The set of scalars λ in the base field (typically C) for which the operator T−λI fails to be invertible in the ambient algebra/topology; for bounded operators on Banach spaces the spectrum splits into point, continuous and residual parts.

Principle

Spectrum characterizes failure of invertibility rather than mere roots of polynomials; it is a topological-algebraic invariant determined by both algebraic equations and the operator topology and controls functional calculus and stability.

Demonstration

For a finite-dimensional matrix, the spectrum coincides with the multiset of eigenvalues (roots of the characteristic polynomial). For the unilateral shift on l^2, the spectrum is the closed unit disk even though there are no eigenvalues in the open disk.

Misapplication

Assuming the spectrum equals the set of eigenvalues in infinite-dimensional settings leads to mistakes: there can be spectral values that are not eigenvalues (continuous spectrum) and such values still influence dynamics and resolvent behavior.

Consequence

Knowing the spectrum permits application of the spectral mapping theorem, spectral radius calculations, and construction of resolvents and functional calculi; it governs long-term growth of iterates and solvability of linear equations (T−λI)x=y).

Reversal

The complementary notion is the resolvent set: scalars for which T−λI is boundedly invertible and whose resolvent operator exists and is analytic in λ. Viewing the problem via resolvent rather than spectrum emphasizes invertibility and analytic dependence.

Boundary

Spectrum depends on the operator class and topology: definitions and decompositions differ between matrices, bounded operators on Banach spaces, unbounded operators on Hilbert spaces, and elements of Banach algebras. Over non-algebraically closed fields the spectrum may be viewed in an extension field.

Semantic Tension

A tension exists between 'spectrum' and 'eigenvalues': in finite dimensions they coincide, but in infinite dimensions the spectrum strictly contains eigenvalues; distinguishing point, approximate and residual spectra is necessary to avoid conflation.

Synthesis

The spectrum is the complete set of scalar parameters that preclude invertibility of T−λI in the relevant algebraic-topological setting; it is the central invariant that determines resolvent behavior, spectral measures, and the applicability of functional calculus.