MAXIMAL AND SINGULAR INTEGRAL-OPERATORS VIA FOURIER-TRANSFORM ESTIMATES
MAXIMAL AND SINGULAR INTEGRAL-OPERATORS VIA FOURIER-TRANSFORM ESTIMATES
复制标题
DOI:
10.1007/bf01388746
复制
发表时间:
1986-01-01
影响因子:
3.1
通讯作者:
DEFRANCIA, JLR
中科院分区:
文献类型:
--
作者:
DUOANDIKOETXEA, J;DEFRANCIA, JLR
In the classical theory of singular integral operators Tf= K, f, the starting point was a Fourier transform estimate:/~ eL, from which, the inequality I] Tf} l 2< Cllf] l~ follows immediately. This fact together with mild regularity assumptions on the kernel yield a weak type (1, 1)(or L~~ inequality by a nowadays standard procedure whose basic ingredient is the so-called Calderdn-Zygmund decomposition (see [16]). By interpolation and duality, the L rboundedness of T is proved for all 1< p<~. The same scheme was tried for more singular operators, like dt Tf (x)= pv f (x-7 (t))•= K* f (x)--wd (Hilbert transform along the curve 7 in IR"). Under some hypothesis on the curvature of 7, the fact that/s~ can still be proved, but the kernel K is now a distribution supported in the curve,/(a set of measure zero), which is too singular for the Calder6n-Zygmund machinery to be applicable. However, the decay of/~(~)= m (~) is such that one can worsen the multiplier m, obtaining certain m~ _L~ if 0> Re (c0>-a, while, for Re (c~) positive, m, improves, and one actually has m,=/~, with locally integrable kernels K~ which are regular enough to fall under the scope of the classical theory. The L p inequalities for T are then obtained by analytic interpolation of the family of operators T,(T'f)^= fm~. Similar ideas lead to L p inequalities for the associated maximal operators