9
$\begingroup$

I am interested in calculating of $H_2(G,\mathbb{Z})$, where $G$ is a group given by presentation with two relators $\langle a,b| r_1 = r_2 = 1\rangle$.

Moreover, I am interested in such presentations, that $H_2(G,\mathbb{Z}) = \mathbb{Z}^2$, it would be great to learn how to construct a lot of such examples.

$\endgroup$
7
  • 7
    $\begingroup$ If $F$ is the free group with 2 generators, and $R$ the normal subgroup generated by your relators, the Hochschild-Serre spectral sequence provides an isomorphism $H_2(G,\Bbb{Z})\cong R\cap (F,F)/(F,R)$. A more explicit computation is usually quite difficult, and depends very much on the form of the relators. $\endgroup$
    – abx
    Oct 19, 2015 at 7:12
  • $\begingroup$ It is not hard to compute a presentation of a covering group of $G$, which is a central extension of $H_2(G,{\mathbb Z})$ by $G$, but it is not always easy to use that to get a presentation of the central subgroup $H_2(G,{\mathbb Z})$. $\endgroup$
    – Derek Holt
    Oct 19, 2015 at 8:38
  • $\begingroup$ @abx: This formular is known as Hopf's Theorem. $\endgroup$ Oct 19, 2015 at 16:55
  • 1
    $\begingroup$ Of course I know about Hopf's formula. The question is how Hochschild-Serre spectral sequence works, how to define the nature of arrows. $\endgroup$
    – Samarkand
    Oct 19, 2015 at 18:50
  • $\begingroup$ For example, how to compute $H_2$ of $G=\langle a,b| [[a,b],a], a^2b^2 \rangle$? $\endgroup$
    – Samarkand
    Oct 19, 2015 at 18:51

2 Answers 2

5
$\begingroup$

If $G$ is a perfect group (i.e. $G=[G,G]$, i.e. $G_{ab}=0$) which admits a presentation with two generators and two relations, then $H_2G=0$.

More generally, if $G$ is a group which admits a presentation with $n$ generators and $m$ relations then $H_2G$ can be generated by $m-n+\text{rk}_\mathbb{Z}(G_{ab})$ elements. This appears as Exercise II.5.5(b) in Ken Brown's group cohomology book. Here is my old solution:

With the presentation $G=\langle s_1,\cdots,s_n\,|\,r_1,\cdots,r_m\rangle$ we associate the $2$-complex $\;Y=(\bigvee_s S^1)\cup_{r_1}e^2\cdots\cup_{r_m}e^2\;$ so that $\pi_1Y\cong G$. By computing the Euler characteristic $\chi(Y)$ two different ways we obtain the equation $\sum(-1)^i\text{rk}_\mathbb{Z}(H_iY)=\sum(-1)^ic_i$, where $c_i$ is the number of $i$-cells. Then $1-\text{rk}_\mathbb{Z}(G_{ab})+\text{rk}_\mathbb{Z}(H_2Y)=1-n+m$ and so $\text{rk}_\mathbb{Z}(H_2Y)=m-n+r$, where $r=\text{rk}_\mathbb{Z}(G_{ab})=\dim_\mathbb{Q}(\mathbb{Q}\otimes G_{ab})$. Now $H_2Y=\text{ker}(\partial_2)$ is a free abelian group (subgroup of cellular 2-chain group), and by applying Theorem II.5.2[Brown] we get a surjection $H_2Y\rightarrow H_2G$ (from the exact sequence in that theorem -- this theorem is a play on the Hurewicz theorem). Thus $H_2G$ $\textit{can}$ be generated by $m-n+r$ elements. Now in my special scenario, $m-n+r=2-2+0=0$.

$\endgroup$
2
  • $\begingroup$ And in that same chapter of the book, you'll find more results for one relator groups, if that interests you. $\endgroup$ Oct 23, 2015 at 21:42
  • 1
    $\begingroup$ Thanks Chris, but this is more or less trivial results. These can be used to understand something about $H_2$, but I am interested in concrete calculations, not generalities. $\endgroup$
    – Samarkand
    Oct 25, 2015 at 0:49
2
$\begingroup$

First, I am very skeptical that one can, in general, calculate $H_2(G, \mathbb{Z})$ using just a presentation $\langle a,b \, \vert \, r_1=r_2 =1 \rangle$. It is generally known that finite presentations for groups are inadequate to solve many kinds of decision problems. You will probably need to know more about $G$ to calculate $H_2(G, \mathbb{Z})$ using either the Hopf formula or spectral sequences.

Second, if you do know of one example of a group $G$ which has a presentation $\langle a,b \, \vert \, r_1=r_2 =1 \rangle$ and such that $H_2(G, \mathbb{Z})$, is $\mathbb{Z}\times \mathbb{Z}$, then it is not difficult to construct lots of other presentations for $G$. Let $F$ be the free group on $\{a,b\}$. You can use a sequence of regular
elementary Nielsen transformations (See Lyndon and Schupp, Chapter I, Section 2) to find another pair of words, $\{A,B\}$ that also freely generate $F$. Write $R_i, i=1,2$ for the word obtained from $r_i$ by replacing every $a$ by $A$ and every $b$ by $B$ (and doing the same replacement for inverses, of course). Then $\langle A,B \, \vert \, R_1=R_2 =1 \rangle$ is also a presentation for $G$. Then $a$ can be written as a word on $A$ and $B$ and $b$ can be written as a word on $A$ and $B$, so we can write $r_1$ and $r_2$ as words on $A$ and $B$ and obtain a presentation $\langle A,B \, \vert \, r_1=r_2 =1 \rangle$ for $G$. Dually, we can rewrite $R_1$ and $R_2$ as words on $a$ and $b$ and obtain a presentation $\langle a,b \, \vert \, R_1=R_2 =1 \rangle$. Similarly, we could obtain a new presentation for $G$ by swapping $a$ with $b$ in $r_1$ and $r_2$. Also we can replace $r_1$ and $r_2$ by conjugates of these.

Finally, I will mention one example which many might consider to be obvious. Let $G$ be the group on 2 generators which is free in the variety of groups having nilpotence class 2. Then $G$ has numerous presentations of the form $\langle a,b \, \vert \, r_1=r_2 =1 \rangle$ and the Schur multiplier, $H_2(G, \mathbb{Z})$, is $\mathbb{Z}\times \mathbb{Z}$.

Write $\{\gamma_k(H)\}$ for the lower central series of any group $H$: i,e. $\gamma_1(H) = H$ and $\gamma_{k+1}(H) = [\gamma_k(H),H]$. Let $F$ be the free group on two generators, $a$ and $b$. Let $G$, as above, be the group on 2 generators which is free in the variety of groups having nilpotence class 2. Then $G = F/\gamma_3(F)$ and one presentation for $G$ is $\langle a,b \, \vert \, [b,a,a]=[b,a,b]=1\rangle$. Here, the smallest normal subgroup, $R$, containing $[b,a,a]$ and $[b,a,b]$ is $\gamma_3(F)$. It is not too difficult to see that $F/\gamma_4(F)$ is the covering group for $G$ and that $\gamma_3(F)/\gamma_4(F)$ is the Schur multiplier. It is also easy to verify that $\gamma_3(F)/\gamma_4(F)$ is free abelian with generators $[b,a,a]\gamma_4(F)$ and $[b,a,b]\gamma_4(F)$.

Later (November 16, 2015): For another example for $G$, let $G$ be the group having presentation $\langle a,b \, \vert \, [b,a,a] = [b,a,b,b]=1 \rangle$.

$\endgroup$
2
  • 1
    $\begingroup$ It's true H_2 is in general not computable from a presentation. This is s theorem of Cameron Gordon. Of course, the examples that show this have many more than 2 relations. The reference is: C. McA. Gordon. Some embedding theorems and undecidability questions for groups. In Combinatorial and geometric group theory (Edinburgh, 1993), vol- ume 204 of London Math. Soc. Lecture Note Ser., pages 105–110. $\endgroup$
    – HJRW
    Oct 24, 2015 at 6:16
  • $\begingroup$ Dear David, I know this. The question is how to construct presentations of different groups having $H_2 = Z^2$ $\endgroup$
    – Samarkand
    Nov 15, 2015 at 19:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.