A scalar λ for which there exists a nonzero vector v (an eigenvector) satisfying (T−λI)v=0 for a linear operator T, indicating a direction that is invariant up to scaling by λ.

Principle

Eigenvalues are roots of the characteristic polynomial of a finite-dimensional operator; each eigenvalue corresponds to an invariant subspace of nonzero vectors and algebraic information such as multiplicity and possible generalized eigenvectors in the defective case.

Demonstration

For a diagonal matrix with diagonal entries d1,…,dn the eigenvalues are precisely those diagonal entries. A rotation in the plane (nontrivial proper rotation) has complex eigenvalues of modulus 1 but may have no real eigenvalues; a Jordan block of size k for eigenvalue λ implies generalized eigenvectors of chain length k.

Misapplication

Confusing eigenvalues with singular values (metric magnitudes) or assuming the presence of eigenvalues and a full set of eigenvectors in infinite-dimensional spaces without checking compactness or other spectral properties; treating algebraic multiplicity as equal to geometric multiplicity without verification.

Consequence

Knowledge of eigenvalues informs diagonalizability, spectral decomposition, long-term behavior of linear dynamical systems (e.g., growth rates via e^{tA}), and invariants such as determinant (product of eigenvalues, counting multiplicity) and trace (sum of eigenvalues).

Reversal

Reversing the concept gives the resolvent set: scalars for which (T−λI) is invertible so no nonzero eigenvector exists; also consider generalized spectra where λ is in the spectrum but not an eigenvalue (continuous or residual spectrum).

Boundary

In finite-dimensional vector spaces over an algebraically closed field every operator has eigenvalues; over non-algebraically closed fields eigenvalues may lie in an extension field; in infinite dimensions eigenvalues may be absent while the operator still has a nonempty spectrum. Distinguish algebraic vs geometric multiplicity and point vs approximate spectra.

Semantic Tension

Eigenvalue versus singular value or spectral radius: eigenvalues encode algebraic invariant directions (including orientation and complex phases), whereas singular values measure orthogonally invariant magnitudes and spectral radius gives maximal modulus but not directional structure.

Synthesis

An eigenvalue is a scalar that records a nontrivial invariant scaling direction of a linear operator; together with eigenvectors and generalized eigenvectors it organizes the operator's action into decomposable pieces that control algebraic structure and dynamics.