On local oscillatory integrals with variable Calderón-Zygmund kernels. II (Q1293852)

The authors study the oscillatory singular integral operator \(T_\lambda\) defined by \[ T_\lambda f(x)= \text{p.v. }\int_{\mathbb{R}^n} e^{i\lambda \Phi(x-y)} K(x, x-y) \psi(x- y) f(y) dy, \] where \(\lambda\in \mathbb{R}\), \(\Phi\) is a real analytic function or a real \(C^\infty\) function on \(\mathbb{R}^n\), \(\psi\in C^\infty_0(\mathbb{R}^n)\) and \(K\) is a variable Calderón-Zygmund kernel. By assuming that the kernel \(K\) satisfies the condition \[ \max_{|j|\leq 2n}\Biggl\|{\partial^{|j|}\over\partial y^j} K(x,y)\Biggr\|_{L^\infty(\mathbb{R}^n\times S^{n-1})}= M<\infty, \] the authors prove that \(T_\lambda\) is bounded from \(H^1(\mathbb{R}^n, \omega)\) to \(L^1(\mathbb{R}^n,\omega)\) uniformly in \(\lambda\) for any \(\omega\in A_1(\mathbb{R}^n)\). To describe the proof, we take \(n= 2\) for simplicity (the proof for \(n>2\) is essentially the same). By the well-known atomic decomposition, it suffices to show \(\|T_\lambda a\|_{L^1(\mathbb{R}^n, \omega)}< C\) with a constant \(C\) independent of any \(H^1(\mathbb{R}^n, \omega)\) atom \(a(x)\). Since \(K(x,y)= \Omega(x, y/|y|)/y^2\), following the proof in [\textit{A. P. Calderón} and \textit{A. Zygmund}, Appl. Anal. 7, 221-238 (1978; Zbl 0451.42012)], for each fixed \(x\in \mathbb{R}^n\) one can write the kernel \(K\) in terms of its Fourier series \[ K(x,y)= \sum_{m\neq 0} c_m(x) e^{imy'}/|y|^2, \] where \(c_m(x)= (2\pi)^{-1} \int^{2\pi}_0 \Omega(x, e^{i\theta}) e^{-im\theta} d\theta\) with \(e^{i\theta}= y'= y/|y|\). Thus \[ \|T_\lambda a\|_{L^1(\mathbb{R}^n, \omega)}\leq \sum_{m\neq 0}\|c_m(x)\|_\infty \|T_{\lambda, m}a\|_{L^1(\mathbb{R}^n, \omega)}, \] where each \(T_{\lambda, m}\) contains a standard Calderón-Zygmund kernel \(e^{imy'}/|y|^2\), so that by the known result in [\textit{Y. Pan}, Indiana Univ. Math. J. 41, No. 1, 279-293 (1992; Zbl 0779.42008)], \(\|T_{\lambda, m}a\|_{L^1(\mathbb{R}^n, \omega)}\leq C|m|\). Now using the condition on \(K(x,y)\) and integration by parts, one obtains \(\|c_m(x)\|_\infty\leq C|m|^{-4}\) which yields the desired estimate. For Part I see [\textit{S. Lu}, \textit{D. Yang} and \textit{Z. Zhou}, Integral Equations Oper. Theory 33, No. 4, 456-470 (1999; Zbl 0933.42007)].