THE NUMBER OF UNARY CLONES CONTAINING THE PERMUTATIONS ON AN INFINITE SET

Similar documents
Interpolation of κ-compactness and PCF

GUESSING MODELS IMPLY THE SINGULAR CARDINAL HYPOTHESIS arxiv: v1 [math.lo] 25 Mar 2019

Axiomatization of generic extensions by homogeneous partial orderings

Sy D. Friedman. August 28, 2001

arxiv:math/ v1 [math.lo] 15 Jan 1991

COMBINATORICS OF REDUCTIONS BETWEEN EQUIVALENCE RELATIONS

PARTITIONS OF 2 ω AND COMPLETELY ULTRAMETRIZABLE SPACES

CARDINALITIES OF RESIDUE FIELDS OF NOETHERIAN INTEGRAL DOMAINS

DEPTH OF BOOLEAN ALGEBRAS SHIMON GARTI AND SAHARON SHELAH

arxiv:math/ v1 [math.lo] 9 Dec 2006

arxiv: v3 [math.lo] 23 Jul 2018

Silver type theorems for collapses.

Chapter 4. Cardinal Arithmetic.

ADDING A LOT OF COHEN REALS BY ADDING A FEW II. 1. Introduction

The Semi-Weak Square Principle

Cardinal arithmetic: The Silver and Galvin-Hajnal Theorems

The (λ, κ)-fn and the order theory of bases in boolean algebras

CONSECUTIVE SINGULAR CARDINALS AND THE CONTINUUM FUNCTION

HEIKE MILDENBERGER AND SAHARON SHELAH

Level by Level Inequivalence, Strong Compactness, and GCH

being saturated Lemma 0.2 Suppose V = L[E]. Every Woodin cardinal is Woodin with.

Continuous images of closed sets in generalized Baire spaces ESI Workshop: Forcing and Large Cardinals

UPWARD STABILITY TRANSFER FOR TAME ABSTRACT ELEMENTARY CLASSES

Generalising the weak compactness of ω

PERFECT TREE FORCINGS FOR SINGULAR CARDINALS

MITCHELL S THEOREM REVISITED. Contents

arxiv: v2 [math.lo] 13 Feb 2014

The finite lattice representation problem and intervals in subgroup lattices of finite groups

Tall, Strong, and Strongly Compact Cardinals

Determinacy models and good scales at singular cardinals

Laurence Boxer and Ismet KARACA

Generalization by Collapse

Attempt QUESTIONS 1 and 2, and THREE other questions. Do not turn over until you are told to do so by the Invigilator.

SHIMON GARTI AND SAHARON SHELAH

LARGE CARDINALS AND L-LIKE UNIVERSES

arxiv: v2 [math.lo] 21 Mar 2016

2. The ultrapower construction

Characterizing large cardinals in terms of layered partial orders

A relative of the approachability ideal, diamond and non-saturation

Strongly compact Magidor forcing.

Non replication of options

Chain conditions, layered partial orders and weak compactness

A precipitous club guessing ideal on ω 1

Unary PCF is Decidable

arxiv: v1 [math.lo] 27 Mar 2009

ON THE QUOTIENT SHAPES OF VECTORIAL SPACES. Nikica Uglešić

Bounds on coloring numbers

CATEGORICAL SKEW LATTICES

Global singularization and the failure of SCH

Extender based forcings, fresh sets and Aronszajn trees

4: SINGLE-PERIOD MARKET MODELS

STRONGLY UNFOLDABLE CARDINALS MADE INDESTRUCTIBLE

AN INFINITE CARDINAL-VALUED KRULL DIMENSION FOR RINGS

Building Infinite Processes from Regular Conditional Probability Distributions

Short Extenders Forcings II

Chromatic number of infinite graphs

COLLAPSING SUCCESSORS OF SINGULARS

FORCING AND THE HALPERN-LÄUCHLI THEOREM. 1. Introduction This document is a continuation of [1]. It is intended to be part of a larger paper.

Philipp Moritz Lücke

A HIERARCHY OF RAMSEY-LIKE CARDINALS

Laurence Boxer and Ismet KARACA

Fat subsets of P kappa (lambda)

Equivalence between Semimartingales and Itô Processes

Annals of Pure and Applied Logic

On almost precipitous ideals.

Commentationes Mathematicae Universitatis Carolinae

The illustrated zoo of order-preserving functions

Large Cardinals with Few Measures

Gödel algebras free over finite distributive lattices

arxiv: v1 [math.lo] 24 Feb 2014

Math 280B Winter Recursion on Well-Founded Relations. 6.1 Recall: For a binary relation R (may be a proper class): T 0 = A T n+1 = pred R (a)

The Outer Model Programme

Covering properties of derived models

Introduction to Priestley duality 1 / 24

MODIFIED EXTENDER BASED FORCING

ANNALES ACADEMIÆ SCIENTIARUM FENNICÆ DIAMONDS ON LARGE CARDINALS

Theorem 1.3. Every finite lattice has a congruence-preserving embedding to a finite atomistic lattice.

Variations on a theme by Weetman

1. Introduction. As part of his study of functions defined on product spaces, M. Hušek introduced a family of diagonal conditions in a topological

More On λ κ closed sets in generalized topological spaces

On the strengths and weaknesses of weak squares

ON THE SINGULAR CARDINALS. A combinatorial principle of great importance in set theory is the Global principle of Jensen [6]:

Notes to The Resurrection Axioms

COMBINATORICS AT ℵ ω

Game Theory: Normal Form Games

Collinear Triple Hypergraphs and the Finite Plane Kakeya Problem

An effective perfect-set theorem

Research Article On Open-Open Games of Uncountable Length

Strongly Unfoldable Cardinals Made Indestructible

arxiv: v1 [math.lo] 8 Oct 2015

PURITY IN IDEAL LATTICES. Abstract.

RUDIN-KEISLER POSETS OF COMPLETE BOOLEAN ALGEBRAS

Generic embeddings associated to an indestructibly weakly compact cardinal

Finite Additivity in Dubins-Savage Gambling and Stochastic Games. Bill Sudderth University of Minnesota

Satisfaction in outer models

Two Stationary Sets with Different Gaps of the Power Function

LATTICE EFFECT ALGEBRAS DENSELY EMBEDDABLE INTO COMPLETE ONES

INTERVAL DISMANTLABLE LATTICES

6. Recursion on Well-Founded Relations

Set- theore(c methods in model theory

Transcription:

THE NUMBER OF UNARY CLONES CONTAINING THE PERMUTATIONS ON AN INFINITE SET MICHAEL PINSKER Abstract. We calculate the number of unary clones (submonoids of the full transformation monoid) containing the permutations, on an infinite base set. It turns out that this number is quite large, on some cardinals as large as the whole clone lattice. Moreover we find that, with one exception, even the cardinalities of the intervals between the monoid of all permutations and the maximal submonoids of the full transformation monoid are as large. Whether or not the only exception is of the same cardinality as the other intervals depends on additional axioms of set theory. 1. Background and the result Fix a set X and consider for all n 1 the set O (n) of n-ary operations on X. If we take the union O = n 1 O(n) over these sets, we obtain the set of all operations on X of finite arity. A clone is a subset of O which contains all functions of the form π n k (x 1,..., x n ) = x k (1 k n), called the projections, and which is closed under composition of functions. With the order of set-theoretical inclusion, the clones on X form a complete algebraic lattice Cl(X). We wish to describe this lattice for infinite X, in which case it has cardinality 2 2 X. A clone is called unary iff it contains only essentially unary functions, i.e., functions which depend on only one variable. Unary clones correspond in an obvious way to submonoids of the full transformation monoid O (1) and we shall not distinguish between the two notions in the following. We say that a unary clone C O (1) is precomplete or maximal iff C together with any unary function f O (1) \ C generates O (1), i.e. iff the smallest clone containing C as well as f is O (1). In [Pin], the author determined all precomplete submonoids of the full transformation monoid O (1) that contain the permutations for all infinite X, which was a generalization from the countable ([Gav65]). The number of such clones turned out to be rather small compared with the size of the clone lattice: On an infinite set X of 1991 Mathematics Subject Classification. Primary 08A40; secondary 08A05. Key words and phrases. clone lattice, permutations, unary clones, transformation monoid, submonoids. Support by DOC [Doctoral Scholarship Programme of the Austrian Academy of Sciences], and later by the Postdoctoral Fellowship of the Japan Society for the Promotion of Science (JSPS) is gratefully acknowledged. 1

2 M. PINSKER cardinality ℵ α there exist 2 α +5 precomplete unary clones, so in particular there are only five precomplete unary clones on a countably infinite set X. Theorem 1. Let X be an infinite set of cardinality κ. If κ is regular, then the precomplete submonoids of O (1) that contain the permutations are exactly the monoid A and the monoids G ξ and M ξ for ξ = 1 and ℵ 0 ξ κ, ξ a cardinal, where A = {f O (1) : f 1 [{y}] is small for almost all y X} G ξ = {f O (1) : f is ξ-injective or not ξ-surjective} M ξ = {f O (1) : f is ξ-surjective or not ξ-injective} If κ is singular, then the same is true with the monoid A replaced by A = {f O (1) : ξ < κ ( f 1 [{x}] ξ for almost all x X ) }. In the theorem, a set is small iff it has cardinality smaller than the cardinality of X, a property holds for almost all y X iff it holds for all y X except for a small set, a function f O (1) is ξ-surjective iff X \ f[x] < ξ, and it is ξ-injective iff there is a set Y X of cardinality smaller than ξ such that the restriction of f to X \ Y is injective. With this result, the question arose whether it was possible to describe the whole interval [S, O (1) ] of the clone lattice, where S is the set of permutations of X. We show that compared to the number of its dual atoms, this interval is quite large. In particular, on a countably infinite set X it equals the size of the whole clone lattice. Theorem 2. Let X be an infinite set of cardinality κ = ℵ α. Then there exist 2 2λ submonoids of O (1) which contain all permutations, where λ = max{ α, ℵ 0 }. Moreover, if κ is regular, then [S, G ] = 2 2λ for every precomplete monoid above S ; in fact, [S, D] = 2 2λ, where D is the intersection of the precomplete elements of [S, O (1) ]. If κ is singular, then [S, G ] = 2 2λ for all precomplete monoids except A : If λ < κ, then [S, A ] = [S, D] = 2 2λ, but if λ = κ, then [S, A ] = [S, D] = 2 (κ<κ ) (where κ <κ = sup{κ ξ : ξ < κ}). 1.1. Notation. For any set Y, we denote the power set of Y by P(Y ). The smallest clone containing a set of functions F O is denoted by F. If f O (1), we write ker(f) P(X) for the kernel of f. 2. The proof of Theorem 2 Definition 3. Set K = {ξ : ξ a cardinal and ξ κ}; then K = λ. Define for every f O (1) a function s f : K K ξ {A ker(f) : A = ξ} In words, the function assigns to every ξ κ the number of equivalence classes in the kernel of f which have cardinality ξ. We call s f the kernel sequence of f.

UNARY CLONES CONTAINING THE PERMUTATIONS 3 Lemma 4. If f, g O (1) are unary functions satisfying s f = s g and X \ f[x] = X \ g[x], then there exist β, γ S such that f = β g γ. Proof. The assumption s f = s g implies that there is γ S such that ker(f) = ker(g γ). Obviously, f[x] = g[x] = g γ[x] as s f = s g. Together with the fact that X \ f[x] = X \ g[x] this implies that we can find β S such that f[x] = β g γ[x], and since ker(f) = ker(g γ) also so that f = β g γ. Proposition 5. The number µ of submonoids of O (1) containing S is at most 2 2λ. Proof. By the preceding lemma, the clone a function f O (1) generates together with S is determined by s f and the cardinality of X \ f[x]. There exist at most λ λ different kernel sequences and λ possibilities for the cardinality of the complement of the range of a function in O (1). Thus, modulo S there are only λ λ λ = λ λ = 2 λ different functions in O (1). Therefore, µ 2 2λ. We will now show the other inequality. Fix any sequence (n i ) i ω of natural numbers such that j<i n j < n i for all i ω. Set R = {n i } i ω {ξ K : ξ infinite successor}. Then R = K = λ. For all f O (1), write s f = s f R for the restriction of its kernel sequence to R. Observe that for all ξ R we have that η<ξ,η R η < ξ: For ξ finite, this is because we chose the finite elements of R that way, and if ξ is infinite, then it is a successor cardinal so that the left side of the inequality is clearly bounded by its predecessor. We say that A R is unbounded iff ξ A ξ = κ. Assign to every unbounded A R a function f A O (1) satisfying s fa (ξ) = 1 whenever ξ A, and s fa (ξ) = 0 whenever ξ K \ A. The fact that A is unbounded guarantees the existence of f A. Lemma 6. If A R is unbounded and g O (1), then s g fa s fa. Proof. Consider an arbitrary B ker(g f A ) with B = ξ R. We claim there exists C B of cardinality ξ such that C ker(f A ). For suppose to the contrary this is not the case. Being an element of ker(g f A ), B is the union of sets in the kernel of f A : B = i δ B i, for B i ker(f A ) and some ordinal δ. By our assumption, B i < ξ for all i δ. Thus, B = i<δ B i D ker(f A ), D <ξ D = η A,η<ξ η < ξ, contradiction. So for all B ker(g f A ) with B = ξ R we injectively find C ker(f A ) of the same cardinality, which proves the lemma. Lemma 7. Let A, A 1,..., A n R be unbounded and such that A A i for all 1 i n. Then f A / {f A1,..., f An } S. Proof. Clearly, every unary t {f A1,..., f An } S which is not a permutation has a representation of the form t = g f Ai β, where g O (1),

4 M. PINSKER β S and 1 i n. But then s t s fai by the preceding lemma, so that s t s fa and therefore t f A. It is a fact that if Y is any set, then there exists a family I P(Y ) such that I = P(Y ) = 2 Y and such that the sets of I are pairwise incomparable, i.e., A B holds for all distinct A, B I. For example, it is a well-known theorem of Hausdorff that there exist independent families of subsets of Y of that size, where I P(Y ) is called independent iff every nontrivial Boolean combination of sets from I is nonempty, i.e., whenever B 1, B 2 I are finite, nonempty and disjoint, then A (Y \ A). A B 1 A B 2 See the textbook [Jec02, Lemma 7.7] for a proof of this. There is an independent family of unbounded subsets of R which has cardinality 2 λ : If I P(R) is independent of size 2 λ, then either I or I = {R \ A : A I } contains 2 λ unbounded sets, the family of which is independent. Proposition 8. There is an order embedding from P(2 λ ) into [S, O (1) ]. In particular, the number µ of submonoids of O (1) containing S is at least 2 2λ. Proof. Let I P(R) be an independent family of unbounded subsets of R with I = 2 λ. Define for every B I a monoid C B = {f A : A B} S. Then for all B 1, B 2 I we have that if B 1 B 2, then C B1 C B2 : Indeed, by the preceding lemma f A C B1 \C B2 for any A B 1 \B 2. Together with the fact that larger subsets of I yield larger clones, this implies that the mapping ϕ : P(I ) [S, O (1) ] assigning to every B I the clone C B is an order embedding. Hence, there exist P(I ) = 2 2λ distinct monoids containing the permutations. This completes the proof of the first statement of Theorem 2. Proposition 9. Let κ be regular and let D be the intersection of the precomplete submonoids of O (1) containing S. There is an order embedding of P(2 λ ) into [S, D]. Hence, [S, D] = 2 2λ. Proof. Since in the proof of Proposition 8 we considered only functions f O (1) with s f (κ) 1, all those functions were elements of A. Also, we did not care about the size of the complement of the range of f; if we assume it to be of cardinality κ, then all functions of the construction are not κ-surjective and therefore elements of G ξ, for all cardinals ξ = 1 and ℵ 0 ξ κ. Since for any unbounded A R and any small Y X there is ξ A with ξ > Y, the fact that f A has a class of size ξ in its kernel yields that f A is not injective on the complement of Y. Therefore, the f A used in the construction are not κ-injective and hence are elements of M ξ, for all ξ = 1 and ℵ 0 ξ κ. This proves the proposition.

UNARY CLONES CONTAINING THE PERMUTATIONS 5 We now turn to the case when κ is singular. The argument of the preceding proposition yields Proposition 10. Let κ be singular and let G A be a precomplete submonoid of O (1) containing S. There is an order embedding of P(2 λ ) into [S, G ]. In particular, [S, G ] = 2 2λ. Proposition 11. Let κ be singular such that λ < κ. Then [S, A ] = 2 2λ. Proof. Since the functions used in our construction satisfy s f (ξ) 1 for all ξ K, we have f[x] λ < κ, and hence f 1 [{x}] = 0 for almost all x X; therefore those functions are elements of A. Hence, [S, A ] = 2 2λ. Proposition 12. Let κ be singular such that λ = κ. Then [S, A ] = 2 (κ<κ). Proof. We first calculate the number of different kernel sequences of functions in A. Let s f : K K be such a sequence; then f A iff there is ξ < κ such that ξ η κ s f (η) = τ < κ. Fixing ξ and τ, we have κ τ possibilities for the part of s f between ξ and κ. Taking the sum over all τ < κ, we obtain κ <κ possibilities for s f between ξ and κ. Since below ξ there are no conditions on s f in order to make f an element of A, there are exactly κ ξ possibilities for the restriction of s f to ξ, so that we have a total of κ ξ + κ <κ kernel sequences of functions f with ξ η κ s f (η) < κ. Since ξ < κ can be arbitrary, we take the sum over all ξ < κ and find that there are κ <κ distinct kernel sequences of functions in A. Hence, [S, A ] 2 (κ<κ). Claim. There exists a family I of pairwise incomparable small unbounded subsets of R which has cardinality κ <κ. To prove this, we first observe that for all cf(κ) ξ < κ there exists a family I ξ of pairwise incomparable unbounded subsets of R of cardinality ξ such that I ξ = κ ξ (cf(κ) denotes the cofinality of κ). Indeed, write R = R R, where R and R are disjoint, and R is unbounded and of cardinality ξ. Now let I ξ be a family of pairwise incomparable subsets of R of cardinality ξ with I ξ = κξ. To see that I ξ exists, observe that every function f κ ξ is a subset of ξ κ, and that all those functions are incomparable as subsets of ξ κ. Thus a family of size κ ξ of pairwise incomparable sets of size ξ exists on ξ κ, and therefore also on R since R = ξ κ = κ. Now we set I ξ = {A R : A I ξ } to obtain the family I ξ having the desired properties. Finally to prove the claim, write R as a disjoint union R = cf(κ) ξ<κ R ξ of sets R ξ of cardinality κ (which also implies that they are unbounded). Fix a family I ξ of pairwise incomparable unbounded subsets of R ξ of cardinality ξ such that I ξ = κ ξ, for all ξ. Then the family I = cf(κ) ξ<κ I ξ consists of pairwise incomparable small unbounded subsets of R and has cardinality κ <κ. Having small range, the functions corresponding to the sets in I are all members of A, so that we obtain 2 (κ<κ) clones in the interval [S, A ].

6 M. PINSKER Proposition 13. Let κ be singular and let D be the intersection of the precomplete submonoids of O (1) containing S. If λ < κ, then [S, D] = 2 2λ. If λ = κ, then [S, D] = 2 (κ<κ). Proof. One only needs to combine the proofs of Propositions 9, 10, 11, and 12; we leave the details to the reader. Remark 14. If GCH holds, then 2 (κ<κ ) = 2 2κ, so in this case we have [S, D] = 2 2λ on all infinite X. However, for any singular κ it is is also consistent that 2 κ < 2 (κ<κ) < 2 2κ. Therefore, if κ is singular and λ = κ, then the intervals [S, A ] and [S, D] can be smaller than 2 2λ. In particular we have that whether or not the intervals [S, A ] and, say, [S, M 1 ] are of equal cardinality depends on the set-theoretical universe. References [Gav65] G. P. Gavrilov. On functional completeness in countable-valued logic (Russian). Problemy Kibernetiki, 15:5 64, 1965. [Jec02] T. Jech. Set theory. Monographs in Mathematics. Springer, third millenium edition edition, 2002. [Pin] M. Pinsker. Maximal clones on uncountable sets that include all permutations. Algebra Univers. to appear. Algebra, TU Wien, Wiedner Hauptstraße 8-10/104, A-1040 Wien, Austria E-mail address: marula@gmx.at URL: http://www.dmg.tuwien.ac.at