Prove or disprove that if the matrices $P$ and $Q$ represent disjoint permutation cycles in $S_{n}$ with sum of cycle lengths equal to $n,$ then $PQ = P+Q-I$.
MY TRY:
Let's start by an example. Let $P$ and $Q$ be the matrices corresponding to the respective permutations $p = (1 \, 2)$ and $q = (3 \, 4 \, 5)$ in cycle notation. We have that $$
P =
\begin{pmatrix}
0 & 1 & 0 & 0 & 0 \\
1 & 0 & 0 & 0 & 0 \\
0 & 0 & 1 & 0 & 0\\
0 & 0 & 0 & 1 & 0\\
0 & 0 & 0 & 0 & 1
\end{pmatrix} \text{ and } Q = \begin{pmatrix}
1 & 0 & 0 & 0 & 0 \\
0 & 1 & 0 & 0 & 0 \\
0 & 0 & 0 & 1 & 0\\
0 & 0 & 0 & 0 & 1\\
0 & 0 & 1 & 0 & 0
\end{pmatrix}.
$$ It seems obvious that the matrix $PQ$ representing the permutation $pq = (1 \, 2)(3 \, 4 \, 5)$ will be $P+Q-I,$ as the "untouched" $1$s in the matrices are simply canceled by $I$ and the touched $1$s create the derangements. But isn't there some clear method to prove it?
I am new to group theory. Please ask for clarifications in case of any discrepancies. Any hint will be a great help!
Here's a way to transform your great illustration into a proof:
Let $e_1,\dots,e_n$ be the standard basis of $\Bbb R^n$, then $P$ and $Q$ act on these exactly as the corresponding permutations.
So, by the condition, each $e_i$ is either moved by $P$ and fixed by $Q$, or is moved by $Q$ and fixed by $P$, moreover then $Qe_i$ is still in the cycle of $Q$, thus it's also fixed by $P$.
In the first case we have $Qe_i=e_i$, $$PQe_i=Pe_i=Pe_i+e_i-e_i=(P+Q-I)e_i\,.$$ While in the second case we have $Pe_i=e_i$ and $PQe_i=Qe_i$, $$PQe_i=Qe_i=e_i+Qe_i-e_i=(P+Q-I)e_i\,.$$ Since this holds for each element of a basis, it holds for all vectors, and hence $PQ=P+Q-I$.
(Note that this proof also allows $P$ and $Q$ to have more cycles, and the condition that we really need here is that the sets of fixed points of $P$ and $Q$ are disjoint.)