I would like to know why "dilation" on the 3-dimensional Heisenberg group $H^3$ is given by the following operation
$\begin{align} \varphi_{\lambda} : & \qquad H^3 \longrightarrow H^3\\ &(x,y,t)\to \varphi_{\lambda} (x,y,t)= (\lambda x, \lambda y, \lambda^2 t) \end{align}\qquad\mbox{for}\, \lambda\geq0.$
Where the Heisenberg group $H^3$ is the set $\mathbb R^2\times \mathbb R$ endowed with the group law $$(x,x,t)\bullet(x',y',t') =\left (x+x',y+y',t+t'+\frac{1}{2}(x y'-x' y)\right).$$
Thank you in advance
My best guess: it plays well with the group operation on $\mathbb{H}^3$. Notice that there are quadratic terms in the third component for the group operation. The Heisenberg group is kind of a warping of the usual group structure on $\Bbb R^3-$which is just vector addition of course. If you dilate a sum, it should be the sum of the dilations. We would want the same here since we are just warping the usual structure on $\Bbb R^3$, that is
$$\mathcal{D}_{\lambda}((x,y,t)\cdot(x',y',t')) = \mathcal{D}_{\lambda}(x,y,t)\cdot\mathcal{D}_{\lambda}(x',y',t').$$
Written out slightly,
$$\mathcal{D}_{\lambda}\left(x+x',y+y',t+t' + \frac{1}{2}\left(xy'-x'y\right)\right) = \mathcal{D}_{\lambda}(x,y,t)\cdot\mathcal{D}_{\lambda}(x',y',t').$$
Notice the appearance of quadratic terms $xy'$ and $x'y$ in the third component of the group law. If you dilate the $x$, $y$ linearly (which you might as well, or change variables to make the dilation linear), you have to dilate the $t$ quadratically in order for this equation to hold.
In algebraic terms, it's a specific kind of automorphism.