simultaneous spectral decomposition

One self-adjoint or normal operator gets a single orthonormal eigenbasis. A natural question is when two of them can share one basis. The answer is as clean as you could wish: a family of normal operators can be simultaneously unitarily diagonalized if and only if they all commute with one another. Commuting is exactly the obstruction; nothing else stands in the way.

Concretely, if S and T are normal and S T = T S, then there is a single orthonormal basis q_1, ..., q_n in which both are diagonal at once: each q_i is an eigenvector of S and of T simultaneously. The intuition is that commuting operators preserve each other's eigenspaces — T maps each eigenspace of S into itself — so you can restrict T to one eigenspace of S at a time and diagonalize it there without disturbing S. Iterating refines the basis until both are fully diagonal.

Each shared eigenvector now carries a pair of eigenvalues, (lambda from S, mu from T), and the collection of such pairs is the joint spectrum. This is the linear-algebra seed of a profound idea: a commuting family of observables can be measured at once because they share eigenstates, which is the mathematical heart of compatible measurements in quantum mechanics. It also explains why a function f(T) always commutes with T (they share the basis) and underlies the structure of the commutant of a normal operator.