Let and suppose . The sub-mean inequality on every closed disk contained in gives
Equality holds throughout. If were strictly below at one point of the circle, upper semicontinuity would make it uniformly below on a small arc, contradicting equality of the average. Thus on every sufficiently small circle centered at , and hence throughout a neighborhood of .
The set is therefore open. It is also closed because upper semicontinuity makes open. Since the domain is connected and is nonempty, it is all of . This proves the maximum principle for subharmonic functions.
Let
The resolvent set is open, and the resolvent of an element is operator-valued holomorphic on each of its components. Fix in the resolvent set. For any , choose unit vectors such that
The scalar function is holomorphic. Its modulus is subharmonic, so
Letting proves that the resolvent norm is subharmonic on the resolvent component containing .
Use the convention that the reciprocal resolvent norm is zero on the spectrum. Suppose a bounded component of
contained no spectral point. A spectral point in its boundary would belong to the same pseudospectral component, so is contained in the resolvent set. On one has , whereas inside one has . Continuity on the compact set makes the resolvent norm attain a maximum at an interior point. The subharmonic maximum principle would make it constant, contradicting its boundary values. Hence every bounded component of the pseudospectrum contains spectrum:
Factor the perturbed operator on as
Since
the Neumann series makes invertible. Therefore and
The geometric-series bound gives
Now choose a bounded open neighborhood of the isolated spectral component such that
and meets no other component of the spectrum. Compactness of gives
For all sufficiently large , . Applying the first part to shows uniformly that .
The corresponding Riesz projections are
The resolvent identity and the uniform Neumann bound imply . The projection is nonzero because contains the nonempty spectral component . Projections at distance less than one have isomorphic ranges, so for large . Therefore has spectrum inside , and any such point satisfies . Thus
for every sufficiently large .
If has finite rank, the image of its unit ball is bounded in the finite-dimensional space . The closure of a bounded set in a finite-dimensional normed space is compact. Hence every bounded finite-rank operator is a compact operator.
Let be the orthogonal projection onto
Then strongly. Strong convergence is uniform on every compact subset: if is compact, cover it by finitely many small balls and use at their centers. Since the closure of applied to the unit ball is compact,
The adjoint of a compact operator is compact, so the same argument for gives
Since ,
Therefore
This is the finite-section approximation of a compact operator.
Write for the orthogonal projections and regard the compression as acting on . The compactness argument from part iii applies to any strongly convergent sequence of orthogonal projections, so
We first rule out spectral pollution. Suppose and, after taking a subsequence, . Choose unit eigenvectors :
Compactness gives a convergent subsequence of . Because
the relation then makes converge to a nonzero vector , and passage to the limit gives . Thus every nonzero limit of finite-section spectral points belongs to . The only remaining possible limit is zero, which belongs to the spectrum of a compact operator on an infinite-dimensional space.
Conversely, the Riesz–Schauder theorem says that every nonzero is an isolated eigenvalue of finite algebraic multiplicity. Put a small contour around containing no other point of . Norm convergence of gives uniform resolvent convergence on the contour, so the associated Riesz projections converge in norm and eventually have the same positive rank. Hence meets every neighborhood of .
Finally, zero is also approximated. Otherwise some subsequence would have all its eigenvalues bounded away from zero. Outside any small disk, has only finitely many eigenvalues, and the preceding Riesz-projection argument fixes the total algebraic multiplicity of nearby finite-section eigenvalues. This cannot account for
Thus finite-section eigenvalues also approach zero. Both directed spectral distances vanish, proving

Articles by others on the same topic (0)

There are currently no matching articles.