The Gilbert-Howie group is defined by a presentation
with addition understood modulo A recent publication1 notes it is unknown whether is hyperbolic.
In this note we will demonstrate that it is isomorphic to the cyclic group
Define the group to be a semidirect product with the presentation
Using Tietze moves, we can simplify this presentation substantially.
First, eliminate since it can be expressed as Similarly, and so forth. We continue this procedure until our remaining generators are and the relations are
and
Now, define the free group automorphism as It is easy to check that all of the -conjugation relations are consequences of this definition, thus making most of these relation redundant.
Also note that the relation
is equivalent to which is redundant, given that Thus, we keep only the following -conjugation relations
Now, we can cancel the redundant generators in the three relations:
Observe that hence this relator is redundant. Observe also that so this relator is redundant as well.
Finally, note that
and hence
But
and thus
This proves to be redundant.
Finally, and can be eliminated, leaving us with just
Let Using this relation, we eliminate The resulting relator is
This leaves us a short presentation:
Observe that can be put into the form Thus, is equivalent to This leaves us with our final short presentation.
Now, Tomas Rokicki demonstrated, in a MathOverflow answer 2 , that the group with the presentation
is finite and has exactly elements. Clearly, this group is identical to
We observe that and so Thus, as claimed.
Notes
More on the group
We can easily verify that the element group has the small groups ID generated by, for example,
a = (2,11,27)(3,21,16)(4,31,5)(6,14,20)(7,24,9)(8,34,35)(10,17,13)(12,37,28)(15,30,32)(18,23,36)(19,33,25)(22,26,29)
b = (1,34,23,2,9,19,28,25,26)(3,21,15,17,4,33,11,6,20)(5,8,7,32,36,10,31,24,14)(12,18,16,29,37,22,27,13,30)
From these, we can see that the generator shifting action of on is faithful. GroupNames page provides a nice overview of this group.
Software
Rokicki’s proof uses group theory software (kbmag, MAF) to find confluent and terminating rewrite system for this group. I was able to replicate this computation. This presentation (taken as a monoid presentation) was also discovered by Slava Pestov3. The page dedicated to this monoid includes a complete rewriting system and a certificate deriving the rewriting rules directly from the monoid presentation. As such, it constitutes an independent proof that this group is indeed finite, being an enveloping group of a finite monoid.
Attribution
All of the difficult work was done by Stefan Kohl 4 who asked about finiteness of this presentation of and Tomas Rokicki4 who proved finiteness of I am also deeply grateful to Slava Pestov, whose monoid word problem project led me to the discovery of Gilbert-Howie groups embedding into semidirect product groups that can have exceptionally short presentations. Analyzing multiple exceptional cases encountered in Pestov’s work, I was able to identify the pattern and discover the hidden significance of these peresentations. I am also thankful for him bringing the MathOverflow answer4 to my attention, and his indepependent verification of the finiteness of the monoid cousin of
Footnotes
-
Chinyere, Ihechukwu and Williams, Gerald (2021) Hyperbolic groups of Fibonacci type and T(5) cyclically presented groups. Journal of Algebra, 580. pp. 104-126. DOI https://doi.org/10.1016/j.jalgebra.2021.04.003 ↩
-
Tomas Rokicki (https://mathoverflow.net/users/536277/tomas-rokicki), Catalogue of groups with short finite presentations, URL (version: 2024-08-28): https://mathoverflow.net/q/477760 ↩
-
Pestov, S. (2026). #22271 ⟨a, b | aaa=1, ababbb=ba⟩. Slava’s Monoid Zoo. https://monoids.net/2,2/22271.html ↩
-
Stefan Kohl (https://mathoverflow.net/users/28104/stefan-kohl), Catalogue of groups with short finite presentations, URL (version: 2022-06-30): https://mathoverflow.net/q/423541 ↩ ↩2 ↩3