Quasi-\(L^p\)-contractive analytic semigroups generated by elliptic operators with complex unbounded coefficients on arbitrary domains (Q1758223)

This paper is devoted to the study of general elliptic operators of the form \[ Au:=-\sum_{k,j =1}^N D_k(a_{kj} D_j u)+\sum_{k=1}^N b_{1k} D_k u + \sum_{k=1}^N D_k (b_{2k} u) + a_0 u, \] on domains \(\Omega\subset \mathbb R^n\), with measurable coefficients. It is well known that, under suitable assumptions on the coefficients, such operators are analytic semigroup generators on \(L^2(\Omega)\). Much effort has been devoted over the last twenty years to extend the theory to more general operators, with possibly unbounded and/or complex-valued coefficients. The authors closely follow the approach popularized among others in [\textit{E. B. Davies}, Heat kernels and spectral theory. Cambridge University Press (1989; Zbl 0699.35006)] and \textit{E. M. Ouhabaz} [Analysis of heat equations on domains. Princeton University Press (2005; Zbl 1082.35003)] and mainly based on the theory of closed sesquilinear forms: they introduce the natural, possibly non-symmetric form \[ a(u,v):=\sum_{k,j =1}^N \int_\Omega a_{kj} D_j u \overline{D_k v}+\sum_{k=1}^N \int_\Omega b_{1k} D_k u \overline{v} - \sum_{k=1}^N b_{2k} u \overline{D_k v} + \int_\Omega a_0 u \overline{v}, \] and associate with it a realisation naturally defined by means of a form domain \(V\) that is assumed to be a subspace that lies between \(H^1_0(\Omega)\) and \(H^1(\Omega)\) -- or rather, of weighted Sobolev spaces defined in analogy to \(H^1_0(\Omega)\) and \(H^1(\Omega)\), the weights arising naturally from the authors' assumptions on ellipticity and growth conditions of the coefficients of the form. The authors show in Theorem 2.1 that this form is associated with the generator of a quasi-contractive, analytic semigroup on \(L^2(\Omega)\). Under additional assumptions on the lattice structure of \(V\) and/or on the symmetry of the coefficient matrix, Theorem 2.2 states that such semigroup extends to all \(L^p(\Omega)\)-spaces, \(1<p<\infty\). Depending on the imposed assumptions, the proofs are either based on domination results related to the classical theory of modulus semigroups; or else on the Trotter-Kato Theorem, after a tricky decomposition of the form has been performed. These results are extended in Section 5 to elliptic operators in non-divergence form, provided certain compatibility and regularity assumptions on the coefficients are satisfied: these assumptions however include differentiability of the elliptic coefficients, which makes the results less interesting, as the considered non-divergence operators can be promptly transformed into operators considered in the previous sections. Finally, several generalised Black-Scholes and Ornstein-Uhlenbeck operators are shown to fit the proposed setting under suitable compatibility assumptions.