ebook img

Algebras of Sets and Combinatorics PDF

264 Pages·2002·13.698 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Algebras of Sets and Combinatorics

Translations of MATHEMATICAL MONOGRAPHS Volume 214 Algebras of Sets and Combinatorics L. S. Grinblat American Mathematical Society Algebras of Sets and Combinatorics Translations of MATHEMATICAL MONOGRAPHS Volume 214 Algebras of Sets and Combinatorics L. S. Grinblat American Mathematical Society Providence, Rhode Island EDITORIAL COMMITTEE AMS Subcommittee Robert D. MacPherson Grigorii A. Margulis James D. Stasheff (Chair) ASL Subcommittee Steifen Lempp (Chair) IMS Subcommittee Mark I. Freidlin (Chair) Л. Ш. Гринблат АЛГЕБРЫ МНОЖЕСТВ И КОМБИНАТОРИКА Translated from an original Russian manuscript by A. Stoyanovskii. Translation edited by A. B. Sossinsky. 2000 Mathematics Subject Classification. Primary 03E05; Secondary 28A05, 54D35. Library of Congress Cataloging-in-Publication Data Grinblat, L. S. (Leonid S.), 1944- [Algebry mnozhestv i kombinatorika. English] Algebras of sets and combinatorics / L. S. Grinblat. p. cm. — (Translations of mathematical monographs, ISSN 0065-9282 ; v. 214) Includes bibliographical references and index. ISBN 0-8218-2765-0 (alk. paper) 1. Combinatorial set theory. I. Title. II. Series. QA248.G753 2002 511.3,22-dc21 2002074584 CIP Copying and reprinting. Individual readers of this publication, and nonprofit libraries acting for them, are permitted to make fair use of the material, such as to copy a chapter for use in teaching or research. Permission is granted to quote brief passages from this publication in reviews, provided the customary acknowledgment of the source is given. Republication, systematic copying, or multiple reproduction of any material in this publication is permitted only under license from the American Mathematical Society. Requests for such permission should be addressed to the Acquisitions Department, American Mathematical Society, 201 Charles Street, Providence, Rhode Island 02904-2294, USA. Requests can also be made by e-mail to [email protected]. © 2002 by the American Mathematical Society. All rights reserved. The American Mathematical Society retains all rights except those granted to the United States Government. Printed in the United States of America. 0 The paper used in this book is acid-free and falls within the guidelines established to ensure permanence and durability. Visit the AMS home page at http: //www. ams. org/ 10 9 8 7 6 5 4 3 2 1 07 06 05 04 03 02 Contents Chapter 1. Introduction 1 Chapter 2. Main Results 11 Chapter 3. The Main Idea 25 Chapter 4. Finite Sequences of Algebras (1). Proof of Theorems 2.1 and 2.2 39 Chapter 5. Countable Sequences of Algebras (1). Proof of Theorem 2.4 57 Chapter 6. Proof of the Gitik-Shelah Theorem, and More from Set Theory 71 Chapter 7. Proof of Theorems 1.17, 2.7, 2.8 83 Chapter 8. Theorems on Almost cr-Algebras. Proof of Theorem 2.9 93 Chapter 9. Finite Sequences of Algebras (2). The Function g(n) 109 Chapter 10. A Description of the Class of Functions 145 Chapter 11. The General Problem. Proof of Theorems 2.15 and 2.20 163 Chapter 12. Proof of Theorems 2.21(1,3), 2.24 175 Chapter 13. The Inverse Problem 179 Chapter 14. Finite Sequences of Algebras (3). Proof of Theorems 2.27, 2.31, 2.36, 2.38 181 Chapter 15. Preliminary Notions and Lemmas 197 Chapter 16. Finite Sequences of Algebras (4). Proof of Theorems 2.39(1,2), 2.45(1,2) 207 Chapter 17. Countable Sequences of Algebras (2). Proof of Theorems 2.29, 2.32, 2.46 229 Chapter 18. A Refinement of Theorems on cr-Algebras. Proof of Theorems 2.34, 2.44 235 Chapter 19. Semistructures and Structures of Sets. Proof of Theorem 2.48 239 Chapter 20. Final Comments. Generalization of Theorem 2.1 245 Appendix. On a Question of Grinblat by Saharon Shelah 247 CONTENTS VI 251 Bibliography 255 Index CHAPTER 1 Introduction 1.1. This monograph contains the results of [Gr] as well as many new results I have obtained in recent years. The new results and the results from [Gr] are so intertwined that it did not seem reasonable to publish all the new results in a separate very large paper. In contrast to [Gr], in the present monograph the theory is complete. Thus, the statements to which we refer without proofs and which we use for proving the main results are widely known and are simple facts of modern mathematics. This also means that all the main results entirely belong to “naive” set theory (except for the results in the Appendix written by Saharon Shelah). Let us start with some history. George Cantor, the creator of set theory, be­ gan studying sets in 1872. His starting point were papers on trigonometric series inspired by the works of Riemann. In 1878 Cantor proposed the continuum hypoth­ esis (Hi = 2N°) which was placed first in Hilbert’s 1900 list of problems bequeathed by him to the 20th century (see [Hi]). Breaking the chronology of our story, let us mention that the solution of the first Hilbert problem turned out to be striking. First, in the paper [Go] published in 1939, Godel proved the consistency of the continuum hypothesis with the axioms of set theory, and in the early 60’s Paul Co­ hen developed a powerful method for constructing models, the forcing method, and constructed a model of set theory in which Hi < 2**° (see [Col]).1 It became clear that the persistent efforts of Cantor and other mathematicians to prove the con­ tinuum hypothesis were of no avail. The continuum hypothesis, like Euclid’s fifth postulate, can be neither proved nor disproved. The last papers of Cantor date back to 1895-1897; .they are mostly devoted to the theory of completely ordered sets and to the calculus of ordinal numbers.2 In 1902 Lebesgue published his thesis [LI], which became the starting point of modern measure theory. The notions of a Lebesgue measurable and nonmeasurable set became fundamental in mathematics. New set theory problems arose. In the book [L2], published in 1904, Lebesgue posed the following problem. The Lebesgue Problem. Does there exist a nonnegative real function /i on the set of all subsets of the interval X = (0,1) that satisfies the following conditions: (a) if Mi and M2 are isometric sets, then /¿(Mi) = /¿(M2); 1 Although the reader is not assumed to be familiar with the forcing method (except in the Appendix), it is related to our monograph. Some examples use results obtained by the forcing method. The important theorem of Gitik-Shelah stated below was originally proved by the forcing method. The infinite analog of the Lusin theorem obtained by Gitik and Shelah — a result close to our subjects and discussed in Chapter 6 — was also proved by the forcing method, and up to now its purely mathematical proof is not known. An interesting result cited in 5.28 and presented in the Appendix was obtained by Saharon Shelah by the forcing method as the answer to a question of the author. In 6.12 we discuss an interesting result of Shelah obtained using the forcing method. 2The above information about Cantor is taken from [Bol] and [Cofc]. 1 2 1. INTRODUCTION (b) if Mi,..., Mfc,... is an at most countable sequence of pairwise disjoint sets, then M(U^fc) = X X M fc); (c) n(X) = 1? In 1905 Vitali [Vi] obtained the presently well-known negative solution of this problem. 1.2. The following problem was posed by Banach. The Banach Problem. What is the solution of the Lebesgue problem with condition (a) replaced by the following: fJ>{{x}) = 0 for each point x G XI In 1929 Banach and Kuratowski published a negative solution of this problem under the assumption of the continuum hypothesis in [B-Ku].3 Probably around that time, many specialists believed that, even without the continuum hypothesis assumption, the Banach problem has a negative solution. Thus, in a well-known book [Bi] by G. Birkhoff published in 1948, one of the problems is posed as follows: prove the nonexistence of a nontrivial measure with countable additivity4 for all subsets of the continuum, such that any point has measure 0, without assuming the continuum hypothesis. 1.3. Remark. Here it is appropriate to recall the history of the study of other problems whose starting point is the Lebesgue problem formulated in 1.1. (We use the historical survey from [Bo2].) In 1914 Hausdorff proved in [Haul] that there does not exist a nontrivial finitely additive measure that is well defined for all subsets of the unit ball in the 3-dimensional Euclidean space and invariant with respect to isometries. It was natural to investigate whether the same is true for the line and the plane — the problem was solved brilliantly by Banach in 1923, who showed in [B] that, on the contrary, it is just in these two cases that such a measure does exist. In 1929 von Neumann showed in [Nel] that the reason for the difference between the line and the plane on one hand, and the Euclidean space of dimension greater than 2 on the other hand, has to do with the commutativity of the group of rotations of the plane. Therefore the a-additivity condition in the Lebesgue problem cannot be replaced by the finite additivity condition. Using the notion of ultrafilter, one easily shows that the cr-additivity condition in the Banach problem cannot be replaced by the finite additivity condition. Note that the idea of the “averaging” method from [B] was used by Haar, who proved in 1933 (see [H]) the existence of an invariant measure for locally compact groups. This discovery immediately allowed von Neumann to solve (see [Ne2]), for compact groups, the fifth Hilbert problem on the characterization of Lie groups by purely topological properties. 3The result of Banach-Kuratowski follows from Theorem 1.6. It also follows from Theo­ rem 6.8 (see 6.9). 4We consider only real nonnegative finite measure (i.e., the measure of a measurable set is a real nonnegative finite number). A measure is called nontrivial if there exists a measurable set whose measure is greater than 0. As usual, a countable additivity property is also called cr- additivity, and a cr-additive measure is called, for short, a cr-measure. Sometimes we will consider not only cr-additive measures, but also finitely additive measures which are not, in general, cr- additive. 1. INTRODUCTION 3 1.4. In 1971, in the period started by Cohen’s great discovery, the paper [So] by R. Solovay appeared with the following remarkable result: Solovay’s Result. In a certain model there exists a a-additive extension of the Lebesgue measure to all subsets of (0,1). 1.5. Let us return to the Banach-Kuratowski result mentioned in 1.2. After this, S. Ulam proved in [Ul] the following important more general result: Ulam’s Matrix. If a set X has cardinality Ni, then one can construct a ma­ trix of subsets of X, (Ml Ml ... Ml ...\ Ml Ml ... Ml ... Ml Ml ... Ml ... v................ which has No tows and Ni columns, and (a) Ml D M^ = 0 if a /?; (b) # (* \ u m £ )< n0.s 1.6. The following theorem, due to Ulam, is an obvious corollary of the exis­ tence of Ulam’s matrix: Theorem. Let p be a nontrivial a-measure on a set of cardinality Ni, and let p({x}) = 0 for each one-point set {#}. Then there exist Ni pairwise disjoint p-nonmeasurable sets. 1.7. Definition. A finitely additive measure p defined on a set X is said to be two-valued if p({x}) = 0 for all x e X, p(X) = 1, and for any ¿¿-measurable set M either p(M) = 0 or p(M) = 1. If p is also cr-additive, then it is said to be a-two-valued. 1.8. The following Alouglu-Erdos theorem stated in [Er] follows rather easily from Theorem 1.6. The Alouglu-Erdos Theorem. Assume that a countable sequence of a- two-valued measures on a set of cardinality Ni is given. Then there exists a set that is nonmeasurable with respect to each of these measures. It is clear that in the Alouglu-Erdos theorem the cr-two-valuedness condition is inessential. The theorem is valid under the assumption that each of the measures is nontrivial, cr-additive, and each one-point set has measure 0. 1.9. The Alouglu-Erdos theorem gives a partial answer to the following Ulam problem stated in [Er] (see also [Er-Ha]). The Ulam Problem. Find the minimal cardinal6 x* such that, for any fam­ ily (of cardinality less than x*) of a-two-valued measures defined on a set of car­ dinality Ni, there exists a set that is nonmeasurable with respect to each of the measures in the family. 5 6 5In 6.4 we give a construction of the sets Ml. The symbol #(M) denotes the cardinality of the set M. (In the Appendix the cardinality of a set M is denoted by \M\.) 6By a cardinal x we mean an ordinal number such that if a < x, then #(c*) < #{x). In what follows, the cardinality of a set will be treated as a cardinal. 4 1. INTRODUCTION Ulam proved that x* > No- By the Alouglu-Erdos theorem, x* > N1. In [S2] Shelah has shown that in a certain model x* = N1. Note that in the Godel model L, we have x* = N2. 1.10. In [Wo] Woodin improved Shelah’s result. He showed that in a certain model there exists a a-two-valued measure ¡jl on a set X of cardinality N1 and a family {Ma}a<u;i of ju-nonmeasurable sets such that {a | #i(Aia \ M) = 0} ^ 0 for each M C X with ¿¿(M) ^ 0. For each Ma let us define a a-two-valued measure on X as follows: = 0 if and only if fi(M fl Ma) = 0. It is evident that each subset of X is measurable with respect to some measure fia. Therefore in the model considered by Woodin we have x* = N1. 1.11. We will return to the history of the subject later. Now let us pass to the main theme of our monograph. Definition. By an algebra A on a set X we mean a nonempty system of subsets of X possessing the following property: if Mi, M2 G A, then M\ U M2, Mi \ M2 G A. Unless specified otherwise, all algebras and measures are considered on some abstract set X ^ 0. Unless the contrary follows from the context, by a set we always mean a subset of X. 1.12. The following “general question” arises in connection with the Alouglu- Erdos theorem and the Ulam problem. Assume we are given a family of measures {¡i\}xeA (since the question is “general”, we do not specify whether they are a- additive and so forth), and for each A G A there exists a /¿A-nonmeasurable set. Does there exist a set nonmeasurable with respect to each of these measures? She­ lah’s and Woodin’s results are nontrivial examples giving a negative answer to our question. But in these examples we deal with families such that #(A) = N1. And what will be the answer if one considers families with #(A) < Ho? Let us ask our question in a more general and, I believe, more natural form. Let us consider alge­ bras instead of measures. Thus, consider an at most countable family of algebras, each of which is not equal to il(A’) (as usual, ty(X) denotes the set of all subsets of AT). Does there exist a set which does not belong to any of these algebras? If we are given two algebras A\ and A2, and each of them does not coincide with ^(X), then it is easy to show that A ^ A 2 ^ {X ). 1.13. However, if we take not two but three algebras each of which is not equal to 93(AT), then it can happen that a set not belonging to any one of these algebras does not exist. Example. #(X) = 3 and X = {xi,X2,X3}. Define the algebras ^4i,^42)^3 by listing their elements: Ai = {0 , {xi}, {a?2) 2:3}, AT}, A2 = {0> {#2}) 1 > ^3}) -^0 > ^3 = {0 i{®3}»

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.