Infinite-dimensional diagonalization and semisimplicity (Q501855)

Let \(V\) be a vector space over a field \(K\) and \(\mathcal B\) a basis of \(V\). A \(K\)-linear operator \(T\in\mathrm{End}(V)\) is said to be \textit{\(\mathcal B\)-diagonalizable}, if every \(v\in\mathcal B\) is an eigenvector for \(T\). The set of \(\mathcal B\)-diagonalizable operators on \(V\) is a commutative subalgebra of \(\mathrm{End}(V)\), which is denoted by \(\mathrm{diag}(\mathcal B)\). It is proved that \(\mathrm{diag}(\mathcal B)\) is a maximal commutative subalgebra of \(\mathrm{End}(V)\) and \(\mathrm{diag}(\mathcal B)\cong K^{\dim V}\). A subalgebra \(D\subseteq\mathrm{End}(V)\) is said to be \textit{diagonalizable} if \(D\subseteq\mathrm{diag}(\mathcal B)\) for some basis \(\mathcal B\) of \(V\). It is shown that the classification of all diagonalizable subalgebras of \(\mathrm{End}(V)\) reduces to the classification of those ones which are closed in the topology of pointwise convergence on \(\mathrm{End}(V)\). Such subalgebras are described in the following theorem. Theorem. For a closed commutative subalgebra \(C\) of \(\mathrm{End}(V)\) the following are equivalent: {\parindent=0.7cm\begin{itemize}\item[(i)] \(C\) is diagonalizable; \item[(ii)] \(C\) is topologically isomorphic to \(K^\Omega\) for some cardinal \(1\leq\Omega\leq\dim(V)\); \item[(iii)] there is a set of pairwise orthogonal idempotents \(\{E_i\}\subseteq C\), such that \(\overline{\mathrm{span}}\{E_i\}=C\) and the net of finite sums of \(E_i\) converges to \(1\). \end{itemize}} If \(K\) is algebraically closed, then another two equivalent conditions in terms of the so-called \textit{pseudocompactness}, semisimplicity and topological nilpotency are given.