Remarks on Kato's square-root problem (Q1174952)

Kato's square-root problem asks if the domain of the square-root of the differential operator \(\hbox {div}A\nabla\) on \(L_ 2(\mathbb{R}^ n)\) can be obtained as the completion of the domain of \(\hbox{div}A\nabla\) under certain scalar product, when \(A\) is a matrix-valued function such that for some \(\delta>0\) and for all \(x\) in \(\mathbb{R}^ n\) and \(\xi\) in \(\mathbb{C}^ n\), \(\text{Re}(\xi,A(x)\xi)\geq0\). The case \(n=1\) has been solved in the affirmative, and for \(n>1\) a positive answer has been found when \(\| A-I\|<\varepsilon_ n\), with \(\varepsilon_ n\) decaying exponentially with \(n\) [see \textit{E. Fabes et al.}, Proc. Nat. Acad. Sci. USA 79, 5746-5750 (1982; Zbl 0501.35014) and \textit{R. Coifman et al.}, Ann. Inst. Fourier 33, 123-134 (1983; Zbl 0497.35088)]. In this paper, explicit lower bounds for \(\varepsilon_ n\), decaying like \(n^{-1/2}\), are obtained, improving the previously mentioned result. Also it is given a direct proof of the conjectured result for \(n=1\).