_facet Samoilovych, I.
author_sort Samoilovych, I.
title Profinite closures of the iterated monodromy groups associated with quadratic polynomials
title_short Profinite closures of the iterated monodromy groups associated with quadratic polynomials
title_full Profinite closures of the iterated monodromy groups associated with quadratic polynomials
title_fullStr Profinite closures of the iterated monodromy groups associated with quadratic polynomials
title_full_unstemmed Profinite closures of the iterated monodromy groups associated with quadratic polynomials
title_sort profinite closures of the iterated monodromy groups associated with quadratic polynomials
publisher Інститут прикладної математики і механіки НАН України
publishDate 2017
url http://dspace.nbuv.gov.ua/handle/123456789/156027
citation_txt Profinite closures of the iterated monodromy groups associated with quadratic polynomials / I. Samoilovych // Algebra and Discrete Mathematics. — 2017. — Vol. 23, № 2. — С. 285-304. — Бібліогр.: 8 назв. — англ.
series Algebra and Discrete Mathematics
work_keys_str_mv AT samoilovychi profiniteclosuresoftheiteratedmonodromygroupsassociatedwithquadraticpolynomials
first_indexed 2025-07-14T08:17:08Z
last_indexed 2025-07-14T08:17:08Z
_version_ 1837610090486562816
fulltext Algebra and Discrete Mathematics RESEARCH ARTICLE Volume 23 (2017). Number 2, pp. 285–304 c© Journal “Algebra and Discrete Mathematics” Profinite closures of the iterated monodromy groups associated with quadratic polynomials Ihor Samoilovych Communicated by R. I. Grigorchuk Abstract. In this paper we describe the profinite closure of the iterated monodromy groups arising from the arbitrary post- critically finite quadratic polynomial. Introduction In this article we consider the iterated monodromy groups arising from the post-critically finite (i.e. orbit of the critical point is finite) quadratic polynomial z2 − c and we denote such group as IMG(z2 − c). Group IMG(z2 − c) could be represented as the automorphism group of the rooted binary tree (see [1] or [6]). These groups were studied by Bartholdi and Nekrashevych [1]. They show that properties of such group highly depend on whether the orbit of the critical point of z2 − c is periodic (the orbit of the critical point contain critical point itself) or strictly pre-periodic (the orbit of the critical point do not contain critical point). Bartholdi and Nekrashevych [1] have proved that group IMG(z2 − c) is weakly branch on its commutator for the periodic case and also they have proved that group IMG(z2− c) is regular branch for the pre-periodic case (except for the infinite dihedral group). 2010 MSC: 20E08, 20E18, 37F10. Key words and phrases: iterated monodromy groups, quadratic polynomials, profinite grous, profinite limits, group acting on trees. 286 Profinite closures of monodromy groups One interesting example of the iterated monodromy group is Basilica group [5] arising from the complex polynomial z2 − 1. This group is gene- rated by a three-state automaton. It is the first example of an amenable group not belonging to the class of sub-exponentially amenable groups [3]. Closure of some iterated monodromy groups of post-critically finite quadratic polynomials z2 − c was studied by Pink [7]. He studied the Hausdorff dimension, the maximal abelian factor groups, and the norma- lizers. Also he showed that the closure of IMG(z2− c) don’t depend up to conjugacy in AutT on words that define the automorphism group which is associated with IMG(z2 − c) (we improve this result and we show that in many cases for different words we get the same closure of IMG(z2 − c), see Corollary 1 and Theorem 4). In this article, we provide the description of the closure for the group IMG(z2− c). We prove that for the pre-periodic case closure of the group IMG(z2 − c) is a self-similar group of finite type except for the infinite dihedral group. We use another approach compare to Pink [7] that allow us to improve some of his results. 1. Preliminaries Denote by T a rooted binary tree. We can denote vertices of the tree T as words on the alphabet {0, 1}. The empty word ∅ corresponds to the root of the tree T ; edges are (v, vx) for all v ∈ T and x ∈ {0, 1}. Words 1 . . . 1︸ ︷︷ ︸ n and 0 . . . 0︸ ︷︷ ︸ n we denote by 1n and 0n respectively. The set of all words of length n we call as n-th level of the tree T and we denote this set by Tn, n > 0. Let 0Tn denote the set of all words of length n+ 1 that starts with 0 and let 1Tn denote the set of all words of length n+ 1 that starts with 1. The subtree of T of all vertices of the first n levels is denoted by T [n]. The group of all automorphisms of the tree T we denote by AutT . We denote trivial element in AutT by the symbol 1. For an automorphism g ∈ AutT and a vertex v ∈ T we denote by g(v) such action on the tree T that g(vw) = g(v)g(v)(w) for all w ∈ T . A stabilizer of the n-th level of G < AutT , StG(n) (or just St(n)) is defined to be a set of all elements in G which fix all vertices of the level n. Group G|T [n] is a restriction of the group G to the subtree T [n] (i.e. if g ∈ G|T [n] then g ∈ AutT [n] and there exists such h ∈ G that for all I . Samoilovych 287 v ∈ T [n] equality g(v) = h(v) holds). For the element g ∈ G we denote by g|T [n] ∈ G|T [n] the restriction of the element g to the subtree T [n]. The group G = {g ∈ AutT | g(v)|T [n] ∈ F for every v ∈ T} called as self-similar group of finite type with depth n given by the pattern group F < AutT [n]. If self-similar group of finite type with depth n can’t be given by any pattern group F < AutT [n−1] then n is minimal pattern depth. We know [2] that AutT ∼= ≀∞i=1 Sym({0, 1}) where Sym({0, 1}) = {e, σ} is a symmetric group on {0, 1}. Thus AutT is a profinite group AutT = lim←− ≀ n i=1 Sym({0, 1}). Also we know [2] that AutT ∼= AutT ≀ Sym({0, 1}) therefore every element g ∈ AutT we can write by the following wreath recursion g = (g(0), g(1))σ i where i ∈ {0, 1}. Product of elements g = (g1, g2)σi and h = (h1, h2)σj is defined as follows gh = { (g1h1, g2h2)σj , if i = 0, (g1h2, g2h1)σ1−j , otherwise. To describe profinite closure of the group G it is enough to describe all G|T [n] for n > 1. In this article we describe G|T [n] for n > 2 recursively i.e. for g0, g1 ∈ G|T [n−1] and i ∈ {0, 1} we determine necessary and sufficient condition for g = (g0, g1)σi ∈ G|T [n] . Let us fix some integer n > 1 and automorphism g = (g0, g1)σi∅ ∈ AutT , where i∅ ∈ {0, 1}. We define following functions Ln(g) = g(00...00) · g(00...01) · . . . · g(0j2...jn−1jn) · . . . · g(01...11), Rn(g) = g(10...00) · g(10...01) · . . . · g(1j2...jn−1jn) · . . . · g(11...11), ln,0(g) = rn,0(g) = l0(g) = r0(g) = i∅, ln,i(g) = ln,i−1(Ln(g)), for i > 0, rn,i(g) = rn,i−1(Rn(g)), for i > 0. 288 Profinite closures of monodromy groups We can define functions ln,i and rn,i for the element g ∈ AutT [k] where k > ni+ 1 in the same way as above. We use notation gh for h−1gh where g and h are elements of a group. For real number x denote by ⌊x⌋ the largest integer not greater than x and denote by ⌈x⌉ the smallest integer not less than x. 2. Periodic case We consider profinite closures of iterated monodromy groups associated with quadratic polynomials with periodic critical points in this section. Let us fix an integer t > 1 and a word x = x1 . . . xt over the alphabet {0, 1}. We define elements a1, ..., at ∈ AutT by recursions a1 = { (at, 1)σ, if x1 = 0, (1, at)σ, otherwise. and for 2 6 k 6 t ak = { (ak−1, 1), if xk = 0, (1, ak−1), otherwise. We will study the group Gt,x = 〈a1, . . . , at〉 and its profinite closure Gt,x. We consider case x1 = 0 further without limiting the generality because we can get Gt,1x from Gt,0y for some y by switch of the letters of the alphabet {0, 1}. Then a1 = (at, 1). Note that a−1 1 = (1, a−1 t )σ and for 2 6 k 6 t a−1 k = { (a−1 k−1, 1), if xk = 0, (1, a−1 k−1), otherwise. Let u be a word in {a1, a2, . . . , at, a −1 1 , a−1 2 , . . . , a−1 t } that represents ele- ment g ∈ Gt,x. We denote by |u|k the exponent sum of ak in u. By the proposition 3.3 from [1] it follows that this numbers does not depend on word u representing element g ∈ Gt,x. Thus we define |g|k as the exponent sum of ak for the word u. I . Samoilovych 289 Lemma 1. For every integer n > 0 we have a2n k ∈ St(nt + k − 1), 1 6 k 6 t. Proof. We will prove this by induction on n. Statement for n = 0 follow from the definition of elements ak, 1 6 k 6 t. We assume that the statement of the Lemma holds for n− 1 > 0. By the definition of a1 we have a2n 1 = (a2 1)2n−1 = (at, at) 2n−1 = (a2n−1 t , a2n−1 t ), and we see that a2n 1 ∈ St(nt) by the inductive hypothesis. For 2 6 k 6 t the inclusion a2n k ∈ St(nt + k − 1) follows from the equalities a2n k = (a2n k−1, 1) for xk = 0, a2n k = (1, a2n k−1) for xk = 1 and the inclusion a2n k−1 ∈ St(nt+ k − 2). Lemma 2. If g ∈ Gt,x then ∑ v∈0T t−1 |g(v)|1 = ⌈ |g|1 2 ⌉ and ∑ v∈1T t−1 |g(v)|1 = ⌊ |g|1 2 ⌋ . Proof. We could represent element from Gt,x as product over the set {a1, . . . , at, a −1 1 , . . . , a−1 t }. We prove the statement by induction on the length of the product. The base is true as we have the trivial element in that case. We suppose that the statement is true for h ∈ Gt,x h = (h0, h1)σi∅ . We will multiply element from the set {a1, . . . , at, a −1 1 , . . . , a−1 t } on h. We denote the result of multiplying by g. There are 2t possibilities for the element g and we consider all that possibilities. Note that for any v ∈ T t we have (ai)(v) = 1 or (ai)(v) = ai and for any v ∈ T t we have (a−1 i )(v) = 1 or (a−1 i )(v) = a−1 i . i) Case g = a1h. Then |g|1 = |h|1 + 1 and a1h = (ath1, h0)σ1−i∅ . For the element (a1h)(0) we have the following sum ∑ v∈0T t−1 |(a1h)(v)|1 = |a1|1 + ∑ v∈1T t−1 ∣∣∣h(v) ∣∣∣ 1 . 290 Profinite closures of monodromy groups Since the last sum is equal to 1 + ⌊ |h|1 2 ⌋ by the inductive hypothesis then we have 1 + ⌊ |h|1 2 ⌋ = ⌈ |h|1 + 1 2 ⌉ = ⌈ |g|1 2 ⌉ . For the element (a1h)(1) we have ∑ v∈1T t−1 |(a1h)(v)|1 = ∑ v∈0T t−1 |h(v)|1 = ⌈ |h|1 2 ⌉ = ⌊ |h|1 + 1 2 ⌋ = ⌊ |g|1 2 ⌋ . ii) Case g = a−1 1 h. Then |g|1 = |h|1 − 1 and a−1 1 h = (h1, a −1 t h0)σ1−i∅ . We have the following sum for the element (a−1 1 h)(0) ∑ v∈0T t−1 |(a−1 1 h)v|1 = ∑ v∈1T t−1 ∣∣∣h(v) ∣∣∣ 1 . This number is equal to ⌊ |h|1 2 ⌋ by the inductive hypothesis. So we have ∑ v∈1T t−1 ∣∣∣h(v) ∣∣∣ 1 = ⌊ |h|1 2 ⌋ = ⌈ |h|1 − 1 2 ⌉ = ⌈ |g|1 2 ⌉ . For the element (a−1 1 h)(1) we have |a−1 1 |1 + ∑ v∈0T t−1 |h(v)|1 = ⌈ |h|1 2 ⌉ − 1 = ⌊ |h|1 − 1 2 ⌋ = ⌊ |g|1 2 ⌋ . iii) We consider all cases aih and a−1 i h, 2 6 i 6 t together as for them we get the same exponent sum of a1: |aih|1 = |a−1 i h|1 = |h|1 and aih = { (ai−1h0, h1)σi∅ , if xi = 0, (h0, ai−1h1)σi∅ , otherwise. a−1 i h = { (a−1 i−1h0, h1)σi∅ , if xi = 0, (h0, a −1 i−1h1)σi∅ , otherwise. Therefore ∑ v∈0T t−1 |(aih)(v)|1 = ∑ v∈0T t−1 |(a−1 i h)(v)|1 = ∑ v∈0T t−1 |h(v)|1 = ⌈ |h|1 2 ⌉ , ∑ v∈1T t−1 |(aih)(v)|1 = ∑ v∈1T t−1 |(a−1 i h)(v)|1 = ∑ v∈1T t−1 |h(v)|1 = ⌊ |h|1 2 ⌋ . I . Samoilovych 291 We prove that functions lt,i and rt,i depend only on the exponent sum of a1. Lemma 3. For every element g ∈ Gt,x and for every not negative integer i the following equalities hold lt,i(g) = ⌈ |g|1 2i ⌉ mod 2 and rt,i(g) = ⌊ |g|1 2i ⌋ mod 2. Proof. We prove this Lemma by induction on i. In the case i = 0 we have lt,0(g) = rt,0(g) = |g|1 mod 2. Now we suppose that the lemma holds for i− 1 where i > 1. By the definition of lt,i we have lt,i(g) = lt,i−1(Lt(g)) = ⌈ |Lt(g)|1 2i−1 ⌉ mod 2 = ⌈∑ v∈0T t−1 |g(v)|1 2i−1 ⌉ mod 2. We conclude by the Lemma 2 that lt,i(g) = ⌈ ⌈|g|1/2⌉ 2i−1 ⌉ mod 2 = ⌈ |g|1 2i ⌉ mod 2. Analogously, for the function rt,i we have rt,i(g) = rt,i−1(Rt(g)) = ⌊ |Rt(g)|1 2i−1 ⌋ mod 2 = ⌊ ⌊|g|1/2⌋ 2i−1 ⌋ mod 2 = ⌊ |g|1 2i ⌋ mod 2. Theorem 1. For every g ∈ Gt,x and every not negative integer k the following holds k∑ i=0 lt,i(g)2i + k∑ i=0 rt,i(g)2i ≡ 0 mod 2k+1. (1) Proof. It follows from the Lemma 3 that k∑ i=0 lt,i(g)2i = k∑ i=0 (⌈ |g|1 2i ⌉ mod 2 ) 2i, k∑ i=0 rt,i(g)2i = k∑ i=0 (⌊ |g|1 2i ⌋ mod 2 ) 2i. This equalities are the 2−adic representations for −|g|1 mod 2k+1 and |g|1 mod 2k+1 respectively. 292 Profinite closures of monodromy groups Remark 1. Equality 1 also holds for every automorphism g ∈ Gt,x|T [n] where n > kt+ 1. We can choose such element g′ ∈ Gt,x that g′|T [n] = g. Equality 1 holds for g′ and so it holds for g. Lemma 4. If for the automorphism g = (g0, g1)σi∅ ∈ AutT where g0, g1 ∈ Gt,x, i∅ ∈ {0, 1} and for the integer k > 0 following congruence relation holds k∑ i=0 lt,i(g)2i + k∑ i=0 rt,i(g)2i ≡ 0 mod 2k+1, then |g1|t + i∅ ≡ |g0|t mod 2k. Proof. At first we make the following transformation k∑ i=0 lt,i(g)2i + k∑ i=0 rt,i(g)2i = 2 k∑ i=1 lt,i(g)2i−1 + 2 k∑ i=1 rt,i(g)2i−1 + 2i∅ ≡ 0 mod 2k+1. Then we divide last congruence by 2 and we get k−1∑ i=0 lt,i+1(g)2i + k−1∑ i=0 rt,i+1(g)2i + i∅ ≡ 0 mod 2k. By the definition of lt,i+1 and rt,i+1 we have k−1∑ i=0 lt,i+1(g)2i + k−1∑ i=0 rt,i+1(g)2i = k−1∑ i=0 lt,i(Lt(g))2i + k−1∑ i=0 rt,i(Rt(g))2i. As g0, g1 ∈ Gt,x then g(v) ∈ Gt,x for all non-empty v and we can use Lemma 3 for the previous expression and we get k−1∑ i=0 (⌈ |Lt(g)|1 2i ⌉ mod 2 ) 2i + k−1∑ i=0 (⌊ |Rt(g)|1 2i ⌋ mod 2 ) 2i = k−1∑ i=0     ∑ v∈0T t−1 |g(v)|1 2i   mod 2   2i + k−1∑ i=0    ∑ v∈1T t−1 |g(v)|1 2i mod 2   2i. From the definition of generators of Gt,x we have ∑ v∈T t−1 |(g0)(v)|1 = |g0|t and ∑ v∈T t−1 |(g1)(v)|1 = |g1|t. I . Samoilovych 293 Now we have the condition k−1∑ i=0 (⌈ |g0|t 2i ⌉ mod 2 ) 2i + k−1∑ i=0 (⌊ |g1|t 2i ⌋ mod 2 ) 2i + i∅ ≡ 0 mod 2k. Congruence above contains 2-adic representations for −|g0|t mod 2k and |g1|t mod 2k respectively which finally lead us to −|g0|t + |g1|t + i∅ ≡ 0 mod 2k. Theorem 2. We assume that g = (g0, g1)σi∅ ∈ Gt,x|T [n] and g′ = (g′ 0, g ′ 1)σi∅ where g′ 0, g ′ 1 ∈ Gt,x|T [n], g′ 0|T [n−1] = g0 and g′ 1|T [n−1] = g1 for n > 1. Then g′ ∈ Gt,x|T [n+1] if and only if the following congruence holds for k = ⌊ n t ⌋ k∑ i=0 lt,i(g ′)2i + k∑ i=0 rt,i(g ′)2i ≡ 0 mod 2k+1. (2) Proof. Necessary condition follow from the Theorem 1. Here we prove the sufficient condition. We can choose such elements w0, w1 ∈ Gt,x that w0|T [n] = g′ 0 and w1|T [n] = g′ 1. By the Lemma 4 there exists such integer j that |w1|t + i∅ = |w0|t + j2k. By the Lemma 1 we have a2k t ∈ St(kt + t − 1). As n 6 kt + t − 1 then a2k t |T [n] is trivial and w0a j2k t |T [n] = g′ 0. If i∅ = 0 then |w1|t = |w0|t + j2k. Stabilizer St(1) is generated by a2 1 = (at, at), a −xi 1 aia xi 1 = (ai−1, 1) and axi−1 1 aia 1−xi 1 = (1, ai−1) where 2 6 i 6 t. So we have (w0a j2k t , w1) = (a |w1|t t , w1)(a −|w1|t t , a −|w1|t t )(w0a j2k t , a |w1|t t ) ∈ St(1). If i∅ = 1 then we can generate element a1 · (w0a j2k t , w1)σ = (atw1, w0a j2k t ) because of 1 + |w1|t = |w0|t + j2k. Remark 2. The congruence 2 from Theorem 2 gives us a restriction only in the case n = kt for some positive integer k. In other cases congruence 2 holds because kt < n therefore action of g′ on the level n+ 1 not used for the congruence 2. Remark 3. We do not use order in products from definitions of Ln(g) and Rn(g). 294 Profinite closures of monodromy groups Remark 4. We prove our formula for the case x1 = 0 only. But if we switch letters in the alphabet {0, 1} then by the previous Remark we conclude that it is enough to switch functions lt,i and rt,i in the congruence 2 of the Theorem 2 to get the criteria for Gt,x|T [n] for the case x1 = 1. But then we get the same criteria as for the case x1 = 0. Pink [7] already showed that group Gt,x depend only on t up to conjugacy in AutT . The following corollary improves that result for the periodic case. Corollary 1. For all words x1, x2 ∈ {0, 1}t equality Gt,x1 = Gt,x2 holds. Proof. Any group Gt,x is Sym{0, 1} on the first level. Congruence 2 do not depend on x therefore Gt,x do not depend on the word x. Remark 5. In general group Gt,x depend on a word x. For example when t = 3 groups corresponding to “Douady Rabbit” and “Airplane” [6] are not isomorphic. 3. Strictly pre-periodic case In this section, we consider profinite closures of iterated monodromy groups associated with quadratic polynomials with pre-periodic critical points. Let us fix integers t > s > 1 and a word x = x1x2 . . . xt−1 over the alphabet {0, 1}. We define elements a1, ..., at ∈ AutT by recursions a1 = (1, 1)σ, as+1 = { (as, at), if xs = 0, (at, as), otherwise, ai = { (ai−1, 1), if xi−1 = 0, (1, ai−1), otherwise. We will study group Gs,t,x = 〈a1, . . . , at〉 and its profinite closure Gs,t,x. Almost all groups Gs,t,x are self-similar groups of finite type, only one exception is the case s = 1 and t = 2. We consider this exception later. Lemma 5. (Proposition 4.2 from[1]) For all 1 6 i 6 t equality a2 i = 1 holds. Abealinization Gs,t,x/G ′ s,t,x is (Z/(2Z))t generated by the images of the ai. I . Samoilovych 295 Let u be a word in {a1, a2, . . . , at} that represents element g ∈ Gs,t,x. We denote by |u|k the exponent sum of ak in u. By the Lemma 5 it follows that number |u|k mod 2 does not depend on the word u representing element g ∈ Gs,t,x. Thus we define |g|k as |u|k mod 2. Lemma 6. Element g ∈ Gs,t,x belong to the commutator G′ s,t,x iff |g|i = 0 for all 1 6 i 6 t. Proof. Statement follow from the Lemma 5. Lemma 7. If g ∈ St(i) then |g|i = 0 for all 1 6 i 6 t. Proof. We prove this by induction on i. If g ∈ St(1) then |g|1 = 0. Let g = (g1, g2). If g ∈ St(i) then g1, g2 ∈ St(i− 1) and by induction suppose for g1 and g2 we have that numbers |g1|i−1 and |g2|i−1 are zeroes. Note that equality |g|i = (|g1|i−1 + |g2|i−1) mod 2 holds therefore |g|i = 0. Lemma 8. If g ∈ St(t) then g ∈ G′ s,t,x. Proof. Statement follow from the Lemma 6 and the Lemma 7. Theorem 3. Suppose that s 6= 1 and t 6= 2 simultaneously. Then group Gs,t,x is a self-similar group of finite type with the following minimal depth d = { 5, if s = 2 and t = 3, t+ 1, otherwise. Proof. We use Sunic’s [8] criteria from the Theorem 3 for a self-similar group of finite type. It is already proven that the group Gs,t,x is regular branch over some group Hs,t,x (Theorem 4.10 from [1]). It is only left to show that Hs,t,x contain a stabilizer of the level d. Let us to consider following cases. • If s > 2 and t > s+ 2 or if s > 3 and t = s+ 1 than Hs,t,x = G′ s,t,x and statement follow from the Lemma 8. • If s = 1 and t > 3 then Hs,t,x = 〈[ai, aj ]26i<j6t, [a1, aj ]16j<t〉 Gs,t,x . Group G′ s,t,x/Hs,t,x ∼= Z2 generated by [a1, at]. Therefore G′ s,t,x = Hs,t,x ∪ ([a1, at]Hs,t,x). By the Lemma 8 we have StGs,t,x (t) < G′ s,t,x. We prove that Hs,t,x contain a stabilizer of the level t in the following way: for any element g ∈ G′ s,t,x we prove that if g /∈ Hs,t,x then g /∈ StGs,t,x (t). 296 Profinite closures of monodromy groups It is easy to see that for all h = (h1, h2) ∈ Hs,t,x we have |h1|t−1 = |h2|t−1 = 0. As [a1, at]=(at−1, at−1) then for all g=(g1, g2) ∈ [a1, at]Hs,t,x we conclude that |g1|t−1 and |g2|t−1 are ones; but by the Lemma 7 it follow that g1, g2 /∈ StGs,t,x (t− 1). Therefore g /∈ StGs,t,x (t). • Case s = 2, t = 3. We consider case x1 = 0 only without limiting the generality. In this case we have H2,3,x = 〈[a1, a2a3]〉G2,3,x . Also G′ 2,3,x = H2,3,x ∪ ([a1, a2]H2,3,x) ∪ ([a2, a3]H2,3,x) ∪ ([a1, a2][a2, a3]H2,3,x). Note that St(4) < G′ 2,3,x. We prove for any g ∈ G′ 2,3,x that if g /∈ H2,3,x then g /∈ St(4) using GAP in the following way: we compute group H2,3,x|T [4] and we check that e /∈ ([a1, a2]H2,3,x)|T [4] ∪ ([a2, a3]H2,3,x)|T [4] ∪ ([a1, a2][a2, a3]H2,3,x)|T [4] . Minimality for all cases except s = 2 and t = 3 follow from the fact that Gs,t,x|T [t] = Aut(T t). For the case s = 2 and t = 3 it can be seen using GAP. Group Gs,t,x barely depend on the word x. Pink [7] proved that for all words z, y ∈ {0, 1}t−1 there exists such w ∈ AutT that Gs,t,z = wGs,t,yw −1. The following theorem improves that result. Theorem 4. Suppose that s 6= 1 and t 6= 2 simultaneously. • If s > 2 and t > s+2 or if s > 3 and t = s+1. Then for all possible words y = y1 . . . yt−1 we have Gs,t,x = Gs,t,y. • If s = 1 and t > 3 then for all possible words y = y1 . . . yt−1 where xt−1 = yt−1 we have Gs,t,x = Gs,t,y. • If s = 2, t = 3 then all 4 groups G2,3,x for this case are conjugate. Proof. All groups in this theorem are self-similar groups of finite type therefore it is enough to prove the statement for pattern groups. For the case s = 2, t = 3 using GAP we can see that there are 4 different groups G2,3,x as sets. Pink [7] already prove that all G2,3,x are conjugate. But we show some example of conjugator for G2,3,00 and G2,3,01. We find using GAP such elements f1, f2 ∈ AutT [5] and g1, g2, g3, g4 ∈ G2,3,00|T [4] that f1 = (g1f2|T [4] , g2f2|T [4]) and f2 = (g3f1|T [4] , g4f1|T [4]), G2,3,00|T [5] = f1G2,3,01|T [5]f−1 1 and G2,3,00|T [5] = f2G2,3,01|T [5]f−1 2 . I . Samoilovych 297 Then we can choose such elements ĝ1, ĝ2, ĝ3, ĝ4 ∈ G2,3,00 that ĝi|T [4] = gi. Therefore we can define elements f̂1 = (ĝ1f̂2, ĝ2f̂2) and f̂2 = (ĝ3f̂1, ĝ4f̂1). For ĥ = (h1, h2)σi ∈ G2,3,01, i ∈ {0, 1} the following holds f̂1(h1, h2)σif̂1 −1 = (ĝ1f̂2h1f̂2 −1 ĝ1+i −1, ĝ2f̂2h2f̂2 −1 ĝ2−i −1)σi, f̂2(h1, h2)σif̂2 −1 = (ĝ3f̂1h1f̂1 −1 ĝ3+i −1, ĝ4f̂1h2f̂1 −1 ĝ4−i −1)σi. As G2,3,00 is self-similar group of finite type then it follows that f̂1G2,3,01f̂1 −1 = G2,3,00 and f̂2G2,3,01f̂2 −1 = G2,3,00. One possible choice for ĝi is ĝ1 = 1, ĝ2 = a2a 3 3a1, ĝ3 = a2 3a2, ĝ4 = a3a2a 2 3a2a1. We exclude case s = 2, t = 3 from further consideration. Let us to fix the following elements a′ s+1 = { (a′ s, a ′ t), if ys = 0, (a′ t, a ′ s), otherwise. a′ i = { (a′ i−1, 1), if yi−1 = 0, (1, a′ i−1), otherwise. In our proof we use a notation for the Kronecker delta δij = { 0, if i = j, 1, if i 6= j. Then we have for 2 6 i 6 t: ai = a xi−1 1 (ai−1, a δi,s+1 t )a xi−1 1 , a′ i = a yi−1 1 (a′ i−1, (a ′ t) δi,s+1)a yi−1 1 . We define function φ on the set of all words in the alphabet {a1, . . . , at−1} φ : {a1, . . . , at−1} ∗ → {a1, . . . , at} ∗ 298 Profinite closures of monodromy groups recursively as follows φ(∅) = ∅, φ(ai) = axi 1 ai+1a xi 1 , 1 6 i 6 t− 1, φ(αβ) = φ(α)φ(β), where α, β ∈ {a1, . . . , at−1} ∗. Note that a0 i is the empty word therefore φ(a0 i ) = 1. We define function ψ(w) as element from the group Gs,t,x that corre- spond to the word w ∈ {a1, . . . , at} ∗, note that ψ(∅) = 1. Functions ψ and φ are homomorphisms. Let us to show that the following properties hold for all words w ∈ {a1, . . . , at−1} ∗ ψ(φ(w)) = (ψ(w), a |w|s t ), |ψ(φ(w))|1 = 0. We prove these by induction on length of the word. For the first one we have ψ(φ(∅)) = ψ(∅) = 1 = (ψ(∅), 1), ψ(φ(ai)) = ψ(axi 1 ai+1a xi 1 ) = axi 1 ai+1a xi 1 = axi 1 a xi 1 (ai, a δi+1,s+1 t )axi 1 a xi 1 = (ai, a |ai|s t ), ψ(φ(aiw)) = ψ(φ(ai)φ(w)) = ψ(φ(ai))ψ(φ(w)) = (aiψ(w), a |ai|s+|w|s t ). For the second one we have |ψ(φ(∅))|1 = |1|1 = 0, |ψ(φ(ai))|1 = |axi 1 ai+1a xi 1 |1 = 0, 1 6 i 6 t− 1, |ψ(φ(aiw))|1 = |ψ(φ(ai))|1 + |ψ(φ(w))|1 = 0. Without limiting the generality it is enough to show that 〈a′ 1, . . . , a ′ t〉|T [t+1] < 〈a1, . . . , at〉|T [t+1] . We construct such elements from Gs,t,x that act on first t+ 1 levels as a′ i. So we define c1 = ∅, ci = a yi−1 1 φ(ci−1)a xi−1 1 , 2 6 i 6 t, d1 = a′ 1 = a1, di = ψ(ci)aiψ(ci) −1, 2 6 i 6 t. Note that ci ∈ {a1, . . . , at−1} ∗ for 1 6 i 6 t . We prove the following di = a yi−1 1 (di−1, a δi,s+1 t )a yi−1 1 I . Samoilovych 299 by induction on i. c2 = ay1 1 a x1 1 , d2 = ay1 1 a x1 1 a2a x1 1 a y1 1 = ay1 1 (a1, a δ2,s+1 t )ay1 1 . di = ψ(ci)aiψ(ci) −1 = a yi−1 1 ψ(φ(ci−1))a xi−1 1 ai(a yi−1 1 ψ(φ(ci−1))a xi−1 1 )−1 = a yi−1 1 (ψ(ci−1)ai−1ψ(ci−1)−1, a δi,s+1 t )a yi−1 1 = a yi−1 1 (di−1, a δi,s+1 t )a yi−1 1 . We have ds+1 = ays 1 (ds, at)a ys 1 . But we need an element ds+1 = ays 1 (ds, dt)a ys 1 ∈ Gs,t,x. We construct this element for each case. • Case s > 2 and t > s+ 2 or if s > 3 and t = s+ 1. There exist such b1 ∈ 〈a1, a2〉 and b2 ∈ 〈a1, a2, a3〉 that [ds, a b1 t ] = 1 and b2 = (b1, 1). If s > 2 and t > s+ 2 then b1 = a xt−1+ys−1+1 1 . If s > 3 and t = s+ 1 then b1 = a xt−1 1 ψ(φ(a (xt−2+ys−2+1) mod 2 1 ))a ys−1 1 . For both cases b2 = ψ(φ(b1)). Then we compute the following b3 = ays 1 ψ(φ(ct)) a1b2ays 1 = ays 1 ((ar t )b1 , ψ(ct))a ys 1 , r = |ψ(ct)|s, ds+1 = b3ds+1b −1 3 = b3(ays 1 (ds, at)a ys 1 )b−1 3 = ays 1 ((ar t )b1ds((ar t )b1)−1, ψ(ct)atψ(ct) −1)ays 1 = ays 1 (ds, dt)a ys 1 . • Case s = 1 and t > 3 and xt−1 = yt−1. Therefore |ψ(ct)|1 = |ψ(a yt−1 1 φ(ct−1)a xt−1 1 )|1 = (yt−1 + xt−1) mod 2 = 0, b3 = ays 1 ψ(φ(ct)) a1ays 1 = ays 1 (a |ct|1 t , ψ(ct))a ys 1 = ays 1 (1, ψ(ct))a ys 1 , ds+1 = b3ds+1b −1 3 = b3(ays 1 (ds, at)a ys 1 )b−1 3 = ays 1 ((ds, ψ(ct)atψ(ct) −1)ays 1 = ays 1 (ds, dt)a ys 1 . 300 Profinite closures of monodromy groups We prove that di|T [t] = a′ i|T [t] for 1 6 i 6 t by induction on i. Base is true because d1 = a′ 1. We have for 2 6 i 6 t, i 6= s+ 1 the following di|T [t] = a yi−1 1 |T [t](di−1|T [t−1] , 1|T [t−1])a yi−1 1 |T [t] = a yi−1 1 |T [t](a′ i−1|T [t−1] , 1|T [t−1])a yi−1 1 |T [t] = a′ i|T [t] , ds+1|T [t] = ays 1 |T [t](ds|T [t−1] , at|T [t−1])a ys 1 |T [t] = ays 1 |T [t](a′ s|T [t−1] , 1|T [t−1])a ys 1 |T [t] = a′ s+1|T [t] . Therefore di|T [t+1] = a yi−1 1 |T [t+1](di−1|T [t] , 1|T [t])a yi−1 1 |T [t+1] = a yi−1 1 |T [t+1](a′ i−1|T [t] , 1|T [t])a yi−1 1 |T [t+1] = a′ i|T [t+1] , ds+1|T [t+1] = ays 1 |T [t+1](ds|T [t] , dt|T [t])a ys 1 |T [t+1] = ays 1 |T [t+1](a′ s|T [t] , a′ t|T [t])a ys 1 |T [t+1] = a′ s+1|T [t+1] . So 〈a′ 1, . . . , a ′ t〉|T [t+1] < 〈a1, . . . , at〉|T [t+1] . 3.1. Case s = 1 and t = 2 Here we consider the profinite closureG = G1,2,0 of the infinite dihedral group G = G1,2,0 generated by following automorphisms a1 = (1, 1)σ, a2 = (a1, a2). Note that a−1 1 = a1 and a−1 2 = a2. Let us to compute a1a2 and a2a1 a1a2 = (a2, a1)σ, a2a1 = (a1, a2)σ. Then for all integer k we have (a1a2)k = (a1a2)2⌊ k 2⌋+k mod 2 = ((a1a2)2)⌊ k 2⌋(a1a2)k mod 2 = (a2a1, a1a2)⌊ k 2⌋(ak mod 2 2 , ak mod 2 1 )σk mod 2. Therefore (a1a2)k = ((a1a2)−⌊ k 2⌋ak mod 2 2 , (a1a2)⌊ k 2⌋ak mod 2 1 )σk mod 2. (3) Lemma 9. For all integer n > 0 the following equality holds ∏ v∈T n l0(((a1a2)k)(v)) = { 1, if 2n | k and 2n+1 ∤ k, 0, otherwise. I . Samoilovych 301 Proof. Let us prove this by induction on n. One can directly check the equality for n = 0, 1. We assume that Lemma holds for n− 1 where n > 2. If 2n | k and 2n+1 ∤ k then k = k02n for some odd k0 and by equality 3 (a1a2)k = ((a1a2)−k02n−1 , (a1a2)k02n−1 ). So in this case we have ∏ v∈T n l0(((a1a2)k)(v)) = ∏ v∈T n−1 l0(((a1a2)−k02n−1 )(v)) · ∏ v∈T n−1 l0(((a1a2)k02n−1 )(v)) = 1 · 1 = 1. Let us to consider other cases. As n > 2 and a1 act trivially on all levels except first we have for all v ∈ Tn−1 l0(((a1a2)k)(1v)) = l0(((a1a2)⌊ k 2⌋ak mod 2 1 )(v)) = l0(((a1a2)⌊ k 2⌋)(v)). If ⌊ k 2 ⌋ 6= k02n−1 for all odd k0 then ∏ v∈T n l0(((a1a2)k)(v)) = ∏ v∈0T n−1 l0(((a1a2)k)(v)) · ∏ v∈T n−1 l0(((a1a2)⌊ k 2⌋)(v)) = ∏ v∈0T n−1 l0(((a1a2)k)(v)) · 0 = 0. If ⌊ k 2 ⌋ = k02n−1 for some odd integer k0 then k = k02n + 1 (because now we consider k 6= k02n) l0(((a1a2)k)(0v)) = l0(((a1a2)−k02n−1 a2)(v))= l0(((a1a2)−k02n−1 a2a1a1)(v)) = l0(((a1a2)−k02n−1−1a1)(v)) = l0(((a1a2)−k02n−1−1)(v)). Then ∏ v∈T n l0(((a1a2)k)(v)) = ∏ v∈T n−1 l0(((a1a2)−k02n−1−1)(v)) · ∏ v∈1T n−1 l0(((a1a2)k)(v)) = 0. Lemma 10. For all g ∈ G and all integer n > 2 the following congruence holds l0(g(01n−1)) + l0(g(1n)) + ∏ v∈T n−1 l0(g(v)) ≡ 0 mod 2. (4) 302 Profinite closures of monodromy groups Proof. Every element of the group G is the product (a1a2)kak1 1 where k is some integer and k1 ∈ {0, 1}. As a1 act not trivially on the first level only it is enough to prove congruence for (a1a2)k for all integer k. We use the following notations: for every integer n > 0 and every integer k we denote αk n+1 = l0(((a1a2)k)(01n)), βk n = l0(((a1a2)k)(1n)). From the equality 3 we conclude that for all n > 1 and all integer k βk 0 = k mod 2, αk n = l0(((a1a2)−⌊ k 2⌋ak mod 2 2 )(1n−1)), βk n = l0(((a1a2)⌊ k 2⌋ak mod 2 1 )(1n−1)). Number of a1 in the expression (a1a2)⌊ k 2⌋ak mod 2 1 module 2 is equal to the number βk 1 βk 1 = (⌊ k 2 ⌋ + k ) mod 2. As a1 act not trivially on the first level only then we can omit last a1 for n > 2 βk n = l0(((a1a2)⌊ k 2⌋ak mod 2 1 )(1n−1)) = l0(((a1a2)⌊ k 2⌋)(1n−1)) = β ⌊ k 2⌋ n−1 . Then we have βk n = β ⌊ k 2⌋ n−1 = . . . = β ⌊ k 2n−1 ⌋ 1 = (⌊ k 2n ⌋ + ⌊ k 2n−1 ⌋) mod 2 for n > 1. Now we compute αk n for all integer k and all n > 2 αk n = l0(((a1a2)−⌊ k 2⌋ak mod 2 2 )(1n−1)) = l0(((a1a2)−⌊ k 2⌋−k mod 2)(1n−1)) Note that ⌊ k 2 ⌋ + k mod 2 = ⌈ k 2 ⌉ . Then αk n = β −⌈ k 2⌉ n−1 = ( − ⌈ k 2n ⌉ − ⌈ k 2n−1 ⌉) mod 2 = (⌈ k 2n ⌉ + ⌈ k 2n−1 ⌉) mod 2. Now congruence 4 looks like the following ⌈ k 2n ⌉ + ⌈ k 2n−1 ⌉ + ⌊ k 2n ⌋ + ⌊ k 2n−1 ⌋ + ∏ v∈T n−1 l0(((a1a2)k)(v)) ≡ 0 mod 2. I . Samoilovych 303 We have by the properties of floor and ceil that ⌈ k 2n ⌉ − ⌊ k 2n ⌋ = { 0, if 2n | k, 1, otherwise. ⌈ k 2n−1 ⌉ − ⌊ k 2n−1 ⌋ = { 0, if 2n−1 | k, 1, otherwise. If 2n | k then 2n−1 | k and by the Lemma 9 congruence 4 looks like the following ⌈ k 2n ⌉ + ⌊ k 2n ⌋ + ⌈ k 2n−1 ⌉ + ⌊ k 2n−1 ⌋ ≡ 0 + 0 ≡ 0 mod 2. If 2n ∤ k and 2n−1 ∤ k then by the Lemma 9 congruence 4 looks like the following ⌈ k 2n ⌉ + ⌊ k 2n ⌋ + ⌈ k 2n−1 ⌉ + ⌊ k 2n−1 ⌋ ≡ 1 + 1 ≡ 0 mod 2. If 2n ∤ k and 2n−1 | k then by the Lemma 9 congruence 4 looks like the following ⌈ k 2n ⌉ + ⌊ k 2n ⌋ + ⌈ k 2n−1 ⌉ + ⌊ k 2n−1 ⌋ + 1 ≡ 1 + 0 + 1 ≡ 0 mod 2. Theorem 5. We suppose that element g = (g0, g1)σi∅ ∈ G|T [n] and g′ = (g′ 0, g ′ 1)σi∅ ∈ AutT [n+1] where g′ 0, g ′ 1 ∈ G|T [n], g′ 0|T [n−1] = g0 and g′ 1|T [n−1] = g1 for integer n > 3. Then g′ ∈ G|T [n+1] if and only if the following congruence holds l0(g′ (01n−1)) + l0(g′ (1n)) + ∏ v∈T n−1 l0(g(v)) ≡ 0 mod 2. (5) Proof. Necessary condition follow from the Lemma 10. Let us prove the sufficient condition. We know (see [7]) that |G|T [n−1] | = 2n for n > 3 and it imply that |(St(n − 1))n| = 2 for n > 3. So there are 4 possible ways how we can construct g′ by given g. Group G is level transitive and it imply that there are such elements h0, h1 in St(n− 1)|T [n] that l0((h0)(1n−1)) = 0 and l0((h1)(1n−1)) = 1. It imply that all possibilities for the pair (l0(g′ (01n−1)), l0(g′ (1n))) are realized and 304 Profinite closures of monodromy groups pairs (l0(g′ (01n−1)), l0(g′ (1n))) are in one-to-one correspondence to g′ con- structed in such way. For the fixed g the product ∏ v∈T n−1 l0(g(v)) is also fixed. Then there are only two pairs (l0(g′ (01n−1)), l0(g′ (1n))) that satisfy congruence 5. As |St(n)|T [n+1] | = 2 then there are two possibilities for the element g′ that belong to G|T [n+1] . So element g′ that corresponds to the pair (l0(g′ (01n−1)), l0(g′ (1n))) which satisfy congruence 5 belongs to G|T [n+1] . Note that group G|T [2] is AutT [2]. For all elements g ∈ G|T [3] one can directly check following congruences l0(g(0)) ≡ l0(g(10)) + l0(g(11)) mod 2, l0(g(1)) ≡ l0(g(00)) + l0(g(01)) mod 2. This congruences together with Lemma 10 and Theorem 5 give us full explicit description of groups G|T [n] for n > 3. References [1] L. Bartholdi, V.V. Nekrashevych, Iterated monodromy groups of quadratic polyno- mials, I, Groups, Geometry, and Dynamics, Vol. 2, N. 3, 2008, pp. 309–336. [2] A.M. Brunner, S. Sidki, On the Automorphism Group of the One-Rooted Binary Tree, Journal of Algebra, Vol. 195, N. 2, 1997, pp. 465–486. [3] L. Bartholdi, B. Virág, Amenability via random walks, Duke Mathematical Journal, Vol. 130, N. 1, 2005, pp. 39–56. [4] I.V. Bondarenko, I.O. Samoilovych, On finite generation of self-similar groups of finite type, International Journal of Algebra and Computation, Vol. 23, N. 1, 2013, pp. 69–79. [5] R.I. Grigorchuk, A. Żuk, On a torsion-free weakly branch group defined by a three state automaton, International Journal of Algebra and Computation, Vol. 12, N. 01n02, 2002, pp. 223–246. [6] V. Nekrashevych, Self-Similar Groups, American Mathematical Soc., N. 117, 2005. [7] R. Pink, Profinite iterated monodromy groups arising from quadratic polynomial, arXiv preprint arXiv:1307.5678, 2013. [8] Z. Šunić, Hausdorff dimension in a family of self-similar groups, Geometriae Dedicata, Vol. 124, N. 1, 2007, pp. 213–236. Contact information I. Samoilovych Department of Mechanics and Mathematics, Kyiv National Taras Shevchenko University, Volodymyrska, 64, Kyiv 01033, Ukraine E-Mail(s): samoil449@gmail.com Received by the editors: 05.04.2017.