Coburn's alternative for Toeplitz operators #
For a nonzero essentially bounded symbol a on the unit circle, either the Toeplitz operator
T_a or its adjoint T_a† = T_ā is injective. Consequently a Fredholm Toeplitz operator of
index zero is invertible; this is how invertibility of T_a for a continuous symbol a reduces
to the computation of its Fredholm index.
The proof is the Hardy product argument. A Hardy function f lies in the kernel of T_a exactly
when a f has only negative Fourier modes (TauCeti.toeplitz_apply_eq_zero_iff). Suppose that
T_a f = 0 and T_ā g = 0. The conjugate conj g of a Hardy function has only nonpositive
modes, so the integrable product a f conj g has only negative modes. Its complex conjugate
ā g conj f has only negative modes for the same reason, so a f conj g also has only positive
modes. Hence all Fourier coefficients of a f conj g vanish and a f conj g = 0 almost
everywhere. If f and g are nonzero, then by boundary uniqueness they are nonzero almost
everywhere, which forces a = 0.
Main results #
TauCeti.eq_zero_of_toeplitz_apply_eq_zero_of_toeplitz_star_apply_eq_zero: ifT_aandT_āboth have nonzero kernel vectors, thena = 0.TauCeti.ker_toeplitz_eq_bot_or_ker_adjoint_toeplitz_eq_bot: Coburn's alternative: fora ≠ 0,ker T_a = 0orker T_a† = 0.TauCeti.isInvertible_toeplitz_of_isFredholm_of_index_eq_zero: a Fredholm Toeplitz operator of index zero is invertible.
References #
- L. A. Coburn, Weyl's theorem for nonnormal operators, Michigan Math. J. 13 (1966), 285–288.
- E. Park, Toeplitz operators, for the Hardy product argument.
- A. Böttcher and B. Silbermann, Analysis of Toeplitz Operators, second edition, Springer (2006), Chapter 2.
The Hardy product argument. If the Toeplitz operators T_a and T_ā = T_a† have nonzero
kernel vectors f and g, then the symbol a is zero.
Coburn's alternative. For a nonzero essentially bounded symbol a, either the Toeplitz
operator T_a or its adjoint has trivial kernel.
A Fredholm Toeplitz operator of index zero is invertible.