1: \begin{proof}

2: %First, consider a strategy profile $s \notin S^\ast$.

3: %%Suppose there exists a player $a$ who

4: %%gets less than her fair share of the total research potential, that is, $\sum_{a' \neq a} \sum_i \overline{\vec{q_y^{a',a}}}[i] < 1$.

5: %%distributes her research potential among more than one collaborator.

6: %\todo{Fill this in.}

7: 

8: Consider a strategy profile $s^\ast \in S^\ast$.

9: %Suppose for purpose of contradiction that $s^\ast$ is not an equilibrium; then there exists an unstable set of at most two players.

10: It is obvious that no player can improve her utility if all other players' strategies remain the same, since joint papers are not possible without cooperation from both players.

11: %, so there must be an unstable set of exactly two players.

12: %Let $a'_1$ and $a'_2$ be their corresponding pairs in the matching, respectively.

13: Consider any strategy profile $s'$ differing from $s^\ast$ only in the strategies of players $a_1$ and $a_2$, so that under $s'$ both $a_1$ and $a_2$ invest a non-zero fraction of their research potential into a joint paper. By an argument similar to that in the proof of Lemma~\ref{thm:collab-2p-overtake2},

14: %\note{(make more rigorous?)},

15: it is not possible that both $h^{s'}_n(a_1)$ overtakes $h^{s^\ast}_n(a_1)$ and $h^{s'}_n(a_2)$ overtakes $h^{s^\ast}_n(a_2)$, so by definition $a_1$ and $a_2$ do not form an unstable set. Since this is true for all pairs of players, there does not exist an unstable set of at most two players under $s^\ast$. Thus $s^\ast$ is an equilibrium.

16: \end{proof}

17: