2

I've been working through D.L. Johnsons book Symmetries and am currently in the chapter about triangle groups, specifically the part about triangles in the real plane, where there are only three cases. My question now concerns the algebraic properties of these groups. In the exercises it is stated, that two of them should be isomorphic to wallpaper groups and I have the feeling the third one should have a corresponding wallpaper group as well. The isomorphism is supposed to be shown through Tietze transformations, but for the life of me I can not find the right steps. Could someone please point me in the right direction? I am specifically interested in an algebraic/group theoretic solution, not a geometric proof. Below I will state the relevant groups: First the general triangle group $\Delta(l,m,n)=\langle x,y,z\ \vert\ x^2=y^2=z^2=(xy)^l=(yz)^m=(zx)^n\rangle$.

Now these are the isomorphisms I can not produce:

  • $\Delta(2,3,6)\cong G_6^1=\langle a,b,r,s\vert\ ab=ba,s^6=1,s^{-1}as=b,s^{-1}bs=a^{-1}b,r^2=1,r^{-1}ar=b,r^{-1}br=a,(sr)^2=1\rangle$
  • $\Delta(2,4,4)\cong G_4^1=\langle a,b,r,s\vert \ ab=ba,s^4=1,s^{-1}as=b,s^{-1}bs=a^{-1},r^2=1,r^{-1}ar=a,r^{-1}br=b^{-1},(sr)^2=1\rangle$
  • $\Delta(3,3,3)\cong ??$ Here I am still clueless if it is even true.
Ferdi
  • 59
  • 2
    One observation: $\Delta(3,3,3)$ is a subgroup of $\Delta(2,3,6)$, since a triangle with angles 30, 60, 90 degrees is half of a 60, 60, 60 triangle. – Viktor Vaughn Jun 29 '25 at 01:28
  • 3
    Regarding your title question, it is unclear from your post whether you care about it or not, but if so then you should clarify a bit and tell us what is your definition of wallpaper groups. The definition that I know is (equivalent to) a group of rigid motions of the plane which has a finite polygon for a fundamental domain, in which case the answer to your title question is a(n almost) trivial "yes": the triangle used to generate a triangle group is a fundamental domain. But the triviality of that answer makes me think that you have a different definition in mind. – Lee Mosher Jun 29 '25 at 13:45
  • 1
    Beyond the issue of the title question, it is also unclear what kind of answer you are fishing for. Are you working from some particular list of presentations which is unknown to us, from which you copied the presentation for $G^1_6$ and $G^1_4$? Are you asking whether the given presentation for $\Delta(3,3,3)$ is Tietze equivalent to some presentation in that unknown list? – Lee Mosher Jun 29 '25 at 13:52
  • Sorry, if my question was unclear, I am indeed working with a given set of permutations given in the book I mentioned. I will try to clarify my question, the book tells me, that the OP part of $Delta(3,3,3)$ is isomorphic to I believe p3, but the group itself is not isomorphic to p3m1 and p31m, is that correct and if yes, which is the isomorpic wallpaper group? – Ferdi Jun 29 '25 at 19:21
  • 1
    Let me suggest that you use the edit button to clarify your post regarding $\Delta(3,3,3)$ and what you are denoting $p3$, $p3m1$, and $p31m$. – Lee Mosher Jun 29 '25 at 21:04

1 Answers1

2

Doing this with Tietze transformations has to be the most painful way to exhibit these isomorphisms, but that's not going to stop me!

I know you said you are not interested in a geometric argument, but the below manipulations are extremely hard to find if you aren't thinking geometrically. It can be presented purely algebraically, but then it's not too different from a magic trick. I'll try to give the geometric intuition also.


I will show that $$\Delta(2,4,4) := \langle x,y,z \; \mid \; x^2 = y^2 = z^2 = 1 = (xy)^4=(yz)^4 = (xz)^2\rangle$$ is isomorphic to $$\langle a,b, r,s \; \mid \; ab = ba, s^4 = 1, s^{-1}as = b, s^{-1}bs = a^{-1}, r^2 = 1, r^{-1}ar = a, r^{-1}br = b^{-1}, (sr^2 = 1) \rangle $$ by finding Titze transformations which take one presentation to the other. I won't do $\Delta(2,3,6)$ because this is long enough already, but you should be able to use the same techniques.

First, we should notice that $\langle a, b \rangle \cong \mathbb{Z}^2$ represent the group of translations in the wallpaper group---we will think of $a$ as a translation along the $x$ axis and $b$ as a translation along the $y$-axis. Since $r^2 = 1$, we guess that $r$ is a reflection. Since it conjugates $b$ to $b^{-1}$ and commutes with $a$ we furthermore guess that it is the reflection through the $x$-axis. Likewise $s$ has order $4$, so we guess it is a rotation; it conjugates $a$ to $b$ and $b$ to $-a = a^{-1}$, so we guess it is a rotation of $\pi/2$ counterclockwise.

With these guesses in mind, we should imagine a $45$-$45$-$90$ triangle with one of its acute angles at the origin, and leg with length $1$. Then $r$ is the reflection through one of the sides, and is therefore should be $x = r$ in the presentation of $\Delta(2,4,4)$. We need to find the other sides. Let $y$ be the reflection through the hypotenuse. Then a quick geometric argument shows that $yx = s$, and hence $y = sx = sr$. Our first Tietze transformations will be introducting the generators $x, y $ along with the relations $x = r$, $y = sr$, then using those relations to eliminate $s$ and $r$ from the presentation. We now have: $$\langle a,b,x,y \; \mid \; x^2 = y^2 = 1, ab = ba, (yx)^4 = 1, xyayx = b, a^{-1} = xybyx, xax = a, xbx = b^{-1}\rangle$$

(Note that I suppressed all inverses on $x$ and $y$ since they have order 2)

Next we need to introduce the reflection, $z$, through the edge not incident to the origin. Here's a somewhat non-obvious geometric argument to see that $a = zyxy$. The translation $a$ is a composition of the reflection $z$ and the reflection through the y-axis (and the translation distance is twice the lenght of the leg). The reflection through the y-axis is just the reflection through the x-axis conjugated by a 45 degree reflection, ie $yzy$. Then $a = zyxy$.

The previous paragraph is unnecessary motivation for our Tietze transformation argument. The next step will be to introduce the new generator $z$ with the relation $z = ayxy$, or equivalently, $a = zyxy$. Using this new relation, we can eliminate $a$ from the presentation: $$\langle b,x,y,z \; \mid \; \text{see below} \rangle$$

with relations

  1. $x^2 = y^2 = 1$
  2. $zyxyb = bzyxy$
  3. $(yx)^4 = 1$
  4. $b = xyzyxyyx = xyzy$
  5. $yxyz^{-1} = xybyx$
  6. $x(zyxy)x = zyxy$
  7. $xbx = b^{-1}$.

Notice that we have the relation $b = xyzy$, and can therefore eliminate it from our presentation by substituting $b = xyzy$ in every relation where $b$ appears.

Relation 5 becomes $yxyz^{-1} = xyxyzyyx = xyxyzx$. Now move the $x$'s on the right to the left to get $x(yxyz^{-1})x = yxyz$. The left hand side of this is simply the inverse of the left hand side of relation 6, so we conclude $yxyz^{-1} = yxyz$, and therefore $z^{-1} = z$, ie $z^2 = 1.

Relation 6 becomes $xzyxyx = zyxy$. Shuffling around gives $xz = zyxyxyxy = z(yx)^4x = zx$, so relation 6 becomes $xz = zx$, and since $z^2 = 1$, equivalently $(xz)^2 = 1$.

Relation 2 becomes $[(zyxy)(x)]yzy = xyzyzyxy$. Cancel the right $y$'s, and use relation 6 on the bracket. Then $xzyxy yz = xyzyzyx$. Canceling some more gives $zyxz = yzyzyx$. Since $[x,z] = 1$, commute the $x$ and $z$ on the left and cancel the $x$ to give $zyz = yzyzy$, or equivalently $(zy)^4 = 1$.

Relation 7 becomes $x xyzy x = yzyx$, and canceling the $x$'s gives $yzy = yzy$, a tautology.

Then through the above Titeze transformations, we have eliminated relations 4 and 7, while replacing relations 2, 5 and 6 with $(zy)^4 = 1$, $z^2 = 1$ and $(xz)^2 = 1$ respectively. Putting it all together, the final presentation is $$\langle x,y,z \; \mid \; x^2 = y^2 = z^2 = 1 = (xz)^2 = (yx)^4 = (zy)^4 \rangle$$ as required.


To answer the question in the title: Yes, all Euclidean triangle groups are wallpaper groups. As Lee Mosher said in the comments, this statement is somewhat trial with some of the standard definitions of triangle groups/wallpaper group, so it's not entirely clear what kind of answer you want.

The triangle group $\Delta(3,3,3)$ corresponds to the wallpaper group of the hexagonal tiling.

  • Sorry if I was unclear, I will try to clarify my question, the book tells me, that the OP part of $Delta(3,3,3)$ is isomorphic to I believe p3, but the group itself is not isomorphic to p3m1 and p31m, is that correct and if yes, which is the isomorpic wallpaper group? – Ferdi Jun 29 '25 at 19:23
  • Also thank you very much for the Tietze transformations, very insightful. – Ferdi Jun 29 '25 at 19:25
  • $\Delta(3,3,3)$ should be p3m1 (from glancing at the wikipedia page). – Noah Caplinger Jun 29 '25 at 20:26