Free Boolean Topological Group Overview

• Free Boolean topological groups are defined as the free objects in Boolean groups where every element has order 2, viewed as finite subsets of X under symmetric difference.
• They are constructed using Graev and Markov approaches, leveraging continuous pseudometrics and neighborhood bases derived from uniformities and covers.
• Their completeness properties link group topology with Dieudonné completeness of X, extending to variants like non-archimedean, precompact, and extremally disconnected cases.

A free Boolean topological group is the free object generated by a space $X$ in the variety of Boolean topological groups. Since a Boolean group is a group in which every element has order $2$, it is automatically Abelian and can be viewed algebraically as a vector space over $\mathbb F_2$; accordingly, the abstract group underlying $B(X)$ is the free $\mathbb F_2$-vector space on $X$, equivalently the group of finite subsets of $X$ under symmetric difference. In the literature this construction is developed for Tychonoff spaces and, in Sipacheva’s survey formulation, for completely regular Hausdorff spaces. The subject combines universal topological algebra, explicit pseudometric constructions, uniform completeness, zero-dimensional topology, and set-theoretic methods based on filters and ultrafilters on $\omega$ (Sipacheva, 2016, Sipacheva et al., 16 Jul 2025).

1. Algebraic model and universal construction

A Boolean group satisfies $x+x=0$ in additive notation, so every element is self-inverse and the group is Abelian. Abstractly, every Boolean group is a free $\mathbb F_2$-vector space. This makes the canonical algebraic model especially concrete: the free Boolean group on a set $2$0 can be identified with $2$1, the family of finite subsets of $2$2, with group operation given by symmetric difference. In reduced form, an element is written as

$2$3

and repeated letters cancel because $2$4 (Sipacheva et al., 16 Jul 2025, Sipacheva, 2016).

For a nonempty set $2$5 with distinguished point $2$6, the Graev version $2$7 is the abstract free Boolean group whose basis is $2$8, with the neutral element identified with $2$9. If $\mathbb F_2$0 is a Tychonoff space with distinguished point $\mathbb F_2$1, the free Boolean topological group $\mathbb F_2$2 is required to be a topological Boolean group containing $\mathbb F_2$3 as a subspace, generated by $\mathbb F_2$4, and satisfying the universal property that every continuous map $\mathbb F_2$5 into a Boolean topological group $\mathbb F_2$6 with $\mathbb F_2$7 extends to a continuous homomorphism

$\mathbb F_2$8

Up to topological isomorphism, this construction does not depend on the choice of distinguished point (Sipacheva et al., 16 Jul 2025).

The pointed Graev construction and the unpointed Markov construction reduce to one another. If $\mathbb F_2$9 denotes the Markov free Boolean topological group and $B(X)$0 is obtained by adjoining a clopen point, then

$B(X)$1

topologically. The Boolean functor also sits naturally between the standard free and free Abelian constructions: $B(X)$2 and, in the formulation used in later work, $B(X)$3 is the topological quotient of $B(X)$4 by the subgroup $B(X)$5 of squares. These quotient maps are continuous and open, so quotient-stable properties of $B(X)$6 and $B(X)$7 often descend to $B(X)$8 (Sipacheva, 2016, Sipacheva, 2022).

2. Topology, pseudometrics, and finite-word layers

The topology of $B(X)$9 admits several explicit descriptions. Sipacheva records neighborhood bases at zero obtained from sequences of entourages in the universal uniformity, from sequences of open or normal covers, and from continuous pseudometrics on $\mathbb F_2$0. In pseudometric form, if $\mathbb F_2$1 is a continuous pseudometric on $\mathbb F_2$2, then

$\mathbb F_2$3

is a neighborhood of zero, and sets of this form constitute a neighborhood base. In the cover description, one defines $\mathbb F_2$4 and then

$\mathbb F_2$5

which again yields a neighborhood base (Sipacheva, 2016).

The 2025 completeness paper develops a Boolean-specific Graev extension of pseudometrics. For a pseudometric $\mathbb F_2$6 on $\mathbb F_2$7 and $\mathbb F_2$8, one considers all representations

$\mathbb F_2$9

and defines

$X$0

followed by

$X$1

The resulting $X$2 is an invariant pseudometric on $X$3 extending $X$4. A central combinatorial lemma shows that the infimum is actually a minimum realized by a reduced representation with pairwise distinct points, except for the possible appearance of the distinguished zero when the support has odd cardinality. The same paper proves that $X$5 is the maximal invariant pseudometric on $X$6 extending $X$7, and that the sets

$X$8

for continuous pseudometrics $X$9 on $X$0 form a local base at the identity (Sipacheva et al., 16 Jul 2025).

Finite-word layers are built into the structure. Writing

$X$1

one has

$X$2

These sets are always closed in $X$3, and $X$4 itself is closed. The survey emphasizes that the inductive-limit problem for the filtration by $X$5 parallels the corresponding questions for $X$6 and $X$7, but in filter-generated examples the Boolean situation becomes especially explicit (Sipacheva, 2016, Sipacheva et al., 16 Jul 2025).

3. Completeness and the Dieudonné criterion

A major recent advance is the exact completeness theorem for free Boolean topological groups. For a Tychonoff space $X$8,

$X$9

Because Boolean groups are Abelian, the left and two-sided group uniformities coincide, so in this setting Weil completeness and Raĭkov completeness are the same. The theorem therefore gives the equivalent characterization

$\omega$0

This answers positively a question posed by O. Sipacheva in 2015 and places the Boolean case alongside the already known Abelian result for $\omega$1 (Sipacheva et al., 16 Jul 2025).

The conceptual core of the “only if” direction is the identification of the uniformity induced on the embedded copy of $\omega$2 by the group uniformity of $\omega$3. If $\omega$4 denotes the restriction of the group uniformity and $\omega$5 the universal uniformity of $\omega$6, then

$\omega$7

Consequently, if $\omega$8 is complete, then the closed subspace $\omega$9 is complete in the restricted uniformity, hence complete in its universal uniformity, and therefore Dieudonné complete (Sipacheva et al., 16 Jul 2025).

The converse direction is substantially deeper. The proof uses the closed finite levels $x+x=0$0, proves that each $x+x=0$1 is complete whenever $x+x=0$2 is Dieudonné complete, and then employs an additional neighborhood construction built from sequences of continuous pseudometrics $x+x=0$3. For

$x+x=0$4

one gets open neighborhoods $x+x=0$5 suited to a final contradiction argument with a nonconvergent Cauchy filter. The proof is Boolean-specific: it relies on the combinatorics of supports and on the particular Graev extension used for exponent-$x+x=0$6 groups (Sipacheva et al., 16 Jul 2025).

4. Filter spaces, linear Boolean groups, and forcing topologies

A central class of examples is provided by almost discrete spaces associated with filters. For a free filter $x+x=0$7 on $x+x=0$8, let

$x+x=0$9

where points of $\mathbb F_2$0 are isolated and neighborhoods of $\mathbb F_2$1 are of the form $\mathbb F_2$2 with $\mathbb F_2$3. Then

$\mathbb F_2$4

with group operation given by symmetric difference. In these models, free Boolean topological groups become topologies on finite subsets of $\mathbb F_2$5 determined by the combinatorics of $\mathbb F_2$6 (Sipacheva, 2016).

The survey identifies the free Boolean linear topology $\mathbb F_2$7 with the Mathias topology $\mathbb F_2$8, and places the ordinary free Boolean topology between Mathias and Laver: $\mathbb F_2$9 This yields a precise interface between free Boolean groups and forcing notions. One of the sharpest results in the area is the Thümmel theorem: for a filter $2$00 on $2$01, selectivity is equivalent to the coincidence

$2$02

and also equivalent to $2$03 being a group topology together with a further explicit openness condition formulated in terms of sequences $2$04 (Sipacheva, 2016).

These filter-generated models also control inductive-limit behavior and extremal disconnectedness. The free Boolean group $2$05 has the inductive-limit topology with respect to its finite levels $2$06 if and only if $2$07 is a $2$08-filter. Moreover, if $2$09 is selective, then $2$10, hence $2$11, is extremally disconnected; conversely, if $2$12 or $2$13 is extremally disconnected, then $2$14 is a Ramsey ultrafilter. The same survey shows that every countable Boolean topological group has either a discrete closed basis or a discrete basis with $2$15 as unique limit point, and in the latter case it is a continuous isomorphic image of some $2$16. Thus filter spaces serve as canonical countable test objects for Boolean topological group topology (Sipacheva, 2016).

5. Strong disconnectedness and set-theoretic dependence

The study of strong disconnectedness properties reveals a sharp contrast between free Boolean groups and the free or free Abelian constructions. For any space $2$17,

$2$18

Hence the Boolean free construction preserves and reflects the $2$19-space property. However, the behavior of $2$20-spaces is exceptional. If $2$21 is an $2$22-group, then $2$23 contains at most one non-$2$24-point. If $2$25 is not a $2$26-space and $2$27 is an $2$28-group, then there exists a countable space $2$29 with a unique nonisolated point such that $2$30 is an $2$31-quotient image of $2$32 and $2$33 is an extremally disconnected quotient of $2$34 (Sipacheva, 2022).

This leads to the principal set-theoretic equivalence in the disconnectedness theory: $2$35 The existence and nonexistence of selective ultrafilters are both consistent with ZFC, so the existence of non-$2$36 free Boolean $2$37-groups is independent of ZFC in exactly the sense established by this equivalence. The Boolean case is therefore exceptional among the standard free constructions: for free groups $2$38 and free Abelian groups $2$39, being an $2$40-space is equivalent to $2$41 being a $2$42-space, while for $2$43 non-$2$44 $2$45-examples may exist, but only under the selective-ultrafilter hypothesis (Sipacheva, 2022).

Sipacheva’s survey places these results into a broader extremal-disconnectedness picture. It records Malykhin’s theorem that every extremally disconnected group contains an open Boolean subgroup, and proves a general obstruction: if $2$46 or $2$47 is extremally disconnected, then either $2$48 is a $2$49-space or there exists a Ramsey ultrafilter. In particular, the nonexistence of nondiscrete extremally disconnected free Boolean groups is consistent with ZFC in models without measurable cardinals and Ramsey ultrafilters (Sipacheva, 2016).

6. Non-archimedean and precompact variants

A particularly explicit specialized theory arises in the non-archimedean category. For a uniform space $2$50, the free Boolean non-archimedean topological group $2$51 is the free object in the class $2$52 of Hausdorff Boolean non-archimedean groups. When $2$53 is non-archimedean, the universal map is a uniform embedding, the group is algebraically the abstract free Boolean group $2$54, and $2$55 is a closed subspace. If $2$56 is a base of equivalence relations, then the neighborhood base at $2$57 is given by the open subgroups

$2$58

where $2$59 is generated by $2$60, equivalently by $2$61. For ultrametric spaces, the corresponding free Boolean NA group is ultra-normable, and its topology is induced by a Graev-type maximal ultra-norm constructed from finite configurations; for Stone spaces, the free Boolean profinite group $2$62 is identified with the Pontryagin dual $2$63 (Megrelishvili et al., 2013).

This NA theory contrasts sharply with the classical Markov-Graev setting. Sipacheva notes that a nondiscrete free Boolean topological group is never metrizable, whereas the NA theory shows that for metrizable non-archimedean uniformities the free Boolean NA group is also metrizable, indeed ultrametrizable in the ultra-metric case. The difference is categorical: the NA construction works inside a subclass of topological groups with open-subgroup bases and admits a final-topology description directly in terms of equivalence relations (Sipacheva, 2016, Megrelishvili et al., 2013).

Another specialized variant is the free precompact Boolean group $2$64, the free object over a topological space $2$65 in the variety of precompact Boolean groups. For zero-dimensional $2$66, $2$67 is topologically isomorphic to $2$68 equipped with the initial topology induced by all continuous maps $2$69. Over topological sums of elementary spaces—equivalently, disjoint sums of countable maximal spaces—the free precompact Boolean group exhibits strong nonconvergence phenomena: it contains no non-trivial convergent sequences, and in fact no infinite compact subsets (Shakhmatov et al., 2017).

Related work on free nonabelian topological groups clarifies what is not yet known in the Boolean setting. Brazas proved that every open subgroup of a free Graev topological group is again free Graev, and that an open subgroup of a free Markov topological group is free Markov if and only if it is disconnected. However, that machinery does not automatically descend through the Boolean quotient, and no corresponding Boolean subgroup theorem is supplied there (Brazas, 2012).