Problem Statement
Given a graph G with each vertex assigned with a binary state on/off.
Each move consists of choosing a vertex, then change the state of itself and all of its neighbor(s).
Prove that from the all-off starting position, it is possible to reach the all-on position.
Idea
Let $n$ be the number of vertex in $G$.
Let $A$ be the adjacent matrix of $G$:
$$ A_{ij} = \begin{cases} 1 &\text{if } i=j \text{ or }i\text{ and }j \text{ are adjacent in G} \\ 0 &\text{otherwise } \end{cases} $$
Another observation is that performing 2 moves on the same vertex has net zero effect regardless of other moves.
So the problem boils down to whether there exists $v \in \mathbb{Z_2}^n$ such that:
$$Av = \overrightarrow{1} $$
We can prove by induction on $n$. The base case is trivial. Assume that it’s true for $n-1$.
Let $V$ be the row span of $A$. Assume the contradiction that $\overrightarrow{1} \not \in V$.
Denote $\bar{i}$ be the vector that has 0 on row $i$ and 1 on the others.
Using the induction hypothesis on first minor matrix $M_{i,i}$ of A we get:
$$\bar{i} \in V$$
If $n$ is even then $\overrightarrow{1} = \sum_{i} \bar{i} \in V$. So we only continue with the assumption that $n$ is odd.
Any proper subset $S$ of $\lbrace \bar{i} \rbrace$ is linear independent. Let $k = \sum_{i \in S} \bar{i}$ then either $k_{i \in S} = 1$ or $k_{i \not \in S} = 1$.
But $\overrightarrow{1} \not \in V$ so $\lvert V \rvert = 2^{n-1}$ or the row vector of $A$ forms a minimal dependent set.
$$\sum_{i} A_i = \overrightarrow{0} $$
But this is a contradiction as the number of 1 entry in $A$ is always odd if $n$ is odd ($A$ is symmetric and reflexive).
Aftermath
Later I found out there’s a generalized result with much more elegant proof:
For any symmetric matrix $A$, the diagonal vector is always in the row span of $A$.
The combinatoric proof of the problem uses the corrolary of the 10th problem:
Any graph $G$ can be partitioned into 2 possibly empty subgraphs, such that the degree of each vertex in the induce graph of one is odd while the other is even.