論文

© 2019 Mathematical Sciences Publishers. The endpoint estimates for the Kunze-Stein phenomenon associated with real rank one semisimple Lie groups and the Jacobi hypergroup were obtained by A. Ionescu and J. Liu respectively. Recently, a modified endpoint estimate was also proved for complex semisimple Lie groups by J. Liu and the author. Hence the original estimate seems not to be valid for general semisimple Lie groups with higher rank. We treat the case of SU(n,m) and obtain a modified estimate similarly to the complex case. The process using the Abel transform makes it clear why the original endpoint estimate might not be valid for general semisimple Lie groups.

We shall define a Calderón-Zygmund class CZp(Δα,β) on the Jacobi hypergroup (R+, Δα,β, ∗ ) such that, if a function g on R+ belongs to CZp(Δα,β), then the convolution operator g∗ is bounded from Lq(Δα,β) to itself for p≤ q≤ 2. Actually, we shall obtain a relation between the Lp norms of g and the Abel transform Ag and a transference principle between the Lp operator norms of g∗ and the Euclidean operator ϕ⊛, where ϕ(x)=e(2p-1)ρxAg(x). Therefore, to define the Calderón-Zygmund class CZp(Δα,β), we shall obtain some conditions on g under which ϕ belongs to CZ (R). Then, ϕ⊛ is bounded on Lq(R) and, by the transference principle, g∗ is bounded on Lq(Δα,β).

©2018 International Academic Printing Co. Ltd. All Rights Reserved. For the Jacobi hypergroup (R+,Δ,), the weak-L1 estimate of the Hardy-Littlewood maximal operator was obtained by W. Bloom and Z. Xu, later by J. Liu, and the endpoint estimate for the Kunze-Stein phenomenon was obtained by J. Liu. In this paper we shall give alternative proofs based on the inverse Abel transform for the Jacobi hypergroup. The point is that the Abel transform reduces the convolution to the Euclidean convolution. More generally, let T be the Hardy-Littlewood maximal operator, the Poisson maximal operator or the Littlewood- Paley g-function for the Jacobi hypergroup, which are defined by using. Then we shall give a standard shape of Tf for f ∈ L1(Δ), from which its weak-L1 estimate follows. Concerning the endpoint estimate of the Kunze-Stein phenomenon, though Liu used the explicit form of the kernel of the convolution, we shall give a proof without using the kernel form.