P. Hall proved that every finitely generated torsion-free nilpotent group can be faithfully represented by upper unitriangular matrices over $\mathbb Z$. The most famous example is the integral Heisenberg group, which is free nilpotent, of rank $2$ and class $2$, and is isomorphic to $UT(3,\mathbb{Z})$.
I would like to know what is the easiest such representation for the free nilpotent group $F(3,2)$, of rank $3$ and class $2$. Its group presentation is $$F(3,2)=\langle x,y,z \,\|\, [x,[x,y]],[y,[x,y]],[z,[x,y]],[x,[y,z]],[y,[y,z]],[z,[y,z]],[x,[x,z]],[y,[x,z]],[z,[x,z]] \rangle.$$
An algorithm to compute this representation was implemented in GAP, based on the work of de Graaf and Nickel [de Graaf, Willem A.; Nickel, Werner, Constructing faithful representations of finitely-generated torsion-free nilpotent groups, J. Symb. Comput. 33, No. 1, 31-41 (2002). ZBL1021.20006.] Unfortunately I am not proficient with GAP to run it myself.
In the above paper the authors state that the algorithm produces a representation for $F(3,2)$ in $UT(6,\mathbb{Z})$.
Is there a faithful representation of $F(3,2)$, over integer matrices, of degree less than $6$?
I think that I have found an embedding of $F(3,2)$ into ${\rm UT}(5,{\mathbb Z})$. I used a combination of GAP and Magma to do this, but here is the (putative) result in GAP.
The images of the three generators of $F(3,2)$ are
If you calculate the three commutators $[a,b]$, $[a,c]$ and $[b,c]$, you will find that they clearly generate a subgroup isomorphic to ${\mathbb Z}^3$, and they are all centralized by $a$,$b$, and $c$, which I think is enough to show that the group they generate is isomorphic to $F(3,2)$.