Introduction
Descriptive set theory
Borel equivalence relations
Countable Borel equivalence relations
Hyperfinite Borel equivalence relations
Lifts of Borel actions
Dichotomies for Polish modules
Quotients by countable normal subgroups are hyperfinite
Introduction
The theory of countable Borel equivalence relations has been applied in numerous areas of mathematics. In fact, they are equivalent to the universal countable Borel equivalence relationπΈβ, which is the most difficult countable Borel equivalence relation.
Preliminaries and examples
Proofs
Then there is a morphismπΊ βOutπ΅(πΉ), induced by the action ofπΊ onπΊ/π», so we get a liftπΊ βAutπ΅(πΉ). Then for everyπ there is a class-bijective rise πΊπ βAutπ΅(πΈ) such that the following diagram commutes:.
Lifts of Borel actions on quotient spaces
Introduction
Using this result and a variant of [KM04, Corollary 13.3], we show, in Corollary 3.3.11, that the full lifting problem has a positive solution generically for an arbitrary periodic (ie with all its classes infinite) CBERπΈ. A Borel action of a countable group πΊ on a standard Borel space π is pmp if it has an invariant Borel probability measure.
Preliminaries
πΊ The CBER caused by this action, i.e. an equivalence relation whose classes are exactly the orbits of this action. The Feldman-Moore theorem (see e.g. [Kec22, Theorem 3.3]) states that for every CBERπΈ onπ there exists a countable groupπΊ and a Borel actionπΊ onπ such that πΈ =πΈπ.
Borel actions on quotient spaces
So, if πΈ is incompressible and satisfies πΈ π΅ πΈ β (πΈΓπΌ . N), then every non-trivial countable group admits a non-external action on π/πΈ. It is shown in [RM21] that there exists a countable basis of pairs πΈ β πΉ CBER such that there is no Borel action πΊ β·π΅ π/πΈwithπΉ =πΈβ¨πΊ (see Section 3.8 for the exact statement). To see that, since im ππ is a πΈ-compressible perfect section forπΈ, there exists some Borel injectionsππ such that π(π₯) πΈ π₯ for everyπ₯ andππ(π) =im ππ.
LetπΈ be an aperiodic CBER on a Polish spaceπ. Then for every Borel actionπΊ β·π΅ π/πΈ there is a comeager πΈβ¨πΊ invariant Borel subsetπ β π such that πΊ β·π΅π/πΈ has a class bijective lift. Assume there is a free pmp action of a countable group πΊ on a standard probability space (π, π) with the following properties:. i) πΊ is co-Hopfian (that is, injective morphisms of πΊ in itself are surjective) and πΊ has no non-trivial finite normal subgroups (e.g. SL3(Z)); ii).
Outer actions
For CBERsπΈ β πΉ it is possible that πΈ is not normal in πΉ, but there is still a Borel action πΊ β·π΅ π/πΈ such that πΉ = πΈβ¨πΊ, as shown in the example at the beginning of section 3.3. On the other hand, if there is a smooth compound πΏ, then πΉ it must be compressible since it contains the aperiodic smooth πΏ. Then there is a link given by the action of {0} ΓZ, but there is no smooth link, because πΉ is not compressible.
By normality, any two πΈ-classes contained in the same class πΉ have the same cardinality, so by dividing the space into invariant Borel groups, we can assume that there are some N} such that every class πΈ has cardinalityπ. It is clear that if πΊ is a free set, then every external operation of πΊ has a raiser.
Outer actions of finite groups
Letπ» be a countable group, let (πΊπ)πbe a countable family of countable groups, letπ» βπΊπ be morphisms, and letπΊ be the merged free product of πΊπoverπ». If every outer action ofπ» has a class-bijective lift, and everyπ» βπΊπhas the class-bijective lift property, then every outer action ofπΊ lifts. Thus, by the universal property of merged products, there is a lift πΊ β. Let πΊ be a countable group and let π β³ πΊ be a finite normal subgroup such that every outer action of π» =πΊ/π has a class bijective lift.
This allows us to define an action of πΊ by setting π π₯ as the unique element in both [π₯]πΏπΊ and . Let π» be a finite group, let (πΊπ)π<β be finite groups, let π» β πΊπ be morphisms, and let πΊ be the aggregated free product of the πΊπ over π».
Outer actions of amenable groups
Given CBERs πΈ β πΉ, we say that πΉ/πΈ is hyperfinite if there is an increasing sequence(πΉπ)πof finite index expansions ofπΈ such thatπΉ =Γ. An Aβquasi-tiling of π΄ is a tuple C = (πΆπ΅)π΅βA of subsets of π΄ such that π΅π β π΄ for every π β πΆπ΅, and the family is disint} Let πΊ be an amenable group whose conjugation equivalence relation on its space of subgroups is smooth.
Let C be a transverse to the conjugation equivalence relation on the space of subgroups, and for each subgroup π» β€ πΊ, fix some ππ» β πΊ such that ππ»π» πβ1. So if for everyπ» β C we let be the union of the ergodic components with a stabilizerπ», then there is a (πΈ βΎ ππ», πΈβ¨π» βΎ ππ»)β.
Summary of lifting results for outer actions
Then πΈ0 is Borel reducible to the conjugation equivalence relation on π, which is therefore not smooth.
Additional topics
Given a pair πΈ β πΉ of CBERs, we say that From [RM21, Theorem 5], there exists a countable set of constraints to be generated by a Borel action. Two projections πandπ are Murray-von Neumann equivalents, written
Given Polishπ modulesπ andπ, we say thatπ embedsintoπ, denoted π βπ π, if there is a continuous linear injection of πinπ. We show that there is a countable basis of minimal uncountable Abelian Polish groups (one for each prime and one for characteristic 0).
Equidecomposition in cardinal algebras
Introduction
A classic theorem of Thorisson [Tho96] in probability theory states that if π₯and πΎthenπΎ π₯ andπ₯β²are equal in distribution. This characterization in terms of switching-merging has been applied to various areas of probability theory including graphs with random roots [Khe18], Brownian motion [PT15], and point processes [HS13]. Setting ππΎ to be the measure on π defined by ππΎ(π΄):= π({πΎ} Γπ΄), we see that πand βΞonπ such thatπ=Γ.
We thank Alexander Kechris for introducing the author to Thorisson's theorem and suggesting the use of cardinal algebras. We would also like to thank Ruiyuan (Ronnie) Chen for the statement in terms of equidecomposition as well as the reference to the Becker-Kechris lemma.
Preliminaries
More recently they have been used in [KM16] in the study of countable Borel equivalence relations. 17.2] The class of cardinals is a CA under addition (although we strictly require that a CA be a collection). If π is a measurable space, then the set B (π) of measurable sets is a GCA under disjoint union.
In CA induced by a π-complete, π-distributive network, β€ is the partial order induced by the network, i.e. π β€ πiffπ=πβ§π. For the cardinal class π β€ πiff there exists an injection π β©βπ and the fact that this is a partial order is the Cantor-Schroeder-Bernstein theorem.
Proof of main theorem
Applications
This also implies a special case of [Mil12, Theorem 24], which says that if dimπ (π) is uncountable, then there exists a linearly independent perfect set (see Corollary 5.5.2). πΉ β¨π»π₯β©π₯βπΉ, where the union is adopted over everything finiteπΉ β π, so since π is countable, there is a number ofπΉ for which π»πΉ :=β¨π»π₯β©π₯βπΉ is not meager, and therefore open, since π»πΉ is analytical. If π0 and π1 are Polish π modules with Baire measurable submodules π0 and π1, respectively, we write π0/π0 βπ π1/π1 if there is a continuous linear map π0β π1 descending to an injectionπ0/π 0 β©β π1/ π1 .
If π β€ π is a closed subring, then there exists a corresponding norm on π obtained by restricting the norm on π . By Theorem 5.1.1, there exists a minimal element β1(R) that is bistable with the usual space β1 of absolutely summable sequences.
A dichotomy for Polish modules
Introduction
A key feature of many of these theories (and all of the above examples) is the existence of dichotomy theorems, which state that either an object is simple or there is a canonical obstacle contained within it. This is usually stated in terms of preorders, saying that there is a natural basis for preordering objects that are not simple (recall that a basis for a preorder π is a subsetπ΅ β πsuch that for each π βπ there someπ π β π β ). A particularly good aspect of Polish modules is that the concept of "definable" reduction is much simpler than in the general case.
More precisely, we give a countable basis underβπ for Polish modules which are not countably generated. To contextualize this, we note that considering even very basic module homomorphisms (for example, inclusion of QintoRasQ vector spaces) naturally leads us to consider the broader class of quotients of Polish modules by sufficiently definable submodules.
Polish modules
Let πΊ be a Polish group, and let (π»π₯)π₯βR be an analytic, independently generating family of subgroups of πΊ. Let π be a Polish ring with no non-trivial compact subgroups, and let π be a locally compact Polish π module. Letππ: π N β π πdenote the projection to the firstπ coordinates, and letππ=ker(ππβ¦ π), which is a closed submodule ofπ.
Thus, ππ is countable, so if we choose the root (ππ)π <πinπ of standard base from π π, then π is generated by ππβͺ (ππ)π <π and thus countably generated. We know nothing about the preorderβπ bounded on locally compact modules, including the existence of a minimum or maximum element.
Proper normed rings
In general, we do not know whether every locally compact field ring admits a compatible regular norm. If π is a finite regular normed ring, then β1(π ) = π N, which in particular is homeomorphic to the Cantor space. To show this, we will use the characterization according to Dijkstra and van Mill [DM09, Theorem 1.1].
Then π is homeomorphic to complete ErdΕ's space if there is a zero-dimensional metrizable topologyπonπ coarser than the original topology such that every point in π has a neighborhood basis (of the original topology) consisting of closed nowhere dense Polish subspaces of (π, ) . It is enough to show that every closed sphere is a closed nowhere dense Polish subspace of (β1(π ), π).
Special cases
Proof of the main theorems
To satisfy the second condition, when choosing ππ you have to take into account that the set of (ππ)π < π is compact, so the set of Γ. When choosing ππ, keep in mind that the first condition applies to a comeager set of ππ, since π+π π0+π π1 Β· Β· Β· +π ππβ1 is analytic and not open, because π/π would otherwise have a countable dimension. The second condition holds for an open set ofππ, since the set ofπ with |ππ!| β€ π is compact.
When choosing ππ, for the first condition, for fixed π βπΌ and πβ² β π π0+ Β· Β· Β· +π ππβ1, we previously showed that {π π βπ+πβ²} is small, i.e. by quantizing over countably manyπππβ² , the set ππ satisfying the first condition is small. The second condition holds for the open set ππ, since the set π with |ππ!| β€ π is bounded.