[step:Define the polynomial-time reduction from subset sum]
We reduce from $\mathrm{SUBSET}\text{-}\mathrm{SUM}$, whose input is a finite list of positive integers $a_1,\dots,a_n$ and a target integer $t \ge 0$, all encoded in binary. Let
\begin{align*}
A := \sum_{i=1}^{n} a_i.
\end{align*}
If $t>A$, then the subset-sum instance is a no-instance because every subset sum is at most $A$; in this case define the reduction to output the fixed no-instance $(1,2)$ of $\mathrm{PARTITION}$.
Assume now that $0 \le t \le A$. Define the output list
\begin{align*}
b_1,\dots,b_{n+2} := a_1,\dots,a_n,A+t,2A-t.
\end{align*}
The two appended integers are positive, since $A \ge 1$, $t \ge 0$, and $t \le A$. Their binary lengths are polynomially bounded by the binary lengths of $a_1,\dots,a_n,t$, because $A$ is computed by adding the input integers. Thus the map from $(a_1,\dots,a_n,t)$ to $(b_1,\dots,b_{n+2})$ is computable in polynomial time.
[/step]