I want to know why any idempotent matrix $P$ over local ring is similar to a diagonal matrix with elements $0$ and $1$.
I saw a proof in lemma 3.3 of ncatlab, but I don't understand why $P$ is similar to a diagonal matrix if it has an invertible $r$-minor, and why such a matrix can be chosen with entries $0$ and $1$.
Could somebody explain more?
Best Answer
$\newcommand{\Set}[1]{\left\{ #1 \right\}}$Let $V$ be the the underlying free module over the commutative local ring $A$ with unity.
Consider the submodules $$ V_{0} = \Set{ v \in V : P v = 0}, $$ and $$ V_{1} = \Set{ v \in V : P v = v}. $$ Clearly $V_{0} \cap V_{1} = \Set{0}$.
For $v \in V$ one has $$ v = (v - P v) + P v, $$ where
Therefore $V = V_{0} \oplus V_{1}$, and $P$ is zero on $V_{0}$ and the identity on $V_{1}$.
Addendum The point of $A$ being local is that the direct summands $V_{0}, V_{1}$ are also free,