polynomial functional calculus
Functional calculus is the art of feeding an operator into a function. The polynomial functional calculus is the most basic version: given a polynomial p and an operator T, form p(T). What makes it a 'calculus' rather than mere substitution is that it behaves like genuine function application — sums of functions give sums of operators, products give products, and constants give scalar multiples of I.
The structural fact underneath is the evaluation homomorphism p -> p(T). Because it is a ring homomorphism, algebraic identities among polynomials transfer verbatim to operators. If p(x) q(x) = r(x) as polynomials, then p(T) q(T) = r(T) as operators — automatically. You can do algebra in F[x], a comfortable commutative world, and read the answer off as an operator.
The key reduction principle: anything p(T) is already determined by p modulo the minimal polynomial. Since m(T) = 0, two polynomials that agree modulo m produce the same operator. So the algebra of polynomials in T is really the finite-dimensional algebra F[x]/(m), and every p(T) can be replaced by a representative of degree below deg(m). This is how you keep computations bounded.
The payoff is concrete and surprising. When T is invertible you can write T^-1 as a polynomial in T; functions you would not expect to be polynomial — a chosen branch, a spectral projection — become polynomials in T via interpolation; and for self-adjoint or normal operators this polynomial calculus extends to a continuous functional calculus handling sqrt, exp, and beyond. The polynomial case is the seed of all of it.
Polynomial functional calculus turns polynomial identities into operator identities, and the minimal polynomial caps how large the algebra of polynomials in T can be.
The dimension of the algebra F[T] equals deg(m), the degree of the minimal polynomial — not n. So even a huge operator may have a tiny polynomial algebra: for any scalar operator cI the algebra F[T] is just the scalars, since m(x) = x - c has degree 1.