Osamu Hatori, Shiho Oi
4 min
This paper by Osamu Hatori and Shiho Oi advances the spectral theory of doubly power-bounded elements in complex unital Banach algebras, generalizing classical results from operator theory to the algebraic setting. It builds on Gelfand's 1941 theorem (characterizing the identity as the unique doubly power-bounded operator with singleton spectrum {1}) and extends recent work by the authors on power-bounded operators. The focus is on invertible elements with finite spectrum on the unit circle, providing tools to decompose them explicitly and exploring implications for algebra commutativity.
An element in a Banach algebra is doubly power-bounded if it is invertible and . Intuitively, both positive powers (n>0) and negative powers stay norm-bounded. A key lemma shows their spectrum lies on the unit circle : if were in , then but wouldn't be bounded; similarly for . This mirrors power-bounded operators (e.g., surjective isometries) but extends to abstract algebras.
The paper's centerpiece (Theorem 4.2) is a spectral decomposition for doubly power-bounded with finite spectrum : where are orthogonal idempotents (, for , ) satisfying for . This generalizes Gelfand's theorem (m=1 implies ) and uses Riesz projections defined via contour integrals around isolated spectral points: , where encircles . Boundedness of resolvents (from double power-boundedness) ensures these exist in algebras.
For an isolated , the paper proves is an eigenvalue with complemented eigenspace via the Riesz projection , generalizing Koehler-Rosenthal for operators. If is separated, is a complemented subalgebra. This relies on Ilišević's theorem (revisited in Section 2): an element is algebraic (satisfies a polynomial equation) iff for coprime polynomials with , . Applied here, double power-boundedness forces algebraic behavior near isolated points.
Section 5 initiates studying if the closure of doubly power-bounded elements is an algebra (closed under multiplication) implies is commutative. This connects to longstanding problems: e.g., does every element have finite spectrum? Open Question 5.3 asks if such closure forces commutativity, with examples like disk algebras (non-commutative DPB sets). This hints at DPB elements probing deeper algebraic structure.
Why care? These results bridge operator theory (spectral theorem for normal operators) to general Banach algebras, where Gelfand theory gives maximal ideals but not always decompositions. Finite-spectrum DPB elements behave 'almost diagonalizable,' offering intuition for when algebras mimic .
We establish a characterization of doubly power-bounded elements with finite spectrum in Banach algebras. In particular, we present a spectral decomposition for such elements, extending a classical theorem of Gelfand concerning doubly power-bounded elements with singleton spectrum. Furthermore, we generalize a theorem of Koehler and Rosenthal for doubly power-bounded elements to the setting of Banach algebras. In the final section, we are initiating a study to investigate whether the properties of doubly power-bounded elements can offer insight into the commutativity of Banach algebras.
Sam: Yes, the embedding preserves the structure and sizes. The projections for S_b stay in the image, a subalgebra, so the p_j are true idempotents in the original. Non-commutativity doesn't block it—the contours ensure orthogonality.
Alex: They build on results like Koehler-Rosenthal, where isolated points give complemented eigenspaces. For finite spectrum, everything decomposes.
Sam: Yes. The paper shows equivalence: decomposition with orthogonal idempotents summing to identity, or b algebraic with minimal polynomial the product (b - lambda_j identity) equals zero. Projections come from products over the other lambdas.
Alex: So stability checks get clearer for these nearly unitary elements with finite spectra.
Sam: A meaningful step. It gives explicit block decompositions for systems with finite spectral points on the unit circle, aiding work in operator algebras. Toward the end, they ask if doubly power-bounded elements closing under multiplication implies commutativity. They pose it as open for semisimple algebras, with counterexamples in C*-algebras like upper triangular matrices or Toeplitz operators.
Alex: A clear advance for decompositions, with pointers ahead. Thanks, Sam—this illuminates handling nearly unitary elements in challenging settings. That's our look at spectral decompositions in Banach algebras. Thanks for listening to ResearchPod.