Operator-valued Fourier multiplier theorems and maximal \(L_p\)-regularity (Q5947416)

One says that an operator valued function \(M:\mathbb{R}\backslash \{0\}\to B(X,Y)\) is a Fourier multiplier for \(L^{p}(\mathbb{R},X)\) provided the operator \(T:f\mapsto (M(\cdot)[\hat{f}(\cdot)]^{\vee })\) where \(\hat{f}(\xi)=\int e^{-it\xi }f(t)dt\). In the case \(Y=X\) the Mihlin multiplier theorem takes the form:\ if \(M:\mathbb{R}\backslash \{0\}\to B(X,X)\) satisfies \(\|M(t)\|\leq C\) and \(\|tM'(t)\|\leq C\) for all \(t\neq 0\) then \(M\) is a multiplier for \(L^{p}(\mathbb{R},X)\), \(1<p<\infty \), with norm depending only on \(C\), if and only if \(X\) is a Hilbert space. The sufficiency was proved by L. Schwartz, the necessity by G. Pisier. The purpose of the present paper is to prove multiplier theorems for more general Banach spaces. As such, additional hypotheses on \(X\) and on \(M\) are required; in the present context these properties are the UMD\ property -- essentially that the Hilbert transform extends to a bounded operator on \(L^{p}(\mathbb{R},X)\), \(1<p<\infty \), along with the \(R\)-boundedness property which on a subset of \(U\subset B(X,Y)\) which says that any finite subset \( \{T_{1},\dots ,T_{n}\}\subset U\) satisfies \(\int_{0}^{1}\|\sum_{j=1}^{n}r_{j}(u)T_{j}(x_{j})\|_{Y}\leq C\int_{0}^{1}\|\sum_{j=1}^{n}r_{j}(u)x_{j}\|_{X}\). Here \(r_{j}\) is the \(j\)th Rademacher function. The author proves that if \(X,Y\) are UMD spaces and the sets \(\{M(t):t\neq 0\}\) and \(\{tM'(t):t\neq 0\}\) are \(R\)-bounded then \(M\) is a Fourier multiplier for \(L^{p}(\mathbb{R},X)\). The proof relies on ideas of Bourgain linking the \(R\)-boundedness property to Littlewood-Paley type square functions. As an application the author proves that maximal \(L^{p}\) regularity of a bounded analytic semigroup of operators is essentially equivalent, in the case of UMD-spaces, to the \(R\)-boundedness of the a resolvent set of the generator of the semigroup. The latter application relies on analytic extensibility of the \(R\)-boundedness property.