Hölder and \(L^p\) estimates for \(\overline {\partial}\) on convex domains of finite type depending on Catlin's multitype (Q1433417)

The author obtains optimal Hölder and \(L^p\) estimates for \(\overline\partial\) on convex domains of finite type, which generalize results of \textit{K. Diederich}, \textit{B. Fischer} and \textit{J. E. Fornaess} [Math. Z. 232, No. 1, 43--61 (1999; Zbl 0932.32008)] and \textit{B. Fischer} [Math. Z. 236, No. 2, 401--418 (2001; Zbl 0984.32022)]. The results of the author are based on a deep study of the geometry of these domains, linking the Hölder exponent to Catlin's multitype. This is possible using the results of \textit{J.~Yu} [Indiana Univ. Math. J. 41, No. 3, 837--849 (1992; Zbl 0759.32009)] and \textit{J. D.~McNeal} [J. Funct. Anal. 108, No. 2, 361--373 (1992; Zbl 0777.31007)]. More precisely if \(D\) is a convex domain in \(\mathbb{C}^n\), with smooth boundary \(bD\) and if \(k = (k_1, \ldots, k_n)\) is the multi type of \textit{D. Catlin} [Ann. Math. (2) 120, 529--586 (1984; Zbl 0583.32048)] at a point \(p \in bD\), J. Yu has shown (loc. cit) that the entry \(k_{n -q + 1}\) equals the maximal order of contact of \(p\)-dimensional complex manifolds through \(p\) with \(bD\). Yu has also introduced a special basis of \({\mathbb C}^n\) which satisfies some geometric conditions relative to the point \(p\). On the other hand McNeal (loc. cit) introduced another kind of basis associated to each point of a sufficiently small neighborhood of \(bD\), called the \(\varepsilon\)-extremal basis. The author introduces what he calls an \(\varepsilon\)-minimal basis which possesses most of the relevant properties of the McNeal basis but also some useful properties of the Yu basis at \(p\). This allows the author to obtain very precise estimates which inserted in the integral estimates of Diederich-Fischer-Fornaess give the solution with optimal Hölder exponents. The main result states that the maximal order of contact with \(bD\) of \((q + 1)\)-dimensional complex manifolds passing through points of \(bD\) determine the optimal Hölder (and \(L^p\)) estimates for solutions of the \(\bar\partial\)-equation on \((0, q + 1)\)-forms. For example, there exists a bounded linear integral operator \(T_q : C^\circ_{0, q + 1} (\bar{D}) \to \Lambda^{\alpha(q)}{0, q} (D)\), \(q = 0, 1, \ldots, n - 1\) (where \(\Lambda_{0, q}^{\alpha(a)}\) is the space of \((0, q)\) forms with Hölder coefficients of order \(\alpha(a)\)) such that \(\bar\partial T_q f = f\) where \(D \subset\subset {\mathbb C}^n\) is a convex domain with smooth boundary of type \(\mu_n\), and if \(\mu = (\mu_1, \ldots, \mu_n)\) is the multi type of \(bD\) and \(\lambda_q = \mu_{n - q} (n - q) + 2q + 2\) then \(\alpha(q) = \frac{1}{\mu_n - q}\) for \(q = 0,1 \ldots, n \cdot 2\). Similar estimates hold for \(L^p\). To obtain optimal \(L^p\) estimates the author refines Fischer's estimate using the \(\varepsilon\)-minimal basis.