[step:Encode sums of two elements by a larger coordinate box]
Define the original coordinate box
\begin{align*}
B := \{0,1,\dots,L_1-1\} \times \cdots \times \{0,1,\dots,L_d-1\}.
\end{align*}
Define the doubled coordinate box
\begin{align*}
C := \{0,1,\dots,2L_1-2\} \times \cdots \times \{0,1,\dots,2L_d-2\}.
\end{align*}
Define the map
\begin{align*}
\psi: C &\to G \\
(m_1,\dots,m_d) &\mapsto 2x_0 + \sum_{i=1}^d m_i x_i,
\end{align*}
where $2x_0$ denotes $x_0+x_0$.
We claim that $P+P \subseteq \psi(C)$. Let $y \in P+P$. Then there exist $p,q \in P$ such that $y=p+q$. By the definition of $P$, there exist $(n_1,\dots,n_d),(k_1,\dots,k_d) \in B$ such that
\begin{align*}
p &= x_0 + \sum_{i=1}^d n_i x_i, &
q &= x_0 + \sum_{i=1}^d k_i x_i.
\end{align*}
For each $i \in \{1,\dots,d\}$, define $m_i := n_i+k_i$. Since $0 \leq n_i,k_i \leq L_i-1$, we have
\begin{align*}
0 \leq m_i \leq 2L_i-2.
\end{align*}
Hence $(m_1,\dots,m_d) \in C$. Using commutativity in $G$,
\begin{align*}
y
&= p+q \\
&= \left(x_0 + \sum_{i=1}^d n_i x_i\right)
+ \left(x_0 + \sum_{i=1}^d k_i x_i\right) \\
&= 2x_0 + \sum_{i=1}^d (n_i+k_i)x_i \\
&= 2x_0 + \sum_{i=1}^d m_i x_i \\
&= \psi(m_1,\dots,m_d).
\end{align*}
Therefore $P+P \subseteq \psi(C)$.
[/step]