Definition

Let T:D(T)XXT:\mathcal D(T)\subseteq X\to X be a on a complex , and assume that it is . A scalar λ\lambda in the belongs to the point spectrum when TλIT-\lambda I is not injective, equivalently when λ\lambda is an . If TλIT-\lambda I is injective, then λ\lambda belongs to the continuous spectrum when its range is dense but not all of XX, and to the residual spectrum when its range is not dense. With this disjoint convention, these three sets partition the spectrum of a closed operator.

Hilbert-space description

For a densely defined operator on a complex ,

Ran(TλI)=ker(TλI).\overline{\operatorname{Ran}(T-\lambda I)} =\ker(T^*-\overline{\lambda}I)^\perp.

Consequently, an injective TλIT-\lambda I has non-dense range exactly when λ\overline{\lambda} is an eigenvalue of the TT^*. A has no residual spectrum: its spectrum is the disjoint union of its point and continuous spectra.

Representative examples

In finite-dimensional spaces every spectral value is an eigenvalue, so the continuous and residual spectra are empty. Multiplication by the coordinate function on L2([0,1])L^2([0,1]) has spectrum [0,1][0,1] and no eigenvalues; every point of the interval lies in its continuous spectrum. The unilateral shift

S(x0,x1,)=(0,x0,x1,)S(x_0,x_1,\ldots)=(0,x_0,x_1,\ldots)

on 2(N)\ell^2(\mathbb N) is injective, but its range is the closed proper subspace of sequences whose first coordinate is zero. Thus 00 lies in its residual spectrum.

Conventions and scope
References
  1. Michael Reed and Barry Simon, Methods of Modern Mathematical Physics I: Functional Analysis, Academic Press, 1972. Publisher record. Relevant: Chapter VII on the point, continuous, and residual spectra.
  2. Nelson Dunford and Jacob T. Schwartz, Linear Operators, Part II: Spectral Theory, Interscience, 1963. Publisher record. Relevant: Chapter VII on spectral subdivisions.