Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space

We use our new type of bounded locally homeomorphic quasiregular mappings in the unit 3-ball to address long standing problems for such mappings, including the Vuorinen injectivity problem. The construction of such mappings comes from our construction of non-trivial compact 4-dimensional cobordisms...

Повний опис

Збережено в:
Бібліографічні деталі
Опубліковано в: :Український математичний вісник
Дата:2019
Автор: Apanasov, B.N.
Формат: Стаття
Мова:English
Опубліковано: Інститут прикладної математики і механіки НАН України 2019
Онлайн доступ:https://nasplib.isofts.kiev.ua/handle/123456789/169429
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:Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space / B.N. Apanasov // Український математичний вісник. — 2019. — Т. 16, № 1. — С. 10-27. — Бібліогр.: 26 назв. — англ.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-169429
record_format dspace
spelling Apanasov, B.N.
2020-06-13T08:18:40Z
2020-06-13T08:18:40Z
2019
Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space / B.N. Apanasov // Український математичний вісник. — 2019. — Т. 16, № 1. — С. 10-27. — Бібліогр.: 26 назв. — англ.
1810-3200
2000 MSC. 30C65, 57Q60, 20F55, 32T99, 30F40, 32H30, 57M30
https://nasplib.isofts.kiev.ua/handle/123456789/169429
We use our new type of bounded locally homeomorphic quasiregular mappings in the unit 3-ball to address long standing problems for such mappings, including the Vuorinen injectivity problem. The construction of such mappings comes from our construction of non-trivial compact 4-dimensional cobordisms M with symmetric boundary components and whose interiors have complete 4-dimensional real hyperbolic structures. Such bounded locally homeomorphic quasiregular mappings are defined in the unit 3-ball B³ ⊂ R³ as mappings equivariant with the standard conformal action of uniform hyperbolic lattices Г ⊂ IsomH³ in the unit 3-ball and with its discrete representation G = ρ(Г) ⊂ IsomH⁴. Here, G is the fundamental group of our non-trivial hyperbolic 4-cobordism M = (H⁴∪Ω (G))/G, and the kernel of the homomorphism ρ: Г → G is a free group F₃ on three generators.
en
Інститут прикладної математики і механіки НАН України
Український математичний вісник
Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space
spellingShingle Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space
Apanasov, B.N.
title_short Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space
title_full Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space
title_fullStr Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space
title_full_unstemmed Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space
title_sort hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space
author Apanasov, B.N.
author_facet Apanasov, B.N.
publishDate 2019
language English
container_title Український математичний вісник
publisher Інститут прикладної математики і механіки НАН України
format Article
description We use our new type of bounded locally homeomorphic quasiregular mappings in the unit 3-ball to address long standing problems for such mappings, including the Vuorinen injectivity problem. The construction of such mappings comes from our construction of non-trivial compact 4-dimensional cobordisms M with symmetric boundary components and whose interiors have complete 4-dimensional real hyperbolic structures. Such bounded locally homeomorphic quasiregular mappings are defined in the unit 3-ball B³ ⊂ R³ as mappings equivariant with the standard conformal action of uniform hyperbolic lattices Г ⊂ IsomH³ in the unit 3-ball and with its discrete representation G = ρ(Г) ⊂ IsomH⁴. Here, G is the fundamental group of our non-trivial hyperbolic 4-cobordism M = (H⁴∪Ω (G))/G, and the kernel of the homomorphism ρ: Г → G is a free group F₃ on three generators.
issn 1810-3200
url https://nasplib.isofts.kiev.ua/handle/123456789/169429
citation_txt Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space / B.N. Apanasov // Український математичний вісник. — 2019. — Т. 16, № 1. — С. 10-27. — Бібліогр.: 26 назв. — англ.
work_keys_str_mv AT apanasovbn hyperbolictopologyandboundedlocallyhomeomorphicquasiregularmappingsin3space
first_indexed 2025-11-26T15:34:35Z
last_indexed 2025-11-26T15:34:35Z
_version_ 1850626745132122112
fulltext Український математичний вiсник Том 16 (2019), № 1, 10 – 27 Hyperbolic topology and bounded locally homeomorphic quasiregular mappings in 3-space Boris N. Apanasov (Presented by V. Gutlyanskĭı) Paper dedicated to the memory of my colleague and friend, brilliant mathematician Bogdan Bojarski Abstract. We use our new type of bounded locally homeomorphic quasiregular mappings in the unit 3-ball to address long standing prob- lems for such mappings, including Matti Vuorinen injectivity problem. The construction of such mappings comes from our construction of non- trivial compact 4-dimensional cobordisms M with symmetric boundary components and whose interiors have complete 4-dimensional real hy- perbolic structures. Such bounded locally homeomorphic quasiregular mappings are defined in the unit 3-ball B3 ⊂ R 3 as mappings equivari- ant with the standard conformal action of uniform hyperbolic lattices Γ ⊂ IsomH3 in the unit 3-ball and with its discrete representation G = ρ(Γ) ⊂ IsomH4. Here G is the fundamental group of our non- trivial hyperbolic 4-cobordism M = (H4 ∪ Ω(G))/G and the kernel of the homomorphism ρ :Γ → G is a free group F3 on three generators. 2010 MSC. 30C65, 57Q60, 20F55, 32T99, 30F40, 32H30, 57M30. Key words and phrases. Bounded quasiregular mappings, Vuorinen problem, local homeomorphisms, hyperbolic group action, hyperbolic manifolds, cobordisms, group homomorphism, deformations of geometric structures. 1. Introduction The theory of quasiregular mappings, initiated by the works of M. A. Lavrentiev and later by Reshetnyak and Martio, Rickman and Väisälä, shows that they form (from the geometric function theoretic point of view) the correct generalization of the class of analytic functions Received 27.03.2019 ISSN 1810 – 3200. c© Iнститут прикладної математики i механiки НАН України B. N. Apanasov 11 to higher dimensions. In particular, Reshetnyak proved that non-constant quasiregular mappings are (generalized) branched covers, that is, con- tinuous, discrete and open mappings and hence local homeomorphisms modulo an exceptional set of (topological) codimension at least two, and that they preserve sets of measure zero. For the theory of quasiregular mappings, see the monographs [20, 21] and [26]. Here we address some properties of bounded quasiregular mappings f : B3 → R3 in the unit ball B3 and well known problems on such quasiregular mappings. These results will be heavily based on our recent construction Apanasov [9] (Theorem 4.1) of surjective locally homemo- rphic quasiregular mappings F : S3\S∗ → S3 with topological barriers at points of a dense subset S∗ ⊂ S2 ⊂ S3. Due to its importance for understanding of results of this paper, in the Appendix we will give some details of this construction based on properties of non-trivial compact 4- dimensional cobordisms M4 with symmetric boundary components – see [7–9]. The interiors of these 4-cobordisms have complete 4-dimensional real hyperbolic structures and universally covered by the real hyperbolic space H4, while the boundary components of M4 have (symmetric) 3- dimensional conformally flat structures obtained by deformations of the same hyperbolic 3-manifold whose fundamental group Γ is a uniform lat- tice in IsomH3. Such conformal deformations of hyperbolic manifolds are well understood after their discovery in [3], see [6]. Nevertheless till recently such “symmetric” hyperbolic 4-cobordisms were unknown de- spite our well known constructions of non-trivial hyperbolic homology 4- cobordisms with very assymmetric boundary components – see [10] and [4–6]. The above subset S∗ of the boundary sphere S2 = ∂B3 is a count- able Γ-orbit of a Cantor subset with Hausdorff dimension ln 5/ ln 6 ≈ 0.89822444 (where a uniform hyperbolic lattice Γ conformally acts in the unit ball B3). All its points are essential singularities of the bounded locally homeomorphic quasiregular mapping f : B3 → R3 defined as the restriction to the unit ball B3 of our quasiregular mapping F : S3\S∗ → S3. This bounded quasiregular mapping f : B3 → R3 has no radial limits at all points x ∈ S∗ ⊂ S2 = ∂B3 and gives an advance to the well known Pierre Fatou’s problem on the correct analogue for higher- dimensional quasiregular mappings of the Fatou’s theorem [13] on radial limits of a bounded analytic function of the unit disc. There are several results concerning radial limits of mappings of the unit ball. The most recent progress is due to Kai Rajala who in particular proved that radial limits exist for infinitely many points of the unit sphere, see [19] and references there for some earlier results in this direction. 12 Bounded Locally Homeomorphic Quasiregular Mappings Another application of our construction Apanasov [9] (Theorem 4.1) of surjective locally homemorphic quasiregular mappings F : S3\S∗ → S3 is to a well known open problem on injectivity of quasiregular map- pings in space formulated by Matti Vourinen in 1970–1980s (see Vuorinen [24–25], [26] (page 193, Problem 4) and Problem 7.66 in the Hayman’s list of problems [11, 16]). The problem asks whether a proper quasiregu- lar mapping f in the unit ball Bn, n ≥ 3, with a compact branching set Bf ⊂ Bn is injective. It is false when n = 2. The conjecture is known to be true in the special case f(Bn) = Bn, n ≥ 3 – see Vuorinen [24]. In Section 2 we show that the quasiregular mapping f : B3 → R3 defined as the restriction to the unit ball B3 of our quasiregular map- ping F : S3\S∗ → S3 from Apanasov [9] (Theorem 4.1) is essentially a counter-example to this conjecture. This mapping f is essentially proper in the sense that any compact C ⊂ f(B3) has a compact subset C ′ ⊂ B3 such that f(C ′) = C. This mapping f is bounded locally homeomorphic but not injective quasiregular mapping in the unit ball. Its restriction to a round ball Br ⊂ B3 of radius r < 1 arbitrary close to one gives (after re-scaling) a proper bounded quasiregular mapping of the round ball B3 serving as a counter-example to this Vuorinen conjecture (Theorem 2.2). The last task of this paper is to investigate the asymptotic behavior of bounded locally homeomorphic quasiregular mappings in the unit ball in smaller balls Bn(r) ⊂ Bn of radius r close to one. There is an open Matti Vuorinen conjecture that in dimension n ≥ 3 it is not possible that for y ∈ f(Bn) and all r ∈ (1/2, 1), the cardinality card(Bn(r) ∩ (f−1(y))) > 1 (1− r)n−1 (1.1) In Section 3 we show that this question is closely related to the growth function of the kernel (a free group of rank 3) F3 ⊂ Γ of the homomor- phism ρ :Γ → G of our hyperbolic lattice Γ ⊂ IsomH3 to the constructed discrete group G ⊂ IsomH4 – see [9]. We conclude that our bounded locally homeomorphic quasiregular mappings in the unit ball B3 satisfies this conjecture. 1.1. Acknowledgments The author is grateful to Matti Vuorinen for attracting our attention to these problems and for fruitful discussions. 2. Not injective bounded quasiregular mappings Here we apply our construction [9] (see Appendix: Theorems 4.1 and 4.5) of bounded locally homeomorphic quasiregular mapping F :B3 → R3 B. N. Apanasov 13 to solve the Matti Vourinen open problem on injectivity of quasiregular mappings in 3-dimensional space. This well known problem was for- mulated by Matti Vourinen in 1970–1980s as result of investigations of quasiregular mappings in space (see Vuorinen [24–25], [26] (page 193, Problem 4) and Problem 7.66 in the Hayman’s list of problems [11, 16]). The problem asks whether a proper quasiregular mapping f in the unit ball Bn, n ≥ 3, with a compact branching set Bf ⊂ Bn is injective. The mapping f(z) = z2, where z ∈ B2, shows that the conjecture is false when n = 2. The conjecture is known to be true in the special case f(Bn) = Bn, n ≥ 3 – see Vuorinen [24]. Here we give a counter-example to this conjecture for n = 3. Proposition 2.1. Let the uniform hyperbolic lattice Γ ⊂ IsomH3 and its discrete representation ρ :Γ → G ⊂ IsomH4 with the kernel as a free subgroup F3 ⊂ Γ be as in Theorem 4.1. Then the bounded locally home- omorphic quasiregular mapping F : B3 → R3 constructed in Theorem 4.5 as a Γ-equivariant mapping in (4.3) is an essentially proper bounded quasiregular mapping in the unit 3-ball B3 which is locally homeomorphic (BF = ∅) but not injective. Proof. The discrete group G = ρ(Γ) ⊂ IsomH4 ∼= Möb(3) has its invariant bounded connected component Ω1 ⊂ Ω(G) ⊂ S3 where its fundamental polyhedron P1 is quasiconformally homeomorphic to the convex hyperbolic polyhedron P0 fundamental for our hyperbolic lattice Γ ⊂ IsomH3 conformally acting in the unit ball B3(0, 1)), φ−1 1 :P0 → P1. This homeomorphism φ−1 1 from (4.2) maps polyhedral sides of P0 to the corresponding sides of the polyhedron P1 and preserves all dihedral angles. Our bounded locally homeomorphic quasiregular mapping F :B3 → Ω1 ⊂ R3 defined in (4.3) as the equivariant extention of this homeomor- phism φ−1 1 maps the tessellation of B3 by compact Γ-images of P0 to the tessellation of Ω1 by compact G-images of P1. This shows that for any compact subset C ⊂ Ω1 = F (B3) (covered by finitely many polyhedra g(P1), g ∈ G), there is a compact subset C ′ ⊂ B3 (covered by finitely many corresponding polyhedra γ(P1), γ ∈ Γ) such that F (C ′) = C. On the other hand, this locally homeomorphic quasiregular mapping F is not injective in B3. In fact, for any element γ 6= 1 in the kernel F3 ⊂ Γ of the homomorphism ρ : Γ → G the image F (P0) = P1 of the fundamental polyhedron P0 of the lattice Γ is the same as the image F (γ(P0)) of the translated polyhedron γ(P0), P0 ∩ γ(P0) = ∅. One may restrict our not injective essentially proper bounded quasi- regular mapping F in the unit 3-ball B3 from Proposition 2.1 to a round 14 Bounded Locally Homeomorphic Quasiregular Mappings ball Br ⊂ B3 of radius r < 1 arbitrary close to one. The composition of this restriction Fr :Br → R3 with the stretching of Br to the unit ball B3, i.e. the mapping f :B3 → R 3 , f(x) = Fr(rx) (2.1) is a proper bounded locally homeomorphic quasiregular mapping of the unit ball B3. For any point y ∈ f(B3) the number Ny of its pre-images in B3, i.e. the cardinality of the set {x ∈ B3 :f(x) = y} is finite. In fact due to Vuorinen Lemma 9.22 in [26], this number Ny is independent of y ∈ f(B3) in the image set and is set as Nf . The number Nf of such pre-images of y ∈ f(B3) is determined by the number of images γ(Pker), γ ∈ F3 ⊂ Γ, of a fundamental polyhedron Pker ⊂ H3 in our round ball Br ⊂ B3 of radius r < 1 defining the mapping f in (2.1). Here F3 is the free subgroup F3 ⊂ Γ in the uniform hyperbolic lattice Γ ⊂ IsomH3 (the kernel of the discrete representation ρ : Γ → G ⊂ IsomH4 from Theorems 2.1 and 4.1), and Pker ⊂ H3 is its fundamental polyhedron in the hyperbolic space H3. Making the radius r < 1 sufficiently close to 1 (i.e. changing our quasiregular mapping f :B3 → R3), one can make the number Nf arbitrary large. This proves the following (the Vuorinen conjecture’ counter-example): Theorem 2.2. There are proper bounded quasiregular mappings f :B3 → R3 without branching sets (Bf = ∅) which are locally homeomorphic but not injective. Their pre-images {x ∈ B3 :f(x) = y} are finite and can be made arbitrary large. 3. Asymptotics of bounded quasiregular mappings in the unit ball and growth in free groups Here we investigate the asymptotic behavior of bounded locally home- omorphic quasiregular mappings f in the unit ball. The question is how many pre-images of a point y ∈ f(Bn) do we have in smaller balls Bn(r) ⊂ Bn of radius r close to one. There is an open Matti Vuorinen conjecture that in dimension n ≥ 3 it is not possible that for y ∈ f(Bn) and all r ∈ (1/2, 1), the cardinality of such pre-image in Bn(r) is bigger than (1− r)1−n – see (1.1). As we show this question for our bounded quasiregular mappings in the unit ball B3, F :B3 → R3, constructed in Theorem 4.5 is closely re- lated to the growth function of the kernel F3 ⊂ Γ of the homomorphism ρ : Γ → G of our uniform hyperbolic lattice Γ ⊂ IsomH3 to the con- structed discrete group G ⊂ IsomH4 – see Proposition 4.3 and Lemma 4.4 in Appendix. Here F3 is a free group on 3 generators. B. N. Apanasov 15 For free groups Fm on m generators one can use well known facts about their growth functions, cf. [14]. The growth function γG,Σ of a group (G,Σ) with a generating set Σ counts the number of elements in G whose length (in the word metric) is at most a natural number k: γG,Σ(k) = card{g ∈ G : |g|Σ ≤ k} . (3.1) Lemma 3.1. A free group Fm on m generators (for any free system Σ of generators) has 2m(2m − 1)k−1 elements of length k, and its growth function: γFm(k) = 1 + m m− 1 ((2m− 1)k − 1). (3.2) Proof. Clearly in a free group Fm on m generators we have the number of elements with length i equals to ci = card{g ∈ Fm : |g| = i} = 2m(2m−1)i. Therefore the growth function γFm(k) = 1+2m+2m(2m− 1)+ . . .+2m(2m−1)k−1. This gives the value γFm(k) in the Lemma. In the embedded Cayley graph ϕ(K(Γ,Σ)) ⊂ B3 (i.e. the graph that is dual to the tessellation of B3 by convex hyperbolic polyhedra γ(P0), γ ∈ Γ), we consider its subgraph (a tree) corresponding to our free group F3 ⊂ Γ on 3 generators (the kernel of the homomorphism ρ). The embedding ϕ of the Cayley graph K(Γ,Σ) is a Γ-equivariant proper embedding. It is a pseudo-isometry, i.e. for the word metric (∗, ∗) on K(Γ,Σ) and the hyperbolic metric d in the unit ball B3, there are positive constants K and K ′ such that (a, b)/K ≤ d(ϕ(a), ϕ(b)) ≤ K · (a, b) for all a, b ∈ K(Γ,Σ) satisfying one of the following two conditions: either (a, b) ≥ K ′ or d(ϕ(a), ϕ(b)) ≥ K ′. LetD be the maximum of hyperbolic length of generators of the kernel F3 ⊂ Γ. All vertices of our tree subgraph corresponding to elements in F3 of length at most k are in the hyperbolic ball centered at 0 ∈ B3 with radius R = Dk. This hyperbolic ball corresponds to the Euclidean ball B3(0, r) ⊂ B3(0, 1) of radius r = (eR − 1)/(eR + 1). Multiplying (1.1) by (1− r)n−1, we see that we need to estimate the asymptotics of (1− r)n−1 card(Bn(r) ∩ (F−1(y))). (3.3) for arbitrary small ǫ = (1−r), or for arbitrary large λ = ln((2/(1−r))−1). In the case of our free group F3 on 3 generators (the kernel of the homomorphism ρ), Lemma 3.1 shows that the growth function γF3(k) = 1+3(5k−1)/2. This reduces the asymptotics of (3.3) to the asymptotics of 3(5λ/D−1)/e2λ for arbitrary large λ. Since the last expression tends to 0 when λ tends to ∞, we conclude that our bounded locally homeomorphic 16 Bounded Locally Homeomorphic Quasiregular Mappings quasiregular mappings F :B3 → Ω1 ⊂ R3 in the unit ball B3 satisfy the Vuorinen conjecture (1.1). Remark 3.2. There is an important observation. If in our analysis of the asymptotics of (3.3) (and in our construction of groups Γ and G) the ker- nel of the corresponding homomorphism ρ :Γ → G ⊂ IsomH4 were a free subgroup Fm on a big number m of generators, then our last expression would tend to ∞ when λ tends to ∞. This would provide a way to con- structing a similar bounded locally homeomorphic quasiregular mapping in the unit ball giving a possible counter example to (1.1). 4. Appendix: Hyperbolic 4-cobordisms and deformations of hyperbolic structures For the readers convenience, here we provide essential details of our construction [9] of .locally homeomorphic quasiregular surjective map- pings F :S3\S∗ → S3 based on the properties of non-trivial “symmmetric” hyperbolic 4-cobordisms M4 = (H4 ∪Ω(G))/G constructed in Apanasov [7]. Properties of the fundamental group π1(M 4) ∼= G ⊂ IsomH4 of such “symmmetric” hyperbolic 4-cobordisms M4 = H4/G acting discretely in the hyperbolic 4-space H4 and in the discontinuity set Ω(G) ⊂ ∂H4 = S3 are very essential for our construction of the quasiregular mapping F . We start with our construction [7] of such discrete group G ⊂ IsomH4 and the corresponding discrete representation ρ :Γ → G of a uniform hy- perbolic lattice Γ ⊂ IsomH3. These discrete groups G and Γ produce a non-trivial (not a product) hyperbolic 4-cobordismsM4 = (H4∪Ω(G))/G whose boundary components N1 and N2 are topologically and geomet- rically symmetric to each other. These N1 and N2 are covered by two G-invariant connected components Ω1 and Ω2 of the discontinuity set Ω(G) ⊂ S3, Ω(G) = Ω1 ∪Ω2. The conformal action of G = ρ(Γ) in these components Ω1 and Ω2 is symmetric and has contractible fundamental polyhedra P1 ⊂ Ω1 and P2 ⊂ Ω2 of the same combinatorial type allowing to realize them as a compact polyhedron P0 in the hyperbolic 3-space, i.e. the dihedral angle data of these polyhedra satisfy the Andreev’s conditions [1]. Nevertheless this geometric symmetry of boundary com- ponents of our hyperbolic 4-cobordism M4(G)) does not make the group G = π1(M 4) quasi-Fuchsian, and our 4-cobordism M4 is non-trivial. Here a Fuchsian group Γ ⊂ IsomH3 ⊂ IsomH4 conformally acts in the 3-sphere S3 = ∂H4 and preserves a round ball B3 ⊂ S3 where it acts as a cocompact discrete group of isometries of H3. Due to the Sullivan structural stability (see Sullivan [22] for n = 2 and Apanasov [4], Theorem 7.2), the space of quasi-Fuchsian representations of a hyperbolic B. N. Apanasov 17 lattice Γ ⊂ IsomH3 into IsomH4 is an open connected component of the Teichmüller space of H3/Γ or the variety of conjugacy classes of discrete representations ρ :Γ → IsomH4. Points in this (quasi-Fuchsian) component of the variety correspond to trivial hyperbolic 4-cobordisms M(G) where the discontinuity set Ω(G) = Ω1 ∪ Ω2 ⊂ S3 = ∂H4 is the union of two topological 3-balls Ωi, i = 1, 2, and M(G) is homeomorphic to the product of N1 and the closed interval [0, 1]. We may consider hyperbolic 4-cobordisms M(ρ(Γ)) corresponding to uniform hyperbolic lattices Γ ⊂ IsomH3 generated by reflections. Natu- ral inclusions of these lattices into IsomH4 act at infinity ∂H4 = S3 as Fuchsian groups Γ ⊂ Möb(3) preserving a round ball B3 ⊂ S3. In this case our construction of the mentioned discrete groups Γ and G = ρ(Γ) results in the following (see Apanasov [7]): Theorem 4.1. There exists a discrete Möbius group G ⊂ Möb(3) on the 3-sphere S3 generated by finitely many reflections such that: 1. Its discontinuity set Ω(G) is the union of two invariant components Ω1, Ω2; 2. Its fundamental polyhedron P ⊂ S3 has two contractible compo- nents Pi ⊂ Ωi, i = 1, 2, having the same combinatorial type (of a compact hyperbolic polyhedron P0 ⊂ H3); 3. For the uniform hyperbolic lattice Γ ⊂ IsomH3 generated by re- flections in sides of the hyperbolic polyhedron P0 ⊂ H3 and act- ing on the sphere S3 = ∂H4 as a discrete Fuchsian group i(Γ) ⊂ IsomH4 = Möb(3) preserving a round ball B3 (where i : IsomH3 ⊂ IsomH4 is the natural inclusion), the group G is its image under a homomorphism ρ :Γ → G but it is not quasiconformally (topolog- ically) conjugate in S3 to i(Γ). Construction: We define a finite collection Σ of reflecting 2-spheres Si ⊂ S3, 1 ≤ i ≤ N . As the first three spheres S1, S2 and S3 we take the coordinate planes {x ∈ R3 : xi = 0}, and S4 = S2(0, R) is the round sphere of some radius R > 0 centered at the origin. The value of the radius R will be determined later. Let B = ⋃ 1≤i≤4Bi be the union of the closed balls bounded by these four spheres, and let ∂B be its boundary (a topological 2-sphere) having four vertices which are the intersection points of four triples of our spheres. We consider a simple closed loop α ⊂ ∂B which does not contain any of our vertices and which symmetrically separates two pairs of these vertices from each other as the white loop does on the tennis ball. This loop α can be considered 18 Bounded Locally Homeomorphic Quasiregular Mappings as the boundary of a topological 2-disc σ embedded in the complement D = S3\B of our four balls. Our geometric construction needs a detailed description of such a 2-disc σ and its boundary loop α = ∂σ obtained as it is shown in Figure 1. The desired disc σ ⊂ D = S3 \B can be described as the boundary in the domain D of the union of a finite chain of adjacent blocks Qi (regular cubes) with disjoint interiors whose centers lie on the coordinate planes S1 and S2 and whose sides are parallel to the coordinate planes. This chain starts from the unit cube whose center lies in the second coordinate axis, in e2 ·R+ ⊂ S1∩S3. Then our chain goes up through small adjacent cubes centered in the coordinate plane S1, at some point changes its direction to the horizontal one toward the third coordinate axis, where it turns its horizontal direction by a right angle again (along the coordinate plane S2), goes toward the vertical line passing through the second unit cube centered in e1 ·R+ ⊂ S2∩S3, then goes down along that vertical line and finally ends at that second unit cube, see Figure 1. We will define the size of small cubes Qi in our block chain and the distance of the centers of two unit cubes to the origin in the next step of our construction. Figure 1: Configuration of blocks and the white loop α ⊂ ∂B. B. N. Apanasov 19 Figure 2: Big and small cube sizes and ball covering. Let us consider one of our cubes Qi, i.e. a block of our chain, and let f be its square side having a nontrivial intersection with our 2-disc σ ⊂ D. For that side f we consider spheres Sj centered at its vertices and having a radius such that each two spheres centered at the ends of an edge of f intersect each other with angle π/3. In particular, for the unit cubes such spheres have radius √ 3/3. From such defined spheres we select those spheres that have centers in our domain D and then include them in the collection Σ of reflecting spheres. Now we define the distance of the centers of our big (unit) cubes to the origin. It is determined by the condition that the sphere S4 = S2(0, R) is orthogonal to the sphere Sj ∈ Σ centered at the vertex of such a cube closest to the origin. As in Figure 2, let f be a square side of one of our cubic blocks Qi having a nontrivial intersection fσ = f ∩ σ with our 2-disc σ ⊂ D. We consider a ring of four spheres Si whose centers are interior points of f which lie outside of the four previously defined spheres Sj centered at vertices of f and such that each sphere Si intersects two adjacent spheres Si−1 and Si+1 (we numerate spheres Si mod 4) with angle π/3. In addition these spheres Si are orthogonal to the previously defined ring of bigger spheres Sj, see Figure 2. From such defined spheres Si we select those spheres that have nontrivial intersections with our domain D outside the previously defined spheres Sj, and then include them in the collection Σ of reflecting spheres. If our side f is not the top side of one of the two unit cubes we add another sphere Sk ∈ Σ. It is centered at the center 20 Bounded Locally Homeomorphic Quasiregular Mappings of this side f and is orthogonal to the four previously defined spheres Si with centers in f , see Figure 2. Now let f be the top side of one of the two unit cubes of our chain. Then, as before, we consider another ring of four spheres Sk. Their centers are interior points of f , lie outside of the four previously defined spheres Si closer to the center of f and such that each sphere Sk intersects two adjacent spheres Sk−1 and Sk+1 (we numerate spheres Sk mod 4) with angle π/3. In addition these new four spheres Sk are orthogonal to the previously defined ring of bigger spheres Si, see Figure 2. We note that the centers of these four new spheres Sk are vertices of a small square fs ⊂ f whose edges are parallel to the edges of f , see Figure 2. We set this square fs as the bottom side of the small cubic box adjacent to the unit one. This finishes our definition of the family of twelve round spheres whose interiors cover the square ring f\fs on the top side of one of the two unit cubes in our cube chain and tells us which two spheres among the four new defined spheres Sk were already included in the collection Σ of reflecting spheres (as the spheres Sj ∈ Σ associated to small cubes in the first step). This also defines the size of small cubes in our block chain. Now we can vary the remaining free parameter R (which is the radius of the sphere S4 ∈ Σ) in order to make two horizontal rows of small blocks with centers in S1 and S2, correspondingly, to share a common cubic block centered at a point in e3 · R+ ⊂ S1 ∩ S2, see Figure 1. We can use the collection Σ of reflecting spheres Si to define a discrete reflection group G = GΣ ⊂ Möb(3). Important properties of Σ are: (1) the closure of the disc σ ⊂ D is covered by balls Bj ; (2) any two spheres Sj , Sj′ ∈ Σ either are disjoint or intersect with angle π/2 or π/3; (3) the complement of all balls Bj, 1 ≤ j ≤ N , is the union of two disjoint contractible polyhedra P1 and P2 of the same combinatorial type and equal corresponding dihedral angles. So the discontinuity set Ω(G) ⊂ S3 of G consists of two invariant connected components Ω1 and Ω2 which are the unions of the G-orbits of P̄1 and P̄2, and this defines a Heegaard splitting of the 3-sphere S3 (see [9]): Lemma 4.2. The splitting of the discontinuity set Ω ⊂ S3 of our discrete reflection group G = GΣ ⊂ Möb(3) into G-invariant components Ω1 and Ω2 defines a Heegaard splitting of the 3-sphere S3 of infinite genus with ergodic word hyperbolic group G action on the separating boundary Λ(G). To finish our construction in Theorem 4.1 we notice that the com- binatorial type (with magnitudes of dihedral angles) of the bounded component P1 of the fundamental polyhedron P ⊂ S3 coincides with B. N. Apanasov 21 the combinatorial type of its unbounded component P2. Applying An- dreev’s theorem on 3-dimensional hyperbolic polyhedra [1], one can see that there exists a compact hyperbolic polyhedron P0 ⊂ H3 of the same combinatorial type with the same dihedral angles (π/2 or π/3). So one can consider a uniform hyperbolic lattice Γ ⊂ IsomH3 generated by reflections in sides of the hyperbolic polyhedron P0. This hyperbolic lat- tice Γ acts in the sphere S3 as a discrete co-compact Fuchsian group i(Γ) ⊂ IsomH4 = Möb(3) (i.e. as the group i(Γ) ⊂ IsomH4 where i : IsomH3 ⊂ IsomH4 is the natural inclusion) preserving a round ball B3 and having its boundary sphere S2 = ∂B3 as the limit set. Obviously there is no self-homeomorphism of the sphere S3 conjugating the action of the groups G and i(Γ) because the limit set Λ(G) is not a topological 2-sphere. So the constructed group G is not a quasi-Fuchsian group. � One can construct a natural homomorphism ρ : Γ → G, ρ ∈ R3(Γ), between these two Gromov hyperbolic groups Γ ⊂ IsomH3 and G ⊂ IsomH4 defined by the correspondence between sides of the hyperbolic polyhedron P0 ⊂ H3 and reflecting spheres Si in the collection Σ bound- ing the fundamental polyhedra P1 and P2. Then we have: Proposition 4.3. The homomorphism ρ ∈ R3(Γ), ρ : Γ → G, in The- orem 4.1 is not an isomorphism. Its kernel ker(ρ) = ρ−1(eG) is a free rank 3 subgroup F3 � Γ. Its proof (see [9], Prop.2.4) is based on the following statement (see [9], Lemma 2.5) in combinatorial group theory: Lemma 4.4. Let A = 〈a1, a2 | a21, a22, (a1a2)2〉 ∼= B = 〈b1, b2 | b21, b22, (b1b2) 2〉 ∼= C = 〈c1, c2 | c21, c22, (c1c2)2〉 ∼= Z2 × Z2, and let ϕ :A ∗B → C be a homomorphism of the free product A ∗B into C such that ϕ(a1) = ϕ(b1) = c1 and ϕ(a2) = ϕ(b2) = c2. Then the kernel ker(ϕ) = ϕ−1(eC) of ϕ is a free rank 3 subgroup F3�A∗B generated by elements x = a1b1, y = a2b2 and z = a1a2b2a1 = a1ya1. 4.1. Bending homeomorphisms between polyhedra Here we sketch our construction of a quasiconformal homeomorphism φ1 :P1 → P0 between the fundamental polyhedron P1 ⊂ Ω1 ⊂ Ω(G) ⊂ S3 for the group G action in Ω1 and the fundamental polyhedron P0 ⊂ B3 for conformal action of our hyperbolic lattice Γ ⊂ IsomH3 from Theorem 4.1. This mapping φ1 is a composition of finitely many elementary "bending homeomorphisms". It maps faces to faces, and preserve the combinatorial structure of the polyhedra and their corresponding dihedral angles. 22 Bounded Locally Homeomorphic Quasiregular Mappings First we observe that to each cube Qj , 1 ≤ j ≤ m, used in the above construction of the group G (see Figure 1), we may associate a round ball Bj centered at the center of the cube Qj and such that its bound- ary sphere is orthogonal to the reflection spheres Si from our generating family Σ whose centers are at vertices of the cube Qj. In particular for the unit cubes Q1 and Qm, the reflection spheres Si centered at their ver- tices have radius √ 3/3, so the balls B1 and Bm (whose boundary spheres are orthogonal to those corresponding reflection spheres Si) should have radius √ 5/12. Also we add another extra ball B3(0, R) (which we con- sider as two balls B0 and Bm+1) whose boundary is the reflection sphere S2(0, R) = S4 ∈ Σ centered at the origin and orthogonal to the closest reflection spheres Si centered at vertices of two unit cubes Q1 and Qm. Our different enumeration of this ball will be used when we consider dif- ferent faces of our fundamental polyhedron P1 lying on that reflection sphere S4. Now for each cube Qj , 1 ≤ j ≤ m, we may associate a discrete sub- group Gj ⊂ G ⊂ Möb(3) ∼= IsomH4 generated by reflections in the spheres Si ∈ Σ associated to that cube Qj - see our construction in Theo- rem 4.1. One may think about such a group Gj as a result of quasiconfor- mal bending deformations (see [4], Chapter 5) of a discrete Möbius group preserving the round ball Bj associated to the cube Qj (whose center coincides with the center of the cube Qj). As the first step in such defor- mations, we define two quasiconformal “bending” self-homeomorphisms of S3, f1 and fm+1, preserving the balls B1, . . . , Bm and the set of their reflection spheres Si, i 6= 4, and transferring ∂B0 and ∂Bm+ 1 into 2- spheres orthogonally intersecting ∂B1 and ∂Bm along round circles b1 and bm+1, respectively - see (3.1) and Figure 6 in [9]. In the next steps in our bending deformations, for two adjacent cubes Qj−1 and Qj, let us denote Gj−1,j ⊂ G the subgroup generated by re- flections with respect to the spheres Si ⊂ Σ centered at common vertices of these cubes. This subgroup preserves the round circle bj = bj−1,j = ∂Bj−1 ∩ ∂Bj . This shows that our group G is a result of the so called "block-building construction" (see [4], Section 5.4) from the block groups Gj by sequential amalgamated products: G = G1 ∗ G1,2 G2 ∗ G2,3 · · · ∗ Gj−2,j−1 Gj−1 ∗ Gj−1,j Gj ∗ Gj,j+1 · · · ∗ Gm−1,m Gm (4.1) The chain of these building balls {Bj}, 1 ≤ j ≤ m, contains the bounded polyhedron P1 ⊂ Ω1. For each pair Bi−1 and Bi with the common boundary circle bi = ∂Bi−1 ∩ ∂Bi, 1 ≤ i ≤ m, we construct a quasi-conformal bending homeomorphism fi that transfers Bi ∪ Bi−1 onto the ball Bi and which is conformal in dihedral ζi-neighborhoods B. N. Apanasov 23 of the spherical disks ∂Bi\Bi−1 and ∂Bi−1\Bi - see (3.3) and Figure 7 in [9]. In each i-th step, 2 ≤ i ≤ m, we reduce the number of balls Bj in our chain by one. The composition fm+1fifi−1 · · · f2f1 transfers all spheres from Σ to spheres orthogonal to the boundary sphere of our last ball Bm which we renormalize as the unit ball B(0, 1). We note that all intersection angles between these spheres do not change. We define our quasiconformal homeomorphism φ1 :P1 → P0 (4.2) as the restriction of the composition fm+1fmfm−1 · · · f2f1 of our bending homeomorphisms fj on the fundamental polyhedron P1 ⊂ Ω1. 4.2. Bounded locally homeomorphic quasiregular mappings Now we define bounded quasiregular mappings F : B3 → R3 as in Theorem 4.1 in [9]: Theorem 4.5. Let the uniform hyperbolic lattice Γ ⊂ IsomH3 and its discrete representation ρ : Γ → G ⊂ IsomH4 with the kernel as a free subgroup F3 ⊂ Γ be as in Theorem 4.1. Then there is a bounded locally homeomorphic quasiregular mapping F : B3 → R3 whose all singular- ities lie in an exceptional subset S∗ of the unit sphere S2 ⊂ R3 and form a dense in S2 Γ-orbit of a Cantor subset with Hausdorff dimension ln 5/ ln 6 ≈ 0.89822444. These (essential) singularities create a barrier for F in the sense that at points x ∈ S∗ the map F does not have radial limits. Construction: We construct our quasiregular mapping F :B3 → Ω1 = F (B3) in the unit ball B3 by equivariant extention of the above quasicon- formal homeomorphism φ−1 1 :P0 → P1 which maps polyhedral sides of P0 to the corresponding sides of the polyhedr P1 and preserves combinatorial structures of polyhedra as well as their dihedral angles: F (x) = ρ(γ) ◦ φ−1 1 ◦ γ−1(x) if |x| < 1, x ∈ γ(P0), γ ∈ Γ (4.3) The tesselations of B3 and Ω1 by corresponding Γ- and G-images of their fundamental polyhedra Po and P1 are perfectly similar. This implies that our quasiregular mapping F defined by (4.3) is bounded and locally homeomorphic. It follows from Lemma 4.2 that the limit set Λ(G) ⊂ S3 of the group G ⊂ Möb(3) defines a Heegard splitting of infinite genus of the 3-sphere S3 into two connected components Ω1 and Ω2 of the discontinuity set Ω(G). The action of G on the limit set Λ(G) is an ergodic word hyperbolic action. For this ergodic action the set of fixed 24 Bounded Locally Homeomorphic Quasiregular Mappings points of loxodromic elements g ∈ G (conjugate to similarities in R3) is dense in Λ(G). Preimages γ ∈ Γ of such loxodromic elements g ∈ G for our homomorphism ρ : Γ → G are loxodromic elements in Γ with two fixed points p, q ∈ Λ(Γ) = S2, p 6= q. This and Tukia’s arguments of the group completion (see [23] and [4], Section 4.6) show that our mapping F can be continuously extended to the set of fixed points of such elements γ ∈ Γ, F (Fix(γ)) = Fix(ρ(γ)). The sense of this continuous extension is that if γ ∈ Γ is a loxodromic preimage of a loxodromic element g ∈ G, ρ(γ) = g, and if x ∈ S3\S2 tends to its fixed points p or q along the hyperbolic axis of γ (in B(0, 1) or in its complement B̂(0, 1)) (i.e. radially) then lim|x|→1 F (x) exists and equals to the corresponding fixed point of the loxodromic element g = ρ(γ) ∈ G. In that sense one can say that the limit set Λ(G) (the common boundary of the connected components Ω1,Ω2 ⊂ Ω(G)) is the F -image of points in the unit sphere S2 ⊂ S3. So the mapping F is extended to a map onto the closure Ω1 = Ω1 ∪ Λ(G) ⊂ R3. Nevertheless not all loxodromic elements γ ∈ Γ in the hyperbolic lat- tice Γ ⊂ IsomH3 have their images ρ(γ) ∈ G as loxodromic elements. Proposition 4.3 shows that ker ρ ∼= F3 is a free subgroup on three gener- ators in the lattice Γ, and all elements γ ∈ F3 are loxodromic. Now we look at radial limits limx→p F (x) when x radially tends to a fixed point p ∈ S2 of this loxodromic element γ ∈ F3 ⊂ Γ. LetK(Γ,Σ) be the Cayley graph for a group Γ with a finite generating set Σ. Our lattice Γ ⊂ IsomH3 has an embedding ϕ of its Cayley graph K(Γ,Σ) in the hyperbolic space H3 ∼= B3. For a point 0 ∈ H3 not fixed by any γ ∈ Γ\{1}, vertices γ ∈ K(Γ,Σ) are mapped to γ(0), and edges joining vertices a, b ∈ K(Γ,Σ) are mapped to the hyperbolic geodesic seg- ments [a(0), b(0)]. In other words, ϕ(K(Γ,Σ)) is dual to the tessellation of H3 by polyhedra γ(P0), γ ∈ Γ. The map ϕ is a Γ-equivariant proper embedding: for any compact C ⊂ H3, its pre-image ϕ−1(ϕ(K(Γ,Σ))∩C) is compact. Moreover for any convex cocompact group Γ ⊂ IsomHn this embedding ϕ is a pseudo-isometry (see [12] and [4], Theorem 4.35), i.e. for the word metric on K(Γ,Σ) and the hyperbolic metric d, there are K > 0 and K ′ > 0 such that |a, b|/K ≤ d(ϕ(a), ϕ(b)) ≤ K · |a, b| for all a, b ∈ K(Γ,Σ) such that either |a, b| ≥ K ′ or d(ϕ(a), ϕ(b)) ≥ K ′. This implies (see [4], Theorem 4.38) that the limit set of any convex- cocompact group Γ ⊂ Möb(n) can be identified with its group completion Γ, Γ = K(Γ,Σ) \ K(Γ,Σ). Namely there exists a continuous and Γ- equivariant bijection ϕΓ :Γ → Λ(Γ). For the kernel subgroup F3 = ker ρ ⊂ Γ ⊂ IsomH3 and for the above pseudo-isometric embedding ϕ, we consider its Cayley subgraph in B. N. Apanasov 25 ϕ(K(Γ,Σ)) ⊂ H3 which is a tree - see Figure 5 in [9]. Since the limit set of ker ρ = F3 ⊂ Γ corresponds to the “bondary at infinity” ∂∞F3 of F3 ⊂ Γ (the group completion F3), it is a closed Cantor subset of the unit sphere S2 with Hausdorff dimension ln 5/ ln 6 ∼ 0.89822444. The Γ-orbit Γ(Λ(F3)) of our Cantor set is a dense subset S∗ of S2 = Λ(Γ) because of density in the limit set Λ(Γ) of the Γ-orbit of any limit point. In particular we have such dense Γ-orbit Γ({p, q}) of fixed points p and q of a loxodromic element γ ∈ F3 ⊂ Γ (the images of p and q are fixed points of Γ-conjugates of such loxodromic elements γ ∈ F3 ⊂ Γ). On the other hand let x ∈ lγ where lγ is the hyperbolic axis in B(0, 1) of an element γ ∈ F3 ⊂ Γ. Denoting dγ the translation distance of γ, we have that any segment [x, γ(x)] ⊂ lγ is mapped by our quasiregular mapping F to a non-trivial closed loop F ([x, γ(x)]) ⊂ Ω1, inside of a handle of the handlebody Ω1 (mutually linked with Ω2 - similar to the loops β1 ⊂ Ω1 and β2 ⊂ Ω2 constructed in the proof of Lemma 4.2). Therefore when x ∈ lγ radially tends to a fixed point p (in fix(γ) ∈ S2) of such element γ, its image F (x) goes along that closed loop F ([x, γ(x)]) ⊂ Ω1 because F (γ(x)) = ρ(γ)(F (x)) = F (x). Immediately it implies that the radial limit limx→p F (x) does not exist. This shows that fixed points of any element γ ∈ F3 ⊂ Γ (or its conjugates) are essential (topological) singularities of our quasiregular mapping F . So our quasiregular mapping F has no continuous extension to the subset S∗ ⊂ S2 which is a dense subset of the unit sphere S2 = ∂B3 ⊂ S3. � References [1] E. M. Andreev, On convex polyhedra in Lobachevsky space // Mat. Sbornik, 81 (1970), 445–478 (iIn Russian); Engl. Transl.: Math. USSR Sbornik 10 (1970), 413–440. [2] E. M. Andreev, The intersections of the planes of faces of polyhedra with sharp angles // Mat. Zametki, 8 (1970), 521–527 (in Russian); Engl. Transl.: Math. Notes, 8 (1970), 761–764. [3] B. Apanasov, Nontriviality of Teichmüller space for Kleinian group in space, in: Riemann Surfaces and Related Topics, Proc. 1978 Stony Brook Conference (I. Kra and B. Maskit, eds), Ann. of Math. Studies, 97, Princeton Univ. Press, 1981, 21–31. [4] B. Apanasov, Conformal geometry of discrete groups and manifolds, De Gruyter Exp. Math. 32, W. de Gruyter, Berlin–New York, 2000. [5] B. Apanasov, Quasisymmetric embeddings of a closed ball inextensible in neigh- borhoods of any boundary points // Ann. Acad. Sci. Fenn., Ser. A I Math, 14 (1989), 243–255. [6] B. Apanasov, Nonstandard uniformized conformal structures on hyperbolic man- ifolds // Invent. Math., 105 (1991), 137–152. 26 Bounded Locally Homeomorphic Quasiregular Mappings [7] B. Apanasov, Hyperbolic 4-cobordisms and group homomorphisms with infinite kernel // Atti Semin. Mat. Fis. Univ. Modena Reggio Emilia, 57 (2010), 31–44. [8] B. Apanasov, Group Actions, Teichmüller Spaces and Cobordisms // Lobachevskii J. Math., 38 (2017), 213–228. [9] B. Apanasov, Topological barriers for locally homeomorphic quasiregular mappings in 3-space // Ann. Acad. Sci. Fenn. Math., 43 (2018), 579–596. [10] B. N. Apanasov, A. V. Tetenov, Nontrivial cobordisms with geometrically finite hyperbolic structures // J. of Diff. Geom., 28 (1988), 407–422. [11] K. F. Barth, D. A. Brannan, W. K. Hayman, Research problems in complex analysis // Bull. London Math. Soc., 16 (1984), No. 5, 490–517. [12] J. W. Cannon, The combinatorial structure of cocompact discrete hyperbolic groups // Geom. Dedicata, 16 (1984), 123–148. [13] P. Fatou, Séries trigonométriques et séries de Taylor // Acta Math., 30 (1906), 335–400 (in French). [14] R. Grigorchuk, Milnor’s problem on the growth of groups and its consequences, in: Frontiers in complex dynamics, Princeton Math. Ser., 51, Princeton Univ. Press, 2014, 705–773. [15] M. Gromov, Metric structures for Riemannian and non-Riemannian spaces, Birkhäuser, Boston, 1999. [16] W. K. Hayman, E. F. Lingham, Research problems in function theory, ArXiv: 1809.07200. [17] M. A. Lavrentiev, On a class of continuous mappings // Mat. Sbornik, 42 (1935), 407–424 (in Russian). [18] O. Martio, S. Rickman, J. Väisälä, Distortion and singularities of quasiregular mappings // Ann. Acad. Sci. Fenn. Ser. A I Math, 465 (1970), 1–13. [19] K. Rajala, Radial limits of quasiregular local homeomorphisms // Amer. J. Math., 130 (2008), 269–289. [20] Y. G. Reshetnyak, Space mappings with bounded distortion, Nauka, Novosibirsk, 1982 (in Russian); Engl. transl.: Transl. Math. Monogr., 73, Amer. Math. Soc., Providence, R.I., 1989. [21] S. Rickman, Quasiregular Mappings, Ergeb. Math. Grenzgeb., 26, Springer, Berlin–Heidelberg, 1993. [22] D. Sullivan, Quasiconformal homeomorphisms and dynamics, II: Structural stabil- ity implies hyperbolicity for Kleinian groups // Acta Math., 155 (1985), 243–260. [23] P. Tukia, On isomorphisms of geometrically Kleinian groups // Publ. Math. IHES, 61 (1985), 171–214. B. N. Apanasov 27 [24] M. Vuorinen, Cluster sets and boundary behavior of quasiregular mappings, Math. Scand., 45 (1979), No. 2, 267–281. [25] M. Vuorinen, Queries No 249 // Notices Amer. Math. Soc., 28 (1981), No. 7, 607. [26] M. Vuorinen, Conformal geometry and quasiregular mappings, Lecture Notes in Math, 1319, Springer, Berlin–Heidelberg, 1988. Contact information Boris Nikolaevich Apanasov Department of Mathematics, University of Oklahoma, Norman, USA E-Mail: apanasov@ou.edu