CC BY
14
Серия «Математика» 2016. Т. 17. С. 62—76
Онлайн-доступ к журналу: http://isu.ru/izvestia
ИЗВЕСТИЯ
Иркутского государственного университета
УДК 510.67:515.12
MSC 03C30, 03C15, 03C50, 54A05
Families of language uniform theories
and their generating sets *
S. V. Sudoplatov
Sobolev Institute of Mathematics, Novosibirsk State Technical University, Novosibirsk State University, Institute of Mathematics and Mathematical Modeling
Abstract. We introduce the notion of language uniform theory and study topological properties related to families of language uniform theory and their E-combinations. It is shown that the class of language uniform theories is broad enough. Sufficient conditions for the language similarity of language uniform theories are found. Properties of language domination and of infinite language domination are studied. A characterization for E-closure of a family of language uniform theories in terms of index sets is found. We consider the class of linearly ordered families of language uniform theories and apply that characterization for this special case. The properties of discrete and dense index sets are investigated: it is shown that a discrete index set produces a least generating set whereas a dense index set implies at least continuum many accumulation points and the closure without the least generating set. In particular, having a dense index set the theory of the E-combination does not have e-least models and it is not small. Using the dichotomy for discrete and dense index sets we solve the problem of the existence of least generating set with respect to E-combinations and characterize that existence in terms of orders.
Values for e-spectra of families of language uniform theories are obtained. It is shown that any e-spectrum can be realized by E-combination of language uniform theories. Low estimations for e-spectra relative to cardinalities of language are found.
It is shown that families of language uniform theories produce an arbitrary given Cantor-Bendixson rank and given degree with respect to this rank.
Keywords: E-combination, P-combination, closure operator, generating set, language uniform theory.
We continue to study structural properties of ^-combinations of structures and their theories [5; 6]. The notion of language uniform theory is introduced. For the class of linearly ordered language uniform theories we
* The research is partially supported by the Grants Council (under RF President) for State Aid of Leading Scientific Schools (Grant NSh-6848.2016.1) and by Committee of Science in Education and Science Ministry of the Republic of Kazakhstan (Grant No. 0830/GF4).
solve the problem of the existence of least generating set with respect to ^-combinations and characterize that existence in terms of orders. Values for e-spectra of families of language uniform theories are obtained. It is shown that families of language uniform theories produce an arbitrary given Cantor-Bendixson rank and given degree.
1. Preliminaries
Throughout the paper we use the following terminology in [5; 6].
Definition 1. [5]. Let P = (Pi)i^i, be a family of nonempty unary predicates, (Ai)iei be a family of structures such that Pi is the universe of Ai, i £ I, and the symbols Pi are disjoint with languages for the structures Aj, j £ I. The structure Ap ^ U Ai expanded by the predicates Pi is
iei
the P-union of the structures Aj,, and the operator mapping (Ai)iei to
Ap is the P-operator. The structure Ap is called the P-combination of the
structures Ai and denoted by CombP (Ai)iei if Ai = (Ap \ Ai) \ E(Ai),
i £ I. Structures A', which are elementary equivalent to CombP(Ai)iei,
will be also considered as P-combinations.
Clearly, all structures A1 = CombP(Ai)iei are represented as unions of
their restrictions Ai = (A' \ Pi) \ S(Ai) if and only if the set p^(x) =
{-iPi(x) | i £ 1} is inconsistent. If A! / Combp(Ai)iei, we write A! =
Combp (Ai) ie/u{oo}) where A'^ = AI \ H Pi, maybe applying Morleyza-
iei
tion. Moreover, we write Combp(Ai)ieiu{^} for Combp(Ai)iei with the empty structure A^>.
Note that if all predicates Pi are disjoint, a structure Ap is a P-combi-nation and a disjoint union of structures Aj,. In this case the P-combination Ap is called disjoint. Clearly, for any disjoint P-combination Ap, Th(Ap) = Th(A'p), where A'p is obtained from Ap replacing Ai by pairwise disjoint Ai = Ai, i £ I. Thus, in this case, similar to structures the P-operator works for the theories Ti = Th(Ai) producing the theory Tp = Th(Ap), being P-combination of T, which is denoted by Combp(Ti)iei.
For an equivalence relation E replacing disjoint predicates Pi by E-classes we get the structure AE being the E-union of the structures Ai. In this case the operator mapping (Ai)i£i to AE is the E-operator. The structure Ae is also called the E-combination of the structures Ai and denoted by CombE(Ai)iei; here Ai = (Ae \ Ai) \ E(Ai), i £ I. Similar above, structures A', which are elementary equivalent to AE, are denoted by CombE(Aj)j, where Aj are restrictions of A to its E-classes. The E-operator works for the theories Ti = Th(Ai) producing the theory TE = Th(AE), being E-combination of T, which is denoted by CombE(Ti)iei or by CombE(T), where T = {Ti | i £ I}.
Clearly, A = Ap realizing p^(x) is not elementary embeddable into Ap and can not be represented as a disjoint P-combination of Ai = Ai, i £ I .At the same time, there are ^-combinations such that all A = Ae can be represented as E-combinations of some Aj = Ai. We call this representability of A to be the E-representability.
If there is A = AE which is not E-representable, we have the E'-representability replacing E by E' such that E' is obtained from E adding equivalence classes with models for all theories T, where T is a theory of a restriction B of a structure A = AE to some E-class and B is not elementary equivalent to the structures Ai. The resulting structure AE/ (with the E'-representability) is a e-completion, or a e-saturation, of AE. The structure AE itself is called e-complete, or e-saturated, or e-universal, or e-largest.
For a structure AE the number of new structures with respect to the structures Ai, i. e., of the structures B which are pairwise elementary non-equivalent and elementary non-equivalent to the structures Ai, is called the e-spectrum of AE and denoted by e-Sp(AE). The value sup{e-Sp(A')) | A = Ae} is called the e-spectrum of the theory Th(AE) and denoted by
e-Sp(Th(AE)).
If Ae does not have E-classes Ai, which can be removed, with all E-classes Aj = Ai, preserving the theory Th(AE), then AE is called e-prime, or e-minimal.
For a structure A = AE we denote by TH(A') the set of all theories Th(Ai) of E-classes Ai in A'.
By the definition, an e-minimal structure A consists of E-classes with a minimal set TH(A'). If TH(A') is the least for models of Th(A') then A is called e-least.
Definition 2. [6]. Let T be the class of all complete elementary theories of relational languages. For a set T С T we denote by CIE(T) the set of all theories Th(A), where A is a structure of some E-class in A = AE, Ae = CombE(A)i,ei, Th(Ai) £ T. As usual, if T = ClE(T) then T is said to be E-closed.
The operator C1E of I?-closure can be naturally extended to the classes T С T as follows: C1E(T) is the union of all ClE(7o) for subsets % С Т.
For a set T С T of theories in a language E and for a sentence (p with E(^) С E we denote by % the set {T eTIp £ T}.
Proposition 1. [6]. IfT С T is an infinite set and T e T \ T then T £ CIe(T) (i.e., T is an accumulation point for T with respect to E-closure ClE) if and only if for any formula p £ T the set Tv is infinite.
Definition 3. [6]. Let T0 be a closed set in a topological space (T, OE(T)), where Oe(T) = {t\ CIe(T') | T' С T}. A subset TO С To is said to be
generating if T0 = ClE(70'). The generating set T' (for T0) is minimal if TO does not contain proper generating subsets. A minimal generating set TO is least if Ti is contained in each generating set for T0.
Theorem 1. [6]. If TO is a generating set for a E-closed set T0 then the following conditions are equivalent:
(1) TO is the least generating set for T0;
(2) TO is a minimal generating set for T0;
(3) any theory in TO is isolated by some set (T0)p, i.e., for any T £T0 there is p £ T such that (T0)^ = {T};
(4) any theory in TO is isolated by some set (T0)v, i.e., for any T £T0 there is p £ T such that (%)p = {T}.
Proposition 2. [6]. For any closed nonempty set T0 in a topological space (T, OE(T)) and for any TO C T0, the following conditions are equivalent:
(1) TO is the least generating set for T0;
(2) any/some structure AE = CombE(Ai)iei, where {Th(Ai) | i £ I} = TO, is an e-least model of the theory Th(AE) and E-classes of each/some e-largest model of Th(AE) form models of all theories in T0;
(3) any/some structure AE = CombE(Ai)iei, where {Th(Ai) | i £ I} = TO, Ai ^ Aj for i = j, is an e-least model of the theory Th(AE), where E-classes of AE form models of the least set of theories and E-classes of each/some e-largest model of Th(AE) form models of all theories in T0.
Theorem 1 and Proposition 2 answer Question 1 in [6] characterizing the existence of the least generating set. The following question also has been formulated in [6].
Question. Is there exists a theory Th(AE) without the least generating set?
Below we will consider a class of special theories with respect to their languages and answer the question characterizing the existence of the least generating set in these special cases.
2. Language uniform theories and related E-closures
Definition 4. A theory T in a predicate language £ is called language uniform, or a LU-theory if for each arity n any substitution on the set of non-empty n-ary predicates preserves T. The LU-theory T is called IILU-theory if it has non-empty predicates and as soon as there is a non-empty n-ary predicate then there are infinitely many non-empty n-ary predicates and there are infinitely many empty n-ary predicates.
Below we point out some basic examples of LU-theories:
• Any theory T0 of infinitely many independent unary predicates Rk is a LU-theory; expanding T0 by infinitely many empty predicates Ri we get a IILU-theory Ti.
• Replacing independent predicates Rk for T0 and T1 by disjoint unary predicates R'k with a cardinality Л £ (ш + 1) \ {0} such that each R'k has Л elements; the obtained theories are denoted by T? and T? respectively; here, T? and T? are LU-theories, and, moreover, T? is a IILU-theory; we denote TJ and T1 by Tf and Tf; in this case nonempty predicates Rk' are singletons symbolizing constants which are replaced by the predicate languages.
• Any theory T of equal nonempty unary predicates Rk is a LU-theory;
• Similarly, LU-theories and IILU-theories can be constructed using nary predicate symbols of arbitrary arity n.
• The notion of language uniform theory can be extended for an arbitrary language taking graphs for language functions; for instance, theories of free algebras can be considered as LU-theories.
• Acyclic graphs with colored edges (arcs), for which all vertices have same degree with respect to each color, has LU-theories. If there are infinitely many colors and infinitely many empty binary relations then the colored graph has a IILU-theory.
• Generic arc-colored graphs without colors for vertices [1; 4], free poly-gonometries of free groups [2], and cube graphs with coordinated colorings of edges [2; 3] have LU-theories.
The simplest example of a theory, which is not language uniform, can be constructed taking two nonempty unary predicates R1 and R2, where R1 С R2. More generally, if a theory T, with nonempty predicates Rj,, i £ I, of a fixed arity, is language uniform then cardinalities of Rfl (x) Л... Л Rj (x) do not depend on pairwise distinct i1,...,ij.
Remark 1. Any countable theory T of a predicate language E can be
transformed to a LU-theory T'. Indeed, since without loss of generality
(k )
E is countable consisting of predicate symbols Rn , n £ ш, then we can step-by-step replace predicates Rn by predicates R'n in the following way. We put Rj ^ R0. If predicates Rj,...,R'n of arities r0 < ... < rn, respectively, are already defined, we take for R'n+1 a predicate of an arity rn+1 > max{rn,kn+1}, which is obtained from R'n+1 adding rn+1 — kn+1 fictitious variables corresponding to the formula
R (x1, . . . ,xkn+! ) Л (xkn+2 ~ Xkn+2) Л (xrn+l ~ XTn+1).
If the resulted LU-theory T' has non-empty predicates, it can be transformed to a countable IILU-theory T'' copying these non-empty predicated
with same domains countably many times and adding countably many empty predicates for each arity rn.
Clearly, the process of the transformation of T to T' do not hold for uncountable languages, whereas any LU-theory can be transformed to an IILU-theory as above.
Definition 5. Recall that theories T0 and Ti of languages £0 and £ respectively are said to be similar if for any models Mi |= Ti, i = 0,1, there are formulas of Ti, defining in Mi predicates, functions and constants of language £1_i such that the corresponding structure of £1_i is a model of Ti_i.
Theories T0 and T1 of languages £0 and £1 respectively are said to be language similar if T0 can be obtained from T1 by some bijective replacement of language symbols in £1 by language symbols in £0 (and vice versa).
Clearly, any language similar theories are similar, but not vice versa. Note also that, by the definition, any LU-theory T is language similar to any theory Ta which is obtained from T replacing predicate symbols R by a(R), where a is a substitution on the set of predicate symbols in £(T) corresponding to nonempty predicates for T as well as a substitution on the set of predicate symbols in £(T) corresponding to empty predicates for T. Thus we have
Proposition 3. Let T1 and T2 be LU-theories of same language such that T2 is obtained from T1 by a bijection f1 (respectively f2) mapping (non) empty predicates for T1 to (non) empty predicates for T2. Then T1 and T2 are language similar.
Corollary 1. Let T1 and T2 be countable IILU-theories of same language such that the restriction T1' ofT1 to non-empty predicates is language similar to the restriction T2' of T2 to non-empty predicates. Then T1 and T2 are language similar.
Proof. By the hypothesis, there is a bijection f2 for non-empty predicates of T1 and T2. Since T1 and T2 be countable IILU-theories then T1 and T2 have countably many empty predicates of each arity with non-empty predicates, there is a bijection f1 for empty predicates of T1 and T2. Now Corollary is implied by Proposition 3. □
Definition 6. For a theory T in a predicate language £, we denote by Supps(T) the support of £ for T, i. e., the set of all arities n such that some n-ary predicate R for T is not empty.
Clearly, if T1 and T2 are language similar theories, in predicate languages £1 and £2 respectively, then SuppSl (T1) = SuppSa (T2).
Definition 7. Let T1 and T2 be language similar theories of same language £. We say that T2 language dominates T1 and write T1 CL T2 if for any symbol R £ £, if T1 h IIXR(X) then T2 h IXR(X), i. e., all predicates, which are non-empty for T1, are nonempty for T2. If T1 CL T2 and T2 CL T1, we say that T1 and T2 are language domination-equivalent and write T1 T2.
Proposition 4. The relation CL is a partial order on any set of LU-theories.
Proof. Since CL is always reflexive and transitive, it suffices to note that if Ti CL T2 and T2 CL Ti then Ti = T2. It follows as language similar LU-theories coincide having the same set of nonempty predicates. □
Definition 8. We say that T2 infinitely language dominates T1 and write T1 T2 if T1 CL T2 and for some n, there are infinitely many new nonempty predicates for T2 with respect to T1
Since there are infinitely many elements between any distinct comparable elements in a dense order, we have
Proposition 5. If a class of theories T has a dense order CL then T1 T2 for any distinct T1T2 £T with T1 CL T2.
Clearly, if T1 CL T2 then Supps(T1) C Supps(T2) but not vice versa. In particular, there are theories T1 and T2 with T1 T2 and Supps(T1) = Supps(T2).
Let TO be a LU-theory with infinitely many nonempty predicate of some arity n, and I0 be the set of indexes for the symbols of these predicates.
Now for each infinite I C I0 with II| = |I0|, we denote by Ti the theory which is obtained from the complete subtheory of T0 in the language {Rk | k £ I} united with symbols of all arities m = n and expanded by empty predicates Ri for l £ I0 \ I, where \ 11 is equal to the cardinality of the set empty predicates for T0, of the arity n.
By the definition, each Ti is language similar to T0: it suffices to take a bijection f between languages of Ti and T0 such that (non)empty predicates of Ti in the arity n correspond to (non)empty predicates of T0 in the arity n, and f is identical for predicate symbols of the arities m = n. In particular, Let T be an infinite family of theories Ti, and Tj be a theory of the form above (with infinite J C I0 such that J| = |I0|). The following proposition modifies Proposition 1 for the E-closure ClE(T).
Proposition 6. If TJ / T then TJ £ ClE (T) if and only if for any finite set J0 C I0 there are infinitely many Ti with J n J0 = I n J0.
Proof. By the definition each theory Tj is defined by formulas describing Pk = 0 & k £ J. Each such a formula p asserts for a finite set J0 C I0 that
if k £ J0 then Rk = 0 & k £ J .It means that {k £ J0 | Pk = 0} = J n J0. On the other hand, by Proposition 1, Tj £ ClE(T) if and only if each such formula p belongs to infinitely many theories Tj in T, i.e., for infinitely many indexes I we have I n Jo = J n Jo. □
Now we take an infinite family F of infinite indexes I such that F is linearly ordered by C and if I\ C I2 then I2 \ I\ is infinite. The set {Tj | I £ F} is denoted by Tf-
For any infinite F' C F we denote by lim F' the union-set 1J F' and by lim F' intersection-set H-^"- ^ limF' (respectively limF') does not belong to F' then it is called the upper (lower) accumulation point (for F'). If J is an upper or lower accumulation point we simply say that J is an accumulation point.
Corollary 2. If TJ / Tf then TJ £ ClE(Tf) if and only if J is an (upper or lower) accumulation point for some infinite F' C F.
Proof If J = lim F' or J = lim F' then for any finite set Jo C Iq there are infinitely many Tj with J n J0 = I n J0. Indeed, if J = IJ F' then for any finite J0 C I0 there are infinitely many I £ F' such that In J0 contains exactly same elements as J nJ0 since otherwise we have J C U F'. Similarly the assertion holds for J = f] F'. By Proposition 6 we have Tj £ C\e(Tf)-
Now let J / limF' and J / lim F' for any infinite F' C F. In this case for each F' C F, either J contains new index j for a nonempty predicate with respect to F' for each F' C F with F' C J or F' contains new index j' for a nonempty predicate with respect to J for each F' C F with P|F' D J. In the first case, for J0 = {j} there are no I £ F' such that I n J0 = J n J0. In the second case, for J0 = {j'} there are no I £ F' such that I n J0 = J n J0. By Proposition 6 we get Tj £ C1E(7>). □
By Corollary 2 the action of the operator C1e for the families Tf is reduced to unions and intersections of index subsets of F.
Now we consider possibilities for the linearly ordered sets F = (F; C) and their closures T = (F; C) related to C\E.
The structure F is called discrete if F does not contain accumulation points.
By Corollary 2, if F is discrete then for any J £ F, TJ £ ClE(7F\{j}). Thus we get
Proposition 7. For any discrete F, TF is the least generating set for ClE (Tf ).
By Proposition 7, for any discrete F, TF can be reconstructed from ClE(TF) removing accumulation points, which always exist. For instance, if (F; C) is isomorphic to (w; <) or (w*; <) (respectively, isomorphic to (Z; <))
then C\e(Tf) has exactly one (two) new element(s) limF or lim F (both lim F and lim F).
Consider an opposite case: with dense F. Here, if F is countable then, similarly to (Q; <), taking cuts for F, i. е., partitions (F~,F+) of F with F~ < F+, we get the closure F with continuum many elements. Thus, the following proposition holds.
Proposition 8. For any dense F, |F| > 2Ш.
Clearly, there are dense F with dense and non-dense F. If F is dense then, since F = F, there are dense F\ with |Fi| = |Fi|. In particular, it is followed by Dedekind theorem on completeness of R. Answering the question in Section 1 we have
Proposition 9. If F is dense then C\e(Tf) does not contain the least generating set.
Proof. Assume on contrary that ClE(TF) contains the least generating set with a set F0 С F of indexes. By the minimality F0 does not contain both the least element and the greatest element. Thus taking an arbitrary J £ F0 we have that for the cut (F-,F+j), where FJJ = {J- £ F0 | J- С J}
and F0+j = {J+ g F0 | J+ d J}, J = Tim F~j and J = iim^oV Thus' F0 \ {J} is again a set of indexes for a generating set for ClE(TF). Having a contradiction we obtain the required assertion. □
Combining Proposition 2 and Proposition 9 we obtain
Corollary 3. If F is dense then Th(^E) does not have e-least models and, in particular, it is not small.
Remark 2. The condition of the density of F for Proposition 9 is essential. Indeed, we can construct step-by step a countable dense structure F without endpoints such that for each J £ F and for its cut (FJ,Fj+), where FJ = {J' £ F | J' С J} and Fj = {J+ e F \ J+ D J}, J D limFj and J С limFj^. In this case C\e(Tf) contains the least generating set {Tj | J £ F}.
In general case, if an element J of F has a successor J' or a predecessor Jthen J defines a connected component with respect to the operations ■' and j1. Indeed, taking closures of elements in F with respect to ■' and we get a partition of F defining an equivalence relation such that two elements J1 and J2 are equivalent if and only if J2 is obtained from J1 applying ■' or j1 several (maybe zero) times.
Now for any connected component C we have one of the following possibilities:
(i) C is a singleton consisting of an element J such that J / lim FJ and J / limFj^; in this case J is not an accumulation point for F\ {J} and Tj belongs to any generating set for C1e(7f);
(ii) C is a singleton consisting of an element J such that J = lim Fj~ or J = limF+, and lim Fj / lim Fj~; in this case J is an accumulation point for exactly one of F - and F+, J separates F j and F+, and T j can be removed from any generating set for C1e(Tf) preserving the generation of C1e(7f); thus Tj does not belong to minimal generating sets;
(iii) C is a singleton consisting of an element J such that J = lim FJ = limF+; in this case J is a (unique) accumulation point for both Fj and F+, moreover, again Tj can be removed from any generating set for C1e(Tf) preserving the generation of ClE(Tf), and T J does not belong to minimal generating sets;
(iv) \C\ > 1 (in this case, for any intermediate element J of C, Tj belongs to any generating set for C1e(7f)), limC d limFjmC and limC c lim FJ- ; in this case, for the endpoint(s) J * of C, if it (they) exists, Tj* belongs to any generating set for C1e(7f);
(v) \C\ > 1, and limC = limFjmC or limC = limF^j— ; in this case, for the endpoint J* = lim C of C, if it exists, Tj* does not belong to minimal generating sets of C1e(Tf), and for the endpoint J** = limC of C, if it exists, TJ** does not belong to minimal generating sets of ClE(Tf).
Summarizing (i)-(v) we obtain the following assertions.
Proposition 10. A partition of F by the connected components forms discrete intervals or, in particular, singletons of F, where only endpoints J of these intervals can be among elements J** such that TJ** does not belong to minimal generating sets of ClE(Tf).
Proposition 11. If (F~,F+) is a cut of F with limF" = limF+ (respectively limF" c limF+) then any generating set for C1e(7f) is represented as a (disjoint) union of generating set TF°- for ClE(TF-) and of generating set TF0+ for ClE(TF +), moreover, any (disjoint) union of a generating set for ClE(TF-) and of a generating set for ClE(TF +) is a generating set T° for ClE(Tf).
Proposition 11 implies
Corollary 4. If (FJ ,F+) is a cut of F then ClE(TF) has the least generating set if and only if ClE(TF-) and ClE(TF+) have the least generating sets.
Considering c-ordered connected components we have that discretely ordered intervals in F, consisting of discrete connected components and their limits lim and lim, are alternated with densely ordered intervals including their limits. If F contains an (infinite) dense interval, then by Proposition
9, CIe(Tf) does not have the least generating set. Conversely, if F does not contain dense intervals then ClE(Tf) contains the least generating set. Thus, answering Questions 1 and 2 [6] for ClE(TF) including the question in Section 1, we have
Theorem 2. For any linearly ordered set F, the following conditions are equivalent:
(1) CIe(Tf) has the least generating set;
(2) F does not have dense intervals.
Remark 3. Theorem 2 does not hold for some non-linearly ordered F. Indeed, taking countably many disjoint, incomparable with respect to nonempty predicates modulo their intersections, copies Fq, q £ Q, of linearly ordered sets isomorphic to (oj,<) and ordering limits Jq = limFg by the ordinary dense order on Q such that {Jq | q £ Q} is densely ordered, we obtain a dense interval {Jq | q £ Q} whereas the set U{Fq | q £ Q} forms the least generating set T0 of theories for ClE(T0).
The above operation of extensions of theories for {Jq | q £ Q} by theories for Fq as well as expansions of theories of the empty language to theories for {Jq | q £ Q} confirm that the (non)existence of a least/minimal generating set for ClE(T0) is not preserved under restrictions and expansions of theories.
Remark 4. Taking an arbitrary theory T with a non-empty predicate R of an arity n, we can modify Theorem 2 in the following way. Extending the language £(T) by infinitely many n-ary predicates interpreted exactly as R and by infinitely many empty n-ary predicates we get a class Tt,r of theories R-generated by T. The class TT,R satisfies the following: any linearly ordered F as above is isomorphic to some family F', under inclusion, sets of indexes of non-empty predicates for theories in TT,R such that strict inclusions J\ C J2 for elements in F' imply that cardinalities J2 \ J\ are infinite and do not depend on choice of J\ and J2. Theorem 2 holds for linearly ordered F' involving the given theory T.
3. On e-spectra for families of language uniform theories
Remark 5. Remind [5, Proposition 4.1, (7)] that if T = Th(^E) has an e-least model M then e-Sp(T) = e-Sp(M). Then, following [5, Proposition 4.1, (5)], e-Sp(T) = ITo\T01, where TO is the (least) generating set of theories for ^-classes of M, and T0 is the closed set of theories for ^-classes of an e-largest model of T. Note also that e-Sp(T) is infinite if T0 does not have the least generating set.
Remind that, as shown in [5, Propositions 4.3], for any cardinality A there is a theory T = Th( AE) of a language £ such that |£| = IA + 1| and e-Sp(T) = A. Modifying this proposition for the class of LU-theories we obtain
Proposition 12. (1) For any i < w there is an E-combination T = Th(AE) of IILU-theories in a language £ of the cardinality w such that T has an e-least model and e-Sp(T) =
(2) For any uncountable cardinality A there is an E-combination T = Th(AE) of IILU-theories in a language £ of the cardinality A such that T has an e-least model and e-Sp(T) = A.
Proof. In view of Propositions 2, 7 and Remark 5, it suffices to take an E-combination of IILU-theories of a language £ of the cardinality A and with a discrete linearly ordered set F having:
1) I < w accumulation points if A = w;
2) A accumulation points if A > w. We get the required F for (1) taking:
(i) finite F for i = 0;
(ii) i/2 discrete connected components, forming F, with the ordering type (Z; <) and having pairwise distinct accumulation points, if i > 0 is even natural;
(iii) (i — 1)/2 discrete connected components, forming F, with the ordering type (Z; <) and one connected components with the ordering type (w; <) such that all accumulation points are distinct, if i > 0 is odd natural;
(iv) w discrete connected components, forming F, with the ordering type (Z; <), if i = w.
The required F for (2) is formed by (uncountably many) A discrete connected components, forming F, with the ordering type (Z; <). □
Combining Propositions 2, 9, Theorem 2, and Remark 5 with F having dense intervals, we get
Proposition 13. For any infinite cardinality A there is an E-combination T = Th(AE) of IILU-theories in a language £ of cardinality A such that T does not have e-least models and e-Sp(T) > max{2w, A}.
Assertion of Proposition 13 can be improved as follows.
Proposition 14. For any infinite cardinality A there is an E-combination T = Th(AE) of LU-theories in a language £ of cardinality A such that T does not have e-least models and e-Sp(T) = 2A.
Proof. Let £ be a language consisting, for some natural n, of n-ary predicate symbols Ri, i < A. Choose a cardinality i £ (w \ {0}) U {w}. For any £' C £ we take a structure Aof the cardinality i such that Ri = (A^ )n
for Ri G £', and Ri = 0 for Ri G £ \ £'. Clearly, each structure AS' has a LU-theory and As' ^ As» for £' = £". For the ^-combination Ae of the structures AS' we obtain the theory T = Th(AE) having a model of the cardinality A. At the same time AE has 2X distinct theories of the ^-classes As. Thus, e- Sp(T) = 2a. Finally we note that T does not have e-least models by Theorem 1 and arguments for Proposition 6. □
Remark 6. LU-theories in the proof of Proposition 14 can be easily transformed to IILU-theories with the same effect for the e-spectrum.
4. Cantor-Bendixson ranks for language uniform theories
Recall the definition of the Cantor-Bendixson rank. It is defined on the elements of a topological space X by induction: CBx(p) > 0 for all p £ X; CBx (p) > a if and only if for any в < a, p is an accumulation point of the points of CBx-rank at least в. CBx (p) = a if and only if both CBx(p) > a and CBx(p) ^ a + 1 hold; if such an ordinal a does not exist then CBx(p) = x>. Isolated points of X are precisely those having rank 0, points of rank 1 are those which are isolated in the subspace of all non-isolated points, and so on. For a non-empty C С X we define CBx(C) = sup{CBx(p) | p £ C}; in this way CBx(X) is defined and CBx({p}) = CBx(p) holds. If X is compact and C is closed in X then the sup is achieved: CBx(C) is the maximum value of CBx(p) for p £ C; there are finitely many points of maximum rank in C and the number of such points is the CBx-degree of C. If X is countable and compact then CBx (X) is a countable ordinal and every closed subset has ordinal-valued rank and finite CBx-degree.
Clearly, for any set F, where ClE(TF) does not have the least generating set, CBjp(TF) = те.
Theorem 3. For any countable ordinal a and a natural number n > 0, there is an E-closed family TFa of LU-theories such that CBj-Fa (TFa) = a and its CBtf^ -degree is equal to n.
Proof. If a = 0 it suffices to take n singletons F0i1 ,...,F0,n. If a = 1 we take n disjoint copies F1j, j = 1,...,n, of Fq in Remark 3, each of which is ordered as (oj,<) and LlFoj = ДШ-Fijj j = 1, ■ ■ ■ ,n. We set Fq =
n
F0i U.. .UF0,n, F1 = F0 U U F1}j. If a > 1 is finite and Fa is already defined
j=i
then we add ш new disjoint copies Fa+1m of Fq related to each element in Fa \ Fa_i, each of which is ordered as (oj,<) and fm = limFn,+i m, fm £ Fa \ Fa-1. In such a case, CB(F0 j) = a + 1 and CB-degree is equal to n.
In general case, if a is limit we take TFa as the union of Tfp for / < a with uj disjoint copies of Fq such that each element in TF/3 is the limit lim of unique new copy of Fq and vice versa. Otherwise, if a = / + 1, we add w disjoint copies of Fq such that the set of these new copies F are in the bijective correspondence with the set of elements f, added in the step /, and / = lim F.
The inductive process guarantees that CBj^ (TFa) = a and CBj^ -degree is equal to n. □
References
1. Sudoplatov S.V. Complete theories with finitely many countable models. II. Algebra and Logic, 2006, vol. 45, no 3, pp. 180-200.
2. Sudoplatov S.V. Group polygonometries. Novosibirsk, NSTU, 2011, 2013.302 p. [in Russian]
3. Sudoplatov S.V. Models of cubic theories. Bulletin of the Section of Logic, 2014, vol. 43, no 1—2, pp. 19-34.
4. Sudoplatov S.V. Classification of Countable Models of Complete Theories. Novosibirsk, NSTU, 2014. [in Russian]
5. Sudoplatov S.V. Combinations of structures. arXiv:1601.00041v1 [math.LO], 2016. 19 p.
6. Sudoplatov S.V. Closures and generating sets related to combinations of structures. Reports of Irkutsk State University. Series "Mathematics", 2016, vol. 16, pp. 131144.
Sudoplatov Sergey Vladimirovich, Doctor of Sciences (Physics and Mathematics), Docent; Leading Researcher, Sobolev Institute of Mathematics SB RAS, 4, Academician Koptyug Avenue, Novosibirsk, 630090, tel.: (383)3297586; Head of Chair, Novosibirsk State Technical University, 20, K. Marx Avenue, Novosibirsk, 630073, tel.: (383)3461166; Docent, Novosibirsk State University, 1, Pirogova st., Novosibirsk, 630090, tel.: (383)3634020; Principal Researcher, Institute of Mathematics and Mathematical Modeling, 125, Pushkina st., Almaty, Kazakhstan, 050010, tel.: +7(727)2720046 (e-mail: sudoplat@math.nsc.ru)
С. В. Судоплатов
Семейства сигнатурно однородных теорий и их порождающие множества
Аннотация. Вводится понятие сигнатурно однородной теории и изучаются топологические свойства, относящиеся к семействам сигнатурно однородных теорий и их Я-совмещениям. Показано, что класс сигнатурно однородных теорий достаточно широк. Найдены достаточные условия сигнатурного подобия сигнатурно однородных теорий. Изучены свойства сигнатурного доминирования и бесконечного
сигнатурного доминирования. Найдена характеризация для E-замыкания семейства сигнатурно однородных теорий в терминах индексных множеств. Рассмотрен класс линейно упорядоченных семейств сигнатурно однородных теорий и к этому классу применена указанная характеризация. Исследованы свойства дискретных и плотных индексных множеств: показано, что любое дискретное индексное множество задает наименьшее порождающее множество, в то время как плотные индексные множества определяют по меньшей мере континуальное число точек накопления и замыкания без наименьших порождающих множеств. В частности, при наличии плотного индексного множества теория соответствующего E-совмещения не имеет е-наименьшей модели и не является малой. Используя дихотомию для дискретных и плотных индексных множеств, решается проблема существования наименьшего порождающего множества относительно E-совмещений и характеризуется это существование в терминах порядков.
Получены значения е-спектров для семейств сигнатурно однородных теорий. Показано, что любой е-спектр может быть реализован некоторым E-совмещением сиг-натурно однородных теорий. Найдены нижние оценки для е-спектров относительно мощностей сигнатур.
Показано, что семейства сигнатурно однородных теорий задают произвольный ранг Кантора - Бендиксона и произвольную степень относительно этого ранга.
Ключевые слова: E-совмещение, Р-совмещение, оператор замыкания, порождающее множество, сигнатурно однородная теория.
Список литературы
1. Судоплатов С. В. Полные теории с конечным числом счётных моделей. II / С. В. Судоплатов // Алгебра и логика. - 2006. - Т. 45, № 3. - С. 314-353.
2. Судоплатов С. В. Полигонометрии групп / С. В. Судоплатов. - Новосибирск : Изд-во НГТУ, 2011, 2013. - 302 с.
3. Sudoplatov S. V. Models of cubic theories / S. V. Sudoplatov // Bulletin of the Section of Logic. - 2014. - Vol. 43, N 1-2. - P. 19-34.
4. Судоплатов С. В. Классификация счетных моделей полных теорий. Ч. 1, 2 / С. В. Судоплатов. - Новосибирск : Изд-во НГТУ, 2014.
5. Sudoplatov S. V. Combinations of structures / S. V. Sudoplatov. -arXiv:1601.00041v1 [math.LO]. - 2016. - 19 p.
6. Sudoplatov S. V. Closures and generating sets related to combinations of structures / S. V. Sudoplatov // Изв. Иркут. гос. ун-та. Сер. Математика. -2016. - Т. 16. - С. 131-144.
Судоплатов Сергей Владимирович, доктор физико-математических наук, доцент; ведущий научный сотрудник, Институт математики им. С. Л. Соболева СО РАН, 630090, Новосибирск, пр. Академика Коптюга, 4, тел.: (383)3297586; заведующий кафедрой алгебры и математической логики, Новосибирский государственный технический университет, 630073, Новосибирск, пр. К. Маркса, 20, тел. (383)3461166; доцент, кафедра алгебры и математической логики, Новосибирский государственный университет, 630090, Новосибирск, ул. Пирогова, 1, тел. (383)3634020; главный научный сотрудник, Институт математики и математического моделирования МОН РК, 050010, Казахстан, Алматы, ул. Пушкина, 125, тел. +7(727)2720046 (e-mail: sudoplat@math.nsc.ru)
