Given a finite graph $\Gamma$, one has the right-angled Artin group $A(\Gamma )$. Its generators $s_1, \dots s_n$ bijectively correspond to vertices of $\Gamma$ and the relators are $s_is_j=s_js_i$ provided the corresponding vertices are joined by an edge.

Let $A_i(\Gamma)$ be the group obtained from $A(\Gamma)$ by setting $s_i=1$; this corresponds to removing the vertex $i$ from $\Gamma$.

I know very little of these matters but it seems plausible that any nontrivial element of $A(\Gamma)$ projects to a nontrivial element of $A_i(\Gamma)$ for some $i$; is this correct?


No. Let $\Gamma$ be the graph with two vertices and no edges - the non-abelian free group of rank two - and let $g$ be the commutator of the two generators $s_1$ and $s_2$. Then $g$ is certainly non-trivial, but $g$ dies whenever you kill $s_1$ or $s_2$.


For an example with a connected graph, let's take $\Gamma$ to be the straight-line graph with four vertices $a,b,c,d$ (so $[a,b]=[b,c]=[c,d]=1$). Now consider $g=[[c,a],[b,d]]$. Clearly this dies when you kill any generator. On the other hand,


and a well-known solution to the word problem in right-angled Artin groups tells you that $g$ is non-trivial.

  • $\begingroup$ Thanks! I missed this obvious example as I was looking at right-angled Artin groups corresponding to connected graphs. Are there such examples for connected graphs (other than the graph that has only one vertex)? $\endgroup$ – Igor Belegradek Jan 16 '10 at 0:44
  • 1
    $\begingroup$ I've added one. $\endgroup$ – HJRW Jan 16 '10 at 2:02
  • $\begingroup$ Thanks a lot! This neatly kills the application that I had in mind. Too bad. $\endgroup$ – Igor Belegradek Jan 16 '10 at 2:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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