Free groups are CSA #
Let u be a nontrivial element of a free group which is not a proper power. Then the cyclic
subgroup zpowers u is a maximal abelian subgroup, and it is malnormal: an element conjugating a
nonidentity power of u back into zpowers u already lies in zpowers u. Since every maximal
abelian subgroup of a nontrivial free group is of this form, every free group is a CSA group, and
commutation is transitive on its nontrivial elements.
The hypothesis that u is not a proper power is needed: for u = a ^ 2 the element a lies
outside zpowers u but commutes with u. Both malnormality and maximality characterize the
non-proper-powers among the nontrivial elements.
The argument avoids reduced words. If g * u ^ m * g⁻¹ = u ^ n with u ^ m ≠ 1, commutation
transitivity through u ^ n shows that g * u * g⁻¹ commutes with u, so it lies in
zpowers u, the centralizer of u. Since g * u * g⁻¹ is not a proper power, it is u or
u⁻¹. A nontrivial element of a free group is not conjugate to its inverse
(Group.IsCommTransitive.not_isConj_inv), so g commutes with u and lies in zpowers u.
The file also records two groups built from free groups that are not CSA. First,
FreeGroup (Fin 2) × Multiplicative ℤ: the free group on two generators is nonabelian, but its
product with a nontrivial abelian group has a nonidentity central element. Second, the HNN
extension ⟨F, t | [a ^ 2, t] = 1⟩ of a free group F along the identity of zpowers (a ^ 2)
for a basis element a: there t commutes with a ^ 2 but not with a, although a commutes
with a ^ 2. This is why an extension of the centralizer of u must use the full centralizer of
u, here zpowers a, rather than zpowers u.
Main results #
IsFreeGroup.isCommTransitive: commutation is transitive on the nontrivial elements of a free group.IsFreeGroup.isTISubgroup_zpowers: the cyclic subgroup generated by a non-proper-power is malnormal.IsFreeGroup.maximal_isMulCommutative_zpowers: the cyclic subgroup generated by a non-proper-power is maximal abelian.IsFreeGroup.maximal_isMulCommutative_iff: the maximal abelian subgroups of a nontrivial free group are exactly the cyclic subgroups generated by non-proper-powers.IsFreeGroup.isCSA: every free group is CSA.Group.not_isCSA_prod_int:FreeGroup (Fin 2) × Multiplicative ℤis not CSA.Group.not_isCSA_hnnExtension_zpowers_sq: the HNN extension of a free group along the identity ofzpowers (a ^ 2)is not CSA.
References #
- G. Baumslag, A. Myasnikov, V. Remeslennikov, Algebraic geometry over groups I. Algebraic sets and ideal theory, J. Algebra 219 (1999), §1 and §2.1 (CSA groups; free groups are CSA).
- O. Kharlampovich, A. Myasnikov, V. Remeslennikov, D. Serbin, Subgroups of fully residually free groups: algorithmic problems, Contemp. Math. 360 (2004), §4 (malnormal subgroups) and §5, proof of Lemma 21 (centralizers in free groups).
Commutation is transitive on the nontrivial elements of a free group.
Maximal cyclic subgroups of a free group are malnormal. If u is not a proper power,
then an element outside zpowers u conjugates no nonidentity element of zpowers u back into
zpowers u.
In a free group, the cyclic subgroup generated by a non-proper-power is a maximal abelian subgroup.
A nontrivial element u of a free group generates a malnormal subgroup exactly when u is
not a proper power.
A nontrivial element u of a free group generates a maximal abelian subgroup exactly when
u is not a proper power.
The maximal abelian subgroups of a nontrivial free group are exactly the cyclic subgroups generated by elements that are not proper powers.
Free groups are CSA: every maximal abelian subgroup of a free group is malnormal.
The product of the free group on two generators with ℤ is not CSA.
In the HNN extension of a free group along the identity of zpowers (a ^ 2), the image of
the basis element a does not commute with the stable letter, although a ^ 2 does.
The HNN extension of a free group along the identity of zpowers (a ^ 2) is not commutative
transitive: a commutes with a ^ 2, which commutes with the stable letter, but a does not.
The HNN extension of a free group along the identity of zpowers (a ^ 2) is not CSA. This
is the extension of the centralizer of a ^ 2 taken with zpowers (a ^ 2) in place of the full
centralizer zpowers a.