Consider the group $G$ given by the following presentation:
$$G=\langle x,y\mid x^{-1}y^2xy^2=x^{-2}yx^{-2}y^3=1\rangle.$$
In this slides it is noted that this is a torsion-free polycyclic group, which is virtually the discrete Heisenberg group, see p.16.
I am interested in the last piece of information. According to the statement, one should be able to find a copy of the Heisenberg group $H$ in $G$, such that $G/H$ is finite.
One possible presentation of the Heisenberg group is given by $$H=\langle a,b\mid [a,[a,b]]=[b,[a,b]]=1\rangle.$$
How to find such a copy of $H$ in $G$ with finite quotient? Can this be done by hand or are the computer-base algorithms that find such copies? I am not an expert in group theory, so maybe this is obvious?
In the slides you link, it is mentioned this example comes from Lindsay Jennae Soelberg's Master's thesis. She mentions "[using] Magma as a tool to show that $G_Z$ is torsion-free [following] an idea of Derek Holt, given in response to a question at MathOverflow", which I suspect may refer to this answer.
To quote the thesis (p24):
From this presentation, it becomes clear that $K := \langle h_1,h_3\rangle$ is isomorphic to the Heisenberg group, has index $8$ in $H$ and index $64$ in $G_Z$. Do note, however, that while $K$ is normal in $H$, it is not normal in $G$.