Representation of Steinitz's lattice in lattices of substructures of relational structures

General conditions under which certain relational structure contains a lattice of substructures isomorphic to Steinitz's lattice are formulated. Under some natural restrictions we consider relational structures with the lattice containing a sublattice isomorphic to the lattice of positive integ...

Повний опис

Збережено в:
Бібліографічні деталі
Опубліковано в: :Algebra and Discrete Mathematics
Дата:2016
Автори: Bezushchak, O., Oliynyk, B., Sushchansky, V.
Формат: Стаття
Мова:Англійська
Опубліковано: Інститут прикладної математики і механіки НАН України 2016
Онлайн доступ:https://nasplib.isofts.kiev.ua/handle/123456789/155237
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:Representation of Steinitz's lattice in lattices of substructures of relational structures / O. Bezushchak, B. Oliynyk, V. Sushchansky // Algebra and Discrete Mathematics. — 2016. — Vol. 21, № 2. — С. 184-201. — Бібліогр.: 19 назв. — англ.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine
_version_ 1860062373742641152
author Bezushchak, O.
Oliynyk, B.
Sushchansky, V.
author_facet Bezushchak, O.
Oliynyk, B.
Sushchansky, V.
citation_txt Representation of Steinitz's lattice in lattices of substructures of relational structures / O. Bezushchak, B. Oliynyk, V. Sushchansky // Algebra and Discrete Mathematics. — 2016. — Vol. 21, № 2. — С. 184-201. — Бібліогр.: 19 назв. — англ.
collection DSpace DC
container_title Algebra and Discrete Mathematics
description General conditions under which certain relational structure contains a lattice of substructures isomorphic to Steinitz's lattice are formulated. Under some natural restrictions we consider relational structures with the lattice containing a sublattice isomorphic to the lattice of positive integers with respect to divisibility. We apply to this sublattice a construction that could be called ``lattice completion''. This construction can be used for different types of relational structures, in particular for universal algebras, graphs, metric spaces etc. Some examples are considered.
first_indexed 2025-12-07T17:05:08Z
format Article
fulltext Algebra and Discrete Mathematics RESEARCH ARTICLE Volume 21 (2016). Number 2, pp. 184–201 © Journal “Algebra and Discrete Mathematics” Representation of Steinitz’s lattice in lattices of substructures of relational structures Oksana Bezushchak, Bogdana Oliynyk and Vitaliy Sushchansky Abstract. General conditions under which certain relational structure contains a lattice of substructures isomorphic to Steinitz’s lattice are formulated. Under some natural restrictions we consider relational structures with the lattice containing a sublattice isomor- phic to the lattice of positive integers with respect to divisibility. We apply to this sublattice a construction that could be called “lattice completion”. This construction can be used for different types of relational structures, in particular for universal algebras, graphs, metric spaces etc. Some examples are considered. 1. Introduction Steinitz’s lattice was introduced at the beginning of the XX century by German mathematician A.Steinitz for describing the structure of subfields of algebraically closed field of prime characteristic [1]. It can be determined as the lattice of supernatural numbers with a relation of the divisibility. Steinitz’s lattice is complete, i.e. for an arbitrary subset of its elements exists the exact lower and the exact upper bounds. It contains a various sublattices including Boolean algebras. So, an existence of such lattice in the lattice of substructures of mathematical structure shows its inner richness that may be a basis for the use of the structure as an universal object for corresponding class of structures of the same type. 2010 MSC: 03G10, 08A02, 03G05. Key words and phrases: relational structure, lattice, supernatural numbers, Boolean algebra. O. Bezushchak, B. Oliynyk, V. Sushchansky 185 In the paper we formulate the general conditions in which certain relational structure has some lattice of substructures, which is isomorphic to Steinitz’s lattice. If these conditions are satisfied, then the isomorphic representation of Steinitz’s lattice in a lattice of substructures of this relational structure is obtained. Under certain natural restrictions it is enough to view structures with the lattice containing sublattice, that is isomorphic to the lattice of positive integers with divisibility. We apply to this sublattice a construction that could be called “lattice completion”. This construction can be used to different types of relational structures, in particular — universal algebra, graphs, metric spaces etc. For example, Steinitz’s lattice is isomorphic to: (i) the lattice of all subfields of algebraic closure of finite field (see, e.g., [2]); (ii) the lattice of Glimm’s subalgebras of limited C∗-algebra (see, e.g., [3], [4]); (iii) lattices of so-called homogeneously symmetric and homogeneously alternating subgroups of symmetric groups of permutations of nat- ural numbers (see, e.g., [5], [6]). Furthermore, some elements of Steinitz’s lattice of substructures in certain relational structures were studied in many works of various authors (see, e.g., [7]–[10]). The paper is organized as follows. In Section 2 we give the definition of Steinitz’s lattice and its basic properties. In Section 3 we present basic information on the theory of relational structures. In Section 4 basic construction of “lattice completion” of relational structures is described. It is shown how to build the certain isomorphism between Steinitz’s lattice and a lattice that occurs as a result of “lattice completion”. In Section 5 we give examples of an application of the construction of “lattice completion” in the theory of infinite groups and semigroups transforma- tions. This makes it possible to introduce new objects namely, groups and semigroups of periodically defined transformations of natural numbers. Last section describes lattices of subspaces of Besicovitch’s space which are constructed by the use of “lattice completion”, and therefore are isomorphic to Steinitz’s lattice. Results of the article were earlier partially announced in the article [10] of third author. All symbols are commonly used in the paper. For determination of indefinite terms we refer readers to [11]–[13]. 186 Representation of Steinitz’s lattice 2. Steinitz’s lattice 2.1. Let N be a set of natural numbers and let P be its subset of primes. Definition 1. Supernatural number (or Steinitz’s number) is called a formal product ∏ p∈P pkp , kp ∈ N ∪ {0, ∞}. (1) Denote by the symbol SN the set of all supernatural numbers. Every natural number is a supernatural number, so that N ⊂ SN. Numbers from SN \N will be called infinite supernatural numbers. Divisible relation | on N in natural way is extended to SN. Namely, for arbitrary supernatural numbers u = ∏ p∈P pkp , v = ∏ p∈P plp , kp, lp ∈ N ∪ {0, ∞}, (2) we get u | v if and only if for all p ∈ P inequalities kp 6 lp hold (it is assumed that ∞ is more than zero and all natural numbers). Main properties of set SN, ordered by the divisible relation |, are characterized by the following lemma. Lemma 1. The ordered set (SN, |) is a lattice. The lattice (SN, |) is a complete one and contains the largest and the smallest elements, that accordingly are such supernatural numbers I = ∏ p∈P p∞ 1 = ∏ p∈P p0. (3) The proof of this statement is not difficult. The exact lower and the exact upper bounds of supernatural numbers u, v, that are given by their decompositions (2), are defined by the equalities u ∨ v = ∏ p∈P pmax(kp,lp) (4) u ∧ v = ∏ p∈P pmin(kp,lp) (5) where max(k, ∞) = ∞, min(k, ∞) = k for k ∈ N ∪ {0}. Definition 2. Lattice (SN, ∧, ∨) will be called Steinitz’s lattice. The following lemma follows from equations (4), (5). O. Bezushchak, B. Oliynyk, V. Sushchansky 187 Lemma 2. Steinitz’s lattice is a complete distributive lattice. In the set of supernatural numbers we select two subsets. Definition 3. Supernatural number u = ∏ p∈P pkp is called complete, if for each p ∈ P there is inclusion kp ∈ {0, ∞}. According to the definition a complete supernatural number u is uniquely determined by the subset O(u) of that primes p from P, for that kp = ∞. The set C of complete supernatural numbers is closed on the operations ∨, ∧ and contains the numbers I and 1 that is defined by (3). Moreover, on the set C one can define the operation of addition, namely the addition ū of the number u ∈ C is called complete supernatural number that is determined by subset O(ū) = P \ O(u). It is obvious, that u ∨ ū = I, u ∧ ū = 1. Lemma 3. The set C with the operations ∨, ∧, ¯ forms a Boolean algebra with 1 as a zero element and I as an unit element. The Boolean algebra (C, ∨, ∧, ¯ ) is isomorphic to the algebra of subsets of a countable set. Proof. Define the mapping ϕ from the algebra of subsets of the set P to the algebra C in such way. For any subset X ⊂ P put ϕ(X ) = u, were O(u) = X . From mentioned above it follows that ϕ is a bijection. In addition, for any X1, X2 ⊂ P for which ϕ(X1) = u1, ϕ(X2) = u2 we have ϕ(X1 ∪ X2) = u1 ∨ u2, ϕ(X1 ∩ X2) = u1 ∧ u2, ϕ(X 1) = u1. So, the mapping ϕ is an isomorphism, and this, in particular, means that (C, ∨, ∧, ¯ ) is a Boolean algebra. Definition 4. A supernatural number is called notsquare one, if indicators of powers in its canonical decomposition (1) takes on only two values 0, 1. Let B be a set of all notsquare supernatural numbers. It is clear, that B is closed regarding to the operations ∨ and ∧. Moreover, on B can also be defined the complement operation by the rule: for the number u = ∏ p∈X p, we define u = ∏ p∈P\X p. The number J = ∏ p∈P p is the largest element in the set B. Lemma 4. Sublattice B of the lattice (SN, ∧, ∨) with the additional op- eration, namely complement one, forms a Boolean algebra, the largest element of which is number J, and the least element is 1. The algebra (B, ∧, ∨, ¯ ) is isomorphic to the subalgebra of subsets of the set P. 188 Representation of Steinitz’s lattice Proof. Define the mapping, which puts the subset X in correspondence to the number ∏ p∈X p. This mapping is an isomorphism. 2.2. The sequence of positive integers χ = 〈k1, k2, . . .〉 will be called divisible if ki | ki+1 for i = 1, 2, . . . . Let DS be a set of the most possible of divisible sequences over N. Definition 5. The sequence χ = 〈ki〉i∈N divides the sequence χ′ = 〈k′ i〉i∈N, if for any i ∈ N there are j ∈ N for which ki | k′ j . Let | denote the divisibility of sequences. The relation | on DS is: (i) reflexive, i.e. χ|χ for arbitrary sequence χ ∈ DS; (ii) transitive, i.e. from χ1|χ2 and χ2|χ3 follows χ1|χ3 for any χ1, χ2, χ3 ∈ DS. But the relation of divisibility is not symmetric or antisymmetric relation. Definition 6. Sequences χ and χ′ are called exactly divisible if at the same time χ|χ′ and χ′|χ. The exactly divisible relation is equivalence on DS, which we denote by the symbol ∼. An arbitrary sequence χ ∈ DS determine a supernatural number char χ (characteristic χ), which is defined thus (i) each member of the sequence χ be a divisor of char χ; (ii) every natural divisor of char χ be a divisor of some member of the sequence χ. For example, if χ = 〈1, p, p2, . . .〉, p ∈ P, then char χ = p∞, and when χ = 〈1, 2!, 3!, 4!, . . .〉, then char χ = I. From the definition of characteristic we get easy Lemma 5. 1) For arbitrary χ1, χ2 ∈ DS the divisibility χ1|χ2 holds if and only if char χ1| char χ2. 2) The sequences χ1, χ2 ∈ DS are exactly divisible if and only if when char χ1 = char χ2. So, sets of exactly divisible sequences are characterized by supernatural numbers, moreover the correspondence between these classes of objects is a bijective. 3. Relational structures 3.1. Recall that n-arity relation over the set A is called an arbitrary subset of Cartesian degree An. A relational structure over the set A is O. Bezushchak, B. Oliynyk, V. Sushchansky 189 called an a pair of the type ℜ = 〈A, {Φkα α }α∈I〉, (6) where I is some set of indices, and for every α ∈ I Φkα α is some relation of the arity kα over A (kα ∈ N ∪ {0}). The set A is called a support of the relational structure ℜ, the set {Φkα α }α∈I is called its signature, and the set (kα)α∈I is called its type. Examples 1) Every directed graph without multiple edges with the set of vertices V and the set of edges E ⊂ V × V is a relational structure (V, E) with the support V and the signature E of the type (2). 2) Every colored graph with the set of vertices V , whose edges are colored in k colors, is a relational structure with the support V and the signature E1, . . ., Ek of the type (2, . . . , 2︸ ︷︷ ︸ k ). 3) Every metric space (X, d) with the set I of values of the metric is a relational structure 〈X, {Dα}α∈I〉, where Dα = {(x, y) | x, y ∈ X, d(x, y) = d(y, x) = α}. This relational structure has the type (2α)α∈I . 4) Every universal algebra £ = 〈A, {ϕkα α }α∈I〉, where ϕkα α : Akα → A is an operation of the arity kα on A, can seen as a relational structure of the form (6) of the type (kα + 1)α∈I , where Φkα+1 α (x1, . . . , xkα , y) occurs if and only if, then ϕkα α (x1, . . . , xkα ) = y. The relational structures ℜ = 〈A, {Φkα α }α∈I〉 and ℜ′ = 〈B, {Ψ lβ β }β∈J〉 have the same type, if the sets I and J have the same cardinality and there is a bijection f : I ↔ J so that for every α ∈ I the equality kα = lf(α) holds. 190 Representation of Steinitz’s lattice Let the relational structures ℜ and ℜ′ have the same type. The bijection F : A → B is called an isomorphism of these structures, if for any α ∈ I the relations Φkα α (x1, x2, . . . , xkα ), x1, x2, . . . , xkα ∈ A, and Ψ lf(α) f(α) (F (x1), F (x2), . . . , F (xkα )) are isomorphic. In particular, an isomorphism of universal algebras or graphs as rela- tional structures is their isomorphism in the usual sense, but an isomor- phism of relational structures related to metric spaces means that these spaces are isometric. An isomorphism of relational structure itself is called an automor- phism. All automorphisms of relational structure ℜ form a group with the operation of superposition of automorphisms, which denoted by the symbol Autℜ and named the group of automorphisms of the structure ℜ. Let A′ be an arbitrary nonempty subset of the set A. For the subset A′ we can consider restriction Φkα α |A′ of the relation Φkα α (α ∈ I) on the set A′: Φkα α |A′(x1, x2, . . . , xkα ) = Φkα α (x1, x2, . . . , xkα ) ∩ (A′)kα . Note, that in this case some restrictions Φkα α |A′ may be equal to empty relations. The relational structure 〈A′, {Φkα α |A′}α∈I〉 is called a substructure of the relational structure ℜ = 〈A, {Φkα α }α∈I〉. An isomorphism of ℜ onto some substructure of the structure ℜ′ is called isomorphic emmbeding of the structure ℜ in the structure ℜ′. 3.2. The partially ordered set (I,>) is called a directed to the right, if for any a, b ∈ I there exist such element c ∈ I, that a 6 c, b 6 c. Definition 7. The family of structures {ℜi}i∈I and embedding fi,k : ℜi → ℜk, i, k ∈ I, i 6 k, satisfying the following requirements: 1) I be a directed to the right partially ordered set; 2) for arbitrary indices i, k ∈ I, i 6 k, there exist emmbedings fi,k : ℜi → ℜk, where fi,i be identity isomorphism; 3) if i, j, k ∈ I and i 6 j 6 k, then fi,j · fj,k = fi,k, is called an inductive family Σ of relational structures over a set of indices I. O. Bezushchak, B. Oliynyk, V. Sushchansky 191 Every inductive family of relational structures Σ = 〈ℜi, fi,k〉i,k∈I defines the limit structure which is called inductive limit of family Σ and is denoted by the symbol ℜ(Σ) = lim −−→ i (ℜi, fi,k), i, k ∈ I. (7) Elements of the structure (7) are the so-called strings, the relation Φkα α extends to their by the standard way ([13], pp. 151–156). We will apply the construction of an inductive limit in a special case when the index set I be the set of positive integers with the natural order. In this case, the inductive family is the sequence ℜ1, ℜ2, . . ., and morphisms fi,k will be defined as compositions of morphisms fi = fi,i+1 (i, k ∈ N). Moreover, sequences of the type u = akak+1 . . . (k ∈ N) are strings, if the following conditions hold: (i) ai ∈ Ai (i > k); (ii) fi(ai) = ai+1 (i > k); (iii) there is no an element ak−1 ∈ Ak−1, for which fk−1(ak−1) = ak. Let (ui = a (i) li a (i) li+1 . . . , 1 6 i 6 kα), and let Φkα α be a relation from the signature of relational structures ℜi, i ∈ N. From the definition follows that the tuple of strings (u1, u2, . . . , ukα ) is in the relation Φkα α if and only if for l > max{l1, l2, . . . , lkα } the tuples (a (1) l , a (2) l , . . . , a (kα) l ) is in this relation. Let the subset ℜ(k) of the support of the structure ℜ(Σ), be the set of all strings that was began with the elements from Al, l 6 k. Then subset ℜ(k) determines a substructure of the structure ℜ(Σ), and the equality ℜ(Σ) = ∞⋃ k=1 ℜ(k) takes place. 4. Construction of “lattice completion” Let ℜ = 〈A, {Φkα α }α∈I〉 be relational structures with the signature (kα)α∈I . Suppose that for every natural number n the structure ℜ has the only one its own substructure ℜ(n), moreover for the family of sub- structures 〈ℜ(n), n ∈ N〉 following conditions hold: (A) if n1 6= n2, then ℜ(n1) 6= ℜ(n2); 192 Representation of Steinitz’s lattice (B) the inclusion ℜ(n1) ⊆ ℜ(n2) is satisfied if and only if, when n1|n2; It follows from properties (A), (B) that family of substructures ℜ(n), n ∈ N, forms a lattice under the inclusion. A correspondence n ↔ ℜ(n), n ∈ N, is an isomorphism between this lattice and lattice (N, |) of natural numbers. Using the family ℜ(n), n ∈ N of substructures we define a new family from ℜ as follows. Let χ = 〈n1, n2, . . .〉 ∈ DS be an arbitrary divisible sequence of natural number. Construct now by sequence χ a growing chain of substructures of structure ℜ of the form ℜ(n1) ⊆ ℜ(n2) ⊆ · · · . If sequence χ is bounded, then ℜ(χ) = ∞⋃ i=1 ℜ(ni) coincides with one of substructures ℜ(nk), k ∈ N. Therefore it suffices to consider only divisible unbounded sequences. Suppose that for any χ as unbounded divisible sequences the family of substructures ℜ(χ) satisfies the following conditions: (C) union ℜ(χ) is its own substructure of structure ℜ; (D) ℜ(χ) does not coincide with any substructures ℜ(n), n ∈ N. Definition 8. The family of substructures ℜ(χ), χ ∈ DS, of structure ℜ will be called the lattice completion of lattice ℜ(n), n ∈ N. Since DS contains arbitrary bounded divisible sequences, the family ℜ(χ), χ ∈ DS, contains sublattice ℜ(n), n ∈ N. Theorem 1. Suppose, that the family of substructures ℜ(n), n ∈ N, of relational structure ℜ satisfies properties (A)–(D). Then (i) substructures ℜ(χ1) and ℜ(χ2), χ1, χ2 ∈ DS, coincide if and only if char χ1 = char χ2; (ii) the family of substructures ℜ(χ), χ ∈ DS, forms a lattice under the inclusion, which is isomorphic to Steinitz’s lattice. Proof. (i) Suppose ℜ(χ1) and ℜ(χ2) are determined by divisible sequences of natural numbers χ1 = 〈n (1) i 〉i∈N, χ2 = 〈n (2) i 〉i∈N. If ℜ(χ1) = ℜ(χ2), then ℜ(χ1) ⊆ ℜ(χ2) and ℜ(χ2) ⊆ ℜ(χ1). The inclusion ℜ(χ1) ⊆ ℜ(χ2) holds if and only if for any natural i there exists a number j, such that ℜ(n (1) i ) ⊆ ℜ(n (2) j ). According to the property (B) it means that O. Bezushchak, B. Oliynyk, V. Sushchansky 193 n (1) i |n (2) j , that is the sequence χ1 is a divisor of the sequence χ2. On the other hand, the inclusion ℜ(χ2) ⊆ ℜ(χ1) means that for any j ∈ N substructures ℜ(n (2) j ) is contained into some substructures ℜ(n (1) i ), namely for an arbitrary j ∈ N there exists such i ∈ N, that n (2) j |n (1) i . This means that the relation χ2|χ1 holds. Thus, sequences χ2 and χ1 are exactly divisible. So, char χ1 = char χ2 by lemma 5. Now suppose char χ1 = char χ2. Then sequences χ1 and χ2 are exactly divisible, that is χ1|χ2 and χ2|χ1. As properties (A) and (B) hold for the family ℜ(n), n ∈ N, using the considerations similar to above we obtain that ℜ(χ1) ⊆ ℜ(χ2) and ℜ(χ2) ⊆ ℜ(χ1). In other words, these substructures coincide. (ii) Let DS(0) be a set of fixed representatives of each class of exactly divisible sequences from DS. Then {ℜ(χ) | χ ∈ DS} = {ℜ(χ) | χ ∈ DS(0)}. We shall show, that the family of substructures on the right side of this equality satisfies the condition (ii) of this theorem. Then the subset of DS(0) is determined by classes of exactly divisible sequences on DS. Hence, the mapping λ : SN → DS(0), such that λ(u) = χ iff char χ = u, is a bijective by using lemma 5. So, properties (C), (D) of the family ℜ(χ), χ ∈ DS, imply that a mapping λ̄ : DS(0) → {ℜ(χ) | χ ∈ DS} defined by λ̄(χ) = ⋃ n∈χ ℜ(n) = ℜ(χ), χ ∈ DS(0), is a bijective too. Thus the mapping µ = λ · λ̄ : SN → {ℜ(χ) | χ ∈ DS(0)} will also be a bijective. So that the image of arbitrary supernatural number will be own substructure of the structure ℜ. It is also clear that µ(1) = ℜ(1), µ(I) = ∪n∈χℜ(n), where χ be such divisible sequence, that char χ = I. It remains to show that the mapping µ is consistent with the lattice operations ∨ and ∧. To this end, we are going to check that µ is consistent with the divisibility | on the set SN and the inclusion of ⊆ on substructures {ℜ(χ) | χ ∈ DS(0)}. Indeed, let the condition u1|u2 hold for supernatural numbers u1, u2. Then λ(u1)|λ(u2). It follows from (i) that λ̄(λ(u1)) ⊆ λ̄(λ(u2)). So (λ · λ̄)(u1) ⊆ (λ · λ̄)(u2). The theorem is proved. Theorem 1 can be reformulated in terms of inductive limits as follows. Let ℜ(n), n ∈ N be a family of structures. This family satisfies the condition (A). There is an embedding ϕn1,n2 of the structure ℜ(n1) into 194 Representation of Steinitz’s lattice the structure ℜ(n2) for arbitrary n1, n2 ∈ N, such that n1|n2. Assume that for monomorphisms ϕn1,n2 following standard requirements: (a) ϕn,n = Id for any n ∈ N; (b) ϕn1,n3 = ϕn1,n2 · ϕn2,n3 for arbitrary n1, n2, n3 ∈ N such, that n1|n2 and n2|n3; hold. Then we have the direct spectrum 〈ℜ(n), ϕn,m〉n,m∈N of relational structures, and let ℜ be a limit of this spectrum. For arbitrary sequence χ = 〈ni〉i∈N it can be considered a direct spectrum 〈ℜ(ni), ϕni,ni+1〉i∈N and its direct limit ℜ(χ). This structure can considered as a substructure of structure ℜ in an obvious way. Theorem 2. 1) Direct limits, defined by divisible sequences χ1 and χ2, coincide as substructures of ℜ if and only if char χ1 = char χ2. 2) All boundary structures considered as substructures of ℜ and defined by divisible sequences form a lattice with respect to inclusion. This lattice is isomorphic to the Steinitz’s lattice. We shall show examples, described how such construction works in two different situations: for an universal algebras and metric spaces. 5. Groups and semigroups of periodically defined trans- formations of natural numbers Let PTn be a semigroup of all partial defined transformations of set the {1, 2, . . . , n}, let PT (N) be a semigroup of all partial defined transformations of the set of natural numbers N. Definition 9. A transformation π̂ ∈ PT (N) is called periodically defined expansion of the transformation π ∈ PTn on the set N, if its act on natural numbers is defined by the following table ( 1 2 . . . n n + 1 n + 2 . . . 2n . . . . . . 1π 2π . . . nπ n + 1π n + 2π . . . n + nπ . . . . . . ) . A transformation α ∈ PT (N) is called periodically defined with the period of definition n, if there is a transformation π ∈ PTn such, that α = π̂. Every periodically defined transformation has infinitely many of dif- ferent periods that are multiples of the minimal period. Let Tn, ISn, Sn be, respectively, the complete semigroup of everywhere defined transfor- mations of the set {1, 2, . . . , n}, the complete inverse semigroup of partial O. Bezushchak, B. Oliynyk, V. Sushchansky 195 permutations and the complete symmetric group over this set. Thus let T (N), IS(N), S(N) denote these semigroups over the set of natural numbers. Lemma 6. For an arbitrary n ∈ N the mapping ϕn : π → π̂, π ∈ PTn, is a homomorphic emmbeding of the semigroup PTn into PT (N) and its restrictions on Tn, ISn, Sn, respectively, is an emmbeding into T (N), IS(N) and S(N). Proof. The injection of the mapping ϕn follows directly from its definition and its consistency with the operation of the multiplication of permuta- tions is obvious. Moreover, for any partial permutation π ∈ ISn and its inverse permutation π∗ we have ϕn(π∗) = π̂∗ = π̂∗. In particular, for any everywhere defined permutation π ∈ Sn we shall get ϕn(π−1) = π̂−1. So, ϕn will be a monomorphism ISn into IS(N) and Sn into S(N). We will denote by the symbol Gn one of semigroups PTn, Tn, ISn, Sn, n ∈ N, and will denote by the symbol G one of the corresponding semi- groups PT (N), T (N), IS(N) or S(N). Thus, all formulated statements will take place simultaneously for all four series of semigroups. Let Ĝn be the image ϕn(Gn) of the semigroup Gn. Then Ĝn be a subsemigroup in G, which is isomorphic to Gn. A family of semigroups Ĝn, n ∈ N, in G is partially ordered by the inclusion. The main property of this partially ordered set will be given in following lemma. Lemma 7. A partially ordered set ({Ĝn, n ∈ N}, ⊆) is a lattice, which is isomorphic to the lattice of positive integers with a relation of the divisibility. Proof. We shall verify that the correspondence n ↔ Ĝn, n ∈ N, will be an isomorphism of partially ordered sets (N, |) and ({Ĝn, n ∈ N}, ⊆). This correspondence is bijective, hence it is enough to show that for any m, n ∈ N the condition m|n holds if and only if Ĝm ⊆ Ĝn. Let m|n and n = km. So that the permutation π̂(k) given by the equality π̂(k) = ( 1 2 . . . m . . . (k − 1)m + 1 . . . km 1π 2π . . . mπ . . . (k − 1)m + 1π . . . (k − 1)m + mπ ) contains into Gn, and the equality holds: ϕm(π) = ϕ(π̂(k)) = π̂. So, for any transformation α ∈ Ĝ such, that α ∈ Ĝm, we obtain α ∈ Ĝn. Hence, Ĝm ⊆ Ĝn. 196 Representation of Steinitz’s lattice On the other hand, let Ĝm ⊆ Ĝn. In semigroup Ĝm there are permu- tations with the minimal period of a definition m. Any such permutation has the period of a definition n because it belongs to Ĝn. It follows that m|n. Since the conditions (A)–(B) from section 4 for the family {Gn, n ∈ N} hold, then it is possible to apply the construction of lattice completion. Namely, we introduce G(χ) by setting G(χ) = ∞⋃ i=1 Ĝni for an arbitrary sequence χ = 〈n1, n2, . . .〉 ∈ DS. Hence, for so determined subsemigroups of the semigroup G the conditions (C) and (D) from section 4 hold. This allows us to get the following result. Theorem 3. The family of subsemigroups G(χ), χ ∈ DS, forms a lattice regarding to an inclusion in the semigroup G, which is a Steinitz’s lattice. Different elements of this lattice are pairwise non-isomorphic semigroups. Proof. The first part of the statement is a direct consequence of Theorem 1. So, we have to show its second part. Let first G = S(N) be the group of permutations on the set N, and let G(u) = S(u) be the corresponding group of periodically defined permutations (u ∈ SN). Due to [6] groups G(u1) and G(u2) for u1, u2 ∈ SN, u1 6= u2, are non-isomorphic, because one of them contains a permutation a centralizator that can be non isomorphic to a centralizator of any permutation from another group. Suppose now that G(u) is a subsemigroup of such periodically defined permutations from G, that minimum periods are divisors of a supernatural number u. Then the group G(u) of inverse elements coincides with S(u). Hence, for u1 6= u2 semigroups G(u1) and G(u2) are non-isomorphic because of groups of their inverse elements are non-isomorphic too. The semigroup G(u), u ∈ SN, can be defined as a limit of the direct spectrum of semigroups with so-called diagonal emmbeding. According to [14] the emmbeding of the transitive transformations group (G, X) into the transformations group (H, Y ) is called a diagonal emmbeding if orbits of G onto the set Y either are trivial (consist of one point), or have the same orbit cardinality |X|. Note, that the act G on such orbit is isomorphic to the permutation group (G, X). A diagonal emmbeding of a group (G, X) into a permutation group (H, Y ) is called a strictly diagonal emmbeding, O. Bezushchak, B. Oliynyk, V. Sushchansky 197 if there are no trivial orbits of group G onto Y . In this case, with some k ∈ N we have |Y | = k|X|. Note that strictly diagonal emmbeding can be defined for arbitrary, not necessarily transitive permutations groups. So, its can be defined for transformations semigroups too. Definition 10. An emmbeding of a transformation semigroup (V, X) into a transformation semigroup (W, Y ) will be called (strictly) diagonal emmbeding, if there is a partition of Y onto subsets of the capacity |X|, which are invariant under the action of the image of V . Moreover, the action of the image of V onto each of these subsets is isomorphic as a semigroup of transformations to semigroup (V, X). As before, let Gn ∈ {PTn, Tn, ISn, Sn}. Lemma 8. Let the mapping δk : Gn → Gnk is defined by equality δk(π) = π̂(k), for any permutation π ∈ Gn. Then δk is an isomorphic emmbeding of the semigroup Gn into the semigroup Gnk. A proof is done by a direct check as at lemma 6. For an arbitrary divisible sequence χ = 〈n1, n2, . . .〉 ∈ SN, ni+1/ni = ki (i = 1, 2, . . .) we define a direct spectrum of semigroups Gni −→δk1 Gn2 −→δk2 Gn3 −→ · · · . So, we also have to consider the limit semigroup of the spectrum G[χ] = lim →i (Gni , δki ). Theorem 4. For any supernatural number χ ∈ SN semigroups G[χ] and G(χ) are isomorphic as semigroups of transformations. Proof. An isomorphism of semigroups is constructed in the standard way, and the set of an action of G[χ] is naturally identified with N. Finally, we note that our semigroups act equally on the set N. 6. Steinitz’s lattice of subspaces in Besicovitch’s space A normalized Hamming metric on the set Hn of (0, 1)-sequences of a length n is called a metric dH , defined by the following equality dH(x, y) = 1 n n∑ i=1 |xi − yi|, x = (x1, . . . , xn), y = (y1, . . . , yn) ∈ Hn. (8) 198 Representation of Steinitz’s lattice So, let now {0, 1}N be a set of infinite (0, 1)-sequences. We will represent the distance function d̂B by the rule d̂B(x, y) = lim n→∞ sup dH((x1, . . . , xn), (y1, . . . , yn)), (9) x = (x1, . . . , xn), y = (y1, . . . , yn) ∈ {0, 1}n. Since (8) is a metric, it is easy to verify that the function d̂B, defined by (9), is a pseudometric, i.e., it differs from a metric so that there are infinite sequences with the distance between them equals to 0. The binary relation x ∼B y ⇔ d̂B(x, y) = 0 is an equivalence on {0, 1}N. Define XB = {0, 1}N/∼B . The function d̂B is consistent with the equivalence ∼B. Hence, it deter- mines a function dB on XB as follows: dB([x], [y]) = d̂B(x, y), (10) where [x], [y] are equivalence classes of ∼B, and x ∈ [x], y ∈ [y] are arbitrary representatives of these classes. Defined by the equality (10) the function dB is a metric. So, a metric space (XB, dB) is called Besicovitch’s space (see [15]). For any natural number n normalized Hamming space Hn is isometric embedded into Besicovitch’s space XB. Lemma 9. The mapping hn : Hn → XB, defined by setting hn((x1, . . . , xn)) = [(x1, . . . , xn, x1, . . . , xn, . . .)], (x1, . . . , xn) ∈ Hn, is an isometric emmbeding. Proof. By the definition the distance (10) between classes of the equiva- lence ∼B, which defined by periodic sequences x = x1, . . . , xn, x1, . . . , xn, . . . and y = y1, . . . , yn, y1, . . . , yn, . . . , is equal to 1 n dH((x1, . . . , xn), (y1, . . . , yn)). O. Bezushchak, B. Oliynyk, V. Sushchansky 199 Denote by H̃n the image of Hamming space Hn for the mapping hn. Lemma 10. The inclusion H̃n ⊆ H̃m (n, m ∈ N) holds if and only if then n|m. Proof. It is obvious. So that, Besicovitch’s space XB contains the family of subspaces H̃n, n ∈ N. It is easy to see, that for this family conditions (A) − (D) of the lattice completion hold. It follows from Theorem 1, that the space XB contains a family of subspaces H̃u, u ∈ SN, indexing by supernatural numbers. Note, that the subspace H̃u consists of all possible periodic (0, 1)-sequences, such that lengthes of their minimum periods are divisors of a supernatural number u. By properties of the general construction we obtain the following Theorem 5. A family of subspaces H̃u, u ∈ SN, of the space XB forms a lattice over the inclusion which is isomorphic to the lattice of supernatural numbers. If u1 6= u2, then subspaces H̃u1 and H̃u2 are not isometric. Proof. The first part of the assertion follows from Theorem 1. The proof of the third part given in the article [16]. By the Theorem 1 each of spaces H̃u, u ∈ SN \ N, is isometric to inductive limit of the sequence of finite Hamming spaces Hm1 , Hm2 , . . ., where χ = 〈m1, m2, . . .〉 is such divisible sequence that char χ = u. Monomorphisms are the diagonal emmbedings fi : Hmi → Hmi+1 , defined by following equalities fi((x1, . . . , xmi )) = (x1, . . . , xmi , . . . , x1, . . . , xmi︸ ︷︷ ︸ kimi ), (x1, . . . , xmi ) ∈ Hmi , (11) where ki = mi+1/mi, i = 1, 2, . . .. Note, that a construction of limit cube, that is built as inductive limit of the sequence of Hamming spaces H2i of the dimension 2i with emmbedings of following doubling coordinates: δi((x1, . . . , x2i)) = (x1, x1, x2, x2, . . . , x2i , x2i), i = 1, 2, . . . , is considered in [17]. So that, it is easy to understand that the limit space of such direct spectrum is isometric to the space H̃2∞ with emmbedings of the type (11). Note, that in [18] was considered continuum family of subspaces of the Besicovitch space on some alphabet B, naturally parametrized by supernatural numbers. Every subspace is defined as 200 Representation of Steinitz’s lattice a diagonal limit of finite Hamming spaces on the alphabet B. So, our construction is more general. Other generalizations of this construction are proposed in [19], and a generalization of construction of lattice completion for linear groups is considered in the article [14]. References [1] E. Steinitz, Algebraische Theorie der Korper, J. reine angew. Math., Vol. 137, 1910, pp. 167-309 / Reprinted: Algebraische Theorie der Korper, New York: Chelsea Publ, 1950. [2] Joel V. Brawley, George E. Schnibben, Infinite algebraic extensions of finite fields, Cotemporary Math., Vol. 95, Amer. Math. Soc., Providence, Rhode Island, 1989. [3] J. G. Glimm, On a certain class of operator algebras, Trans. Amer. Math. Soc., Vol. 95, N. 2, 1960, pp. 318–340. [4] S.C. Power, Limit algebras: an introduction to subalgebras of C ∗–algebras, Harlow, Essex: Longman Scientific and Technical. New York: Wiley, 1993. [5] N.V. Kroshko, V.I. Sushchanskij, Homogeneous symmetric groups, Dopov. Akad. Nauk Ukr., N.12, 1993, pp. 9-13, (Ukrainian). [6] N. Kroshko, V. Sushchansky, Direct limits of symmetric and alternating groups with strictly diagonal embeddings Arch. Math. (Basel), Vol. 71, 1998, pp. 173–182. [7] E.E. Goryachko, The K0-functor and characters of the group of rational rearrange- ments, (English. Russian original) J. Math. Sci., N.158(6), 2009, pp. 838-844. [8] E.E. Goryachko, F.V. Petrov, Indecomposable characters of the group of rational rearrangements of the segment, (English. Russian original) J. Math. Sci., New York, Vol. 174, 2011, pp. 7–14. [9] A. Bier, V.I. Sushchanskyy, Dense subgroups in the group of interval exchange transformations, Algebra and Discrete Math., Vol. 17, N. 2, 2014, pp. 232–247. [10] V.I. Sushchanskij, The lattice of supernatural numbers as a subalgebra lattice in universal algebras, Dopov. Akad. Nauk Ukr., N.11, 1997, pp. 42–45, (Ukrainian). [11] V.I. Sushchansky, V. S. Sikora, Operations on permutation groups. The theory and application. Chernivtsi, Ruta, 2003, (Ukrainian). [12] Peter J. Cameron, Oligomorphic Permutation Groups, London Math. Soc. Lecture Notes, Vol. 152, Cambridge Univ. Press, 1990. [13] N. Bourbaki, Set theory, Moskva, Mir, 1965 (Russian). [14] A. E. Zalesskii, Group rings of inductive limits of alternating groups, Leningrad Mathematical Journal, Vol. 2, N. 6, 1991, pp. 1287–1303. [15] F. Blanchard, E. Formenti, P. Kurka, Cellular Automata in the Cantor, Besicovitch and Weyl Topological Spaces, Complex Systems, Vol. 11, N. 2, 1997, pp. 107–123. [16] B. V. Oliynyk, V. I. Sushchanskii The isometry groups of Hamming spaces of periodic sequences, Siberian Mathematical Journal, Vol. 54, N. 1, 2013, pp. 124–136. [17] P. J. Cameron, S. Tarzi, Limits of cubes, Topology and its Applications, Vol. 155, N. 14, 2008, pp. 1454–1461. O. Bezushchak, B. Oliynyk, V. Sushchansky 201 [18] B. Oliynyk, The diagonal limits of Hamming spaces, Algebra and Discrete Math.,Vol. 15, N. 2, 2013, pp. 229–236. [19] O. O. Bezushchak, V. I. Sushchansky, Groups of periodically defined linear transfor- mations of an infinite-dimensional vector space, Ukrainian Mathematical Journal, Vol. 67, N. 10, 2016, pp. 1457–1468. Contact information O. Bezushchak Department of Mechanics and Mathematics Kyiv National Taras Shevchenko University Volodymyrska, 64, Kyiv 01033, Ukraine E-Mail(s): bezusch@univ.kiev.ua B. Oliynyk Department of Mathematics, National Univer- sity of Kyiv-Mohyla Academy, Skovorody St. 2, Kyiv, 04655, Ukraine E-Mail(s): oliynyk@ukma.edu.ua V. Sushchansky Institute of Mathematics Silesian University of Technology ul. Kaszubska 23, 44-100 Gliwice, Poland E-Mail(s): Vitaliy.Sushchanskyy@polsl.pl Received by the editors: 11.05.2016.
id nasplib_isofts_kiev_ua-123456789-155237
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn 1726-3255
language English
last_indexed 2025-12-07T17:05:08Z
publishDate 2016
publisher Інститут прикладної математики і механіки НАН України
record_format dspace
spelling Bezushchak, O.
Oliynyk, B.
Sushchansky, V.
2019-06-16T14:26:24Z
2019-06-16T14:26:24Z
2016
Representation of Steinitz's lattice in lattices of substructures of relational structures / O. Bezushchak, B. Oliynyk, V. Sushchansky // Algebra and Discrete Mathematics. — 2016. — Vol. 21, № 2. — С. 184-201. — Бібліогр.: 19 назв. — англ.
1726-3255
2010 MSC:03G10, 08A02, 03G05.
https://nasplib.isofts.kiev.ua/handle/123456789/155237
General conditions under which certain relational structure contains a lattice of substructures isomorphic to Steinitz's lattice are formulated. Under some natural restrictions we consider relational structures with the lattice containing a sublattice isomorphic to the lattice of positive integers with respect to divisibility. We apply to this sublattice a construction that could be called ``lattice completion''. This construction can be used for different types of relational structures, in particular for universal algebras, graphs, metric spaces etc. Some examples are considered.
en
Інститут прикладної математики і механіки НАН України
Algebra and Discrete Mathematics
Representation of Steinitz's lattice in lattices of substructures of relational structures
Article
published earlier
spellingShingle Representation of Steinitz's lattice in lattices of substructures of relational structures
Bezushchak, O.
Oliynyk, B.
Sushchansky, V.
title Representation of Steinitz's lattice in lattices of substructures of relational structures
title_full Representation of Steinitz's lattice in lattices of substructures of relational structures
title_fullStr Representation of Steinitz's lattice in lattices of substructures of relational structures
title_full_unstemmed Representation of Steinitz's lattice in lattices of substructures of relational structures
title_short Representation of Steinitz's lattice in lattices of substructures of relational structures
title_sort representation of steinitz's lattice in lattices of substructures of relational structures
url https://nasplib.isofts.kiev.ua/handle/123456789/155237
work_keys_str_mv AT bezushchako representationofsteinitzslatticeinlatticesofsubstructuresofrelationalstructures
AT oliynykb representationofsteinitzslatticeinlatticesofsubstructuresofrelationalstructures
AT sushchanskyv representationofsteinitzslatticeinlatticesofsubstructuresofrelationalstructures