Bilinear Littlewood-Paley for circle and transference (Q555272): Difference between revisions

Let \(m\) be a bounded measurable function defined on \(\mathbb R^{2n}\) and \[ T(f,g) (x) =\int_{\mathbb R^n}\int_{\mathbb R^n}\widehat{f}(\xi) \widehat{g}(\eta) m(\xi ,\eta) e^{2\pi ix.(\xi +\eta)}\,d\xi \,d\eta \] a bilinear operator associated with the symbol \(m\) for the functions \(f,\) \(g\) belonging to the Schwartz space \(S(\mathbb R^n)\). Assume that \(1\leq p_1, p_2 \leq \infty \) and \( \frac{1}{p_1}+\frac{1}{p_2}=\frac{1}{p_3}\). We say that \(m\) is a bilinear multiplier on \(\mathbb R^n\) for the triplet \((p_1,p_2,p_3)\) if the operator \(T\) extends to a bounded operator from \(L^{p_1}\times L^{p_2}\) into \( L^{p_3}\), that is, \[ \| T(f,g)\|_{p_3}\leq c\| f\|_{p_1}\|g\|_{p_2}, \] where \(c\) is a constant independent of the functions \(f\) and \(g\). In the theory of linear and bilinear multipliers, the smooth and non-smooth Littlewood-Paley square functions play important roles. \textit{M. T. Lacey} [Publ. Mat., Barc. 40, No. 2, 387--396 (1996; Zbl 0869.42005)] introduced the bilinear Littlewood-Paley square function. Let \(\{I_l\}_{l\in\mathbb Z}\) be a sequence of disjoint intervals in \(\mathbb R\). Then the associated non-smooth Littlewood-Paley square operator is defined as \[ S(f,g) =\Bigg(\sum_{l\in\mathbb Z}| S_{I_l}(f,g)(x)| ^2\Bigg) ^{\frac{1}{2}},\qquad f,g\in S(\mathbb R), \] where \(S_{I_l}\) is the bilinear multiplier operator associated with the symbol \(\chi _{I_l}(\xi -\eta)\). The smooth version of this operator is defined replacing \(\chi _{I_l}(\xi -\eta)\) by a smooth function \(\varphi _{l}(\xi -\eta)\). \textit{M. T. Lacey} [loc. cit.] proved the boundedness of a bilinear analogue of the smooth Carleson's Littlewood-Paley operator on \(\mathbb R\) for \(p_{3}=2\). \textit{P. Mohanty} and \textit{S. Shrivastava} [Proc. Am. Math. Soc. 138, No. 6, 2095--2098 (2010; Zbl 1196.42006)] extended this results for \(p_3>\frac{4}{3}\) and \(p_1,p_2>2\). \textit{F. Bernicot} proved the boundedness of a version of the non-smooth bilinear Littlewood-Paley operator [Math. Ann. 351, No. 1, 1--49 (2011; Zbl 1225.42009)]. In Section 2, the authors consider bilinear Littlewood-Paley square functions on the circle group \({\mathbf T}\). They prove the boundedness of this operator for possible exponents. Also, they prove a bilinear analogue of F. Bernicot's result on \({\mathbf T}\). They use suitable vector valued transference techniques to obtain these results. For an almost periodic function \(f,\) its Bohr transform at a point \(\lambda \in \mathbb R\) is defined as \[ \widehat{f}(\lambda) =\lim_{T\to \infty} \frac{1}{2T} \int_{-\infty }^\infty f(x) e^{-2\pi i\lambda x}\,dx. \] Note that here we are using the same notation for the Bohr transform and the Fourier transform. Let \(B^{p}(\mathbb R)\) be the Besicovitch space, where \(0<p\leq \infty\). We say that a bounded function \(m(t,s)\) is an almost periodic bilinear multiplier for the triplet \((p_1,p_2,p_3)\) if the operator \(B\), defined for the trigonometric polynomial \(f\) and \(g\) by \[ B(f,g)(x) =\sum_{\text{finite}}\;\sum_{\text{finite}}m(t,s) \widehat{f}(t) \widehat{g}(s) e^{2\pi ix(t+s)}, \] extends to a bounded operator from \(B^{p_1}(\mathbb R) \times B^{p_2}(\mathbb R)\) into \(B^{p_3}(\mathbb R)\). \textit{K. de Leeuw} proved in [Ann. Math. (2) 81, 364--379 (1965; Zbl 0171.11803)] that the multipliers on \(L^p(\mathbb R)\) are precisely the ones which are multipliers on \(B^p(\mathbb R)\). In Section 3, the authors prove some bilinear analogues of de Leeuw's results concerning multipliers on \(\mathbb R^n\). In the same paper, de Leeuw proved that, if a continuous function \(m\) is an \(L^p(\mathbb R)\) multiplier, then \(m\mid _{\mathbb R^n}\)\ is an \(L^k(\mathbb R)\) multiplier for \(k<n\). A bilinear analogue of this result was proved by \textit{G. Diestel} and \textit{L. Grafakos} [Nagoya Math. J. 185, 151--159 (2007; Zbl 1131.43003)] for \(p_3\geq 1\). In Section 4, the authors extend this result to the entire range of the exponents \(p_1, p_2\) and \(p_3\).