Pseudoconvex fully nonlinear partial differential operators: strong comparison theorems (Q1881152)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Pseudoconvex fully nonlinear partial differential operators: strong comparison theorems
scientific article

    Statements

    Pseudoconvex fully nonlinear partial differential operators: strong comparison theorems (English)
    0 references
    0 references
    0 references
    4 October 2004
    0 references
    The authors introduce a class of partial differential operators describing properties of real hypersurfaces of \({\mathbb C}^{n+1}\) as follows. Let \(bD=\{z\in{\mathbb C}^{n+1}:f(z)=0\}\) be a real submanifold, the boundary of the domain \(D=\{z\in{\mathbb C}^{n+1}:f(z)<0\}\), where \(f:{\mathbb C}^{n+1}\rightarrow {\mathbb R}\) is twice continuously differentiable and \(\partial_pf:=(f_{z_1}(p),\dots,f_{z_{n+1}}(p))\neq 0\) for \(p\in bD\), where derivatives are indicated by subscripts. The complex tangent space to \(bD\) at \(p\) is \(T_p^{\mathbb C}(bD)=\{h\in{\mathbf C}^{n+1}: \langle h, \bar\partial_p f\rangle=0\}\) where \(\langle\cdot,\cdot\rangle\) is the usual Hermitian product on \({\mathbb C}^{n+1}\) and \(\bar\partial_p f:=(f_{\bar z_1}(p),\dots,f_{\bar z_{n+1}}(p))\). The complex Hessian of \(f\) is \(\mathcal{H}_p(f):= \left(f_{z_j,\bar z_k}(p)\right)_{j,k=1,\dots,n+1}\) and the Levi form \(L_p(f,\cdot)\) is the restriction to \(T_p^{\mathbb C}(bD)\) of the Hermitian form \(\zeta \mapsto L_p(f,\zeta):= \langle \mathcal{H}_p^T\zeta,\zeta \rangle\). For a given orthonormal basis \(B=\{u_1,\dots,u_n\}\) of \(T_p^{\mathbb C}(bD)\), the \(B\)-normalized Levi matrix of \(bD\) at \(p\in bD\) is the \(n\times n\) Hermitian matrix \[ L_p(f,B) = \left( {1\over{| \partial_p f| }} \langle \mathcal{H}_p^Tu_j,u_k \rangle \right)_{j,k=1,\dots,n}. \] Clearly \(L_p(f,B)\) depends on \(f\) and \(B\), but its eigenvalues depend only on \(D\) and are denoted by \(\lambda(bD)=(\lambda_1,\dots,\lambda_n)\). As in the real case where various curvature functions such as the mean and Gauss curvatures can be defined in terms of the principal curvatures, the authors define various geometric quantities by considering appropriate symmetric functions of the eigenvalues of \(L_p(f,B)\). The Levi-mean curvature is \({1\over n}\sum \lambda_j\), while the total Levi-curvature of \(bD\) is \(S_n(bD)=\prod \lambda_j\). The intermediate quantities \(S_q(bD)\), \(q=2,\dots,n-1\), are defined to be the normalized elementary symmetric functions \(\sigma^{(q)}\) of \(\lambda_1,\dots,\lambda_n\). As in the real case, the functions \(\sigma^{(q)}\) are considered only on the corresponding cone \(\Sigma_q=\{\lambda\in{\mathbb R}^n: \sigma^{(j)}(\lambda)>0 \;\forall\;j=1,\dots,q\}\). A hypersurface \(bD\) with \(\lambda(bD)\in\Sigma_q\) at each point is then said to be \(s^{(q)}\)- pseudoconvex. More general functions than \(S^{(q)}\) are also considered. Some of these equations have been considered previously by \textit{E.~Bedford} and \textit{B.~Gaveau} [Indiana Univ. Math. J. 27, 867--873 (1978; Zbl 0365.32011)], \textit{A.~Montanari} and \textit{F.~ Lascialfari} [J. Geom. Anal. 14, No. 2, 331--353 (2004; Zbl 1217.35082)], and \textit{Z.~Slodkowski} and \textit{G.~Tomassini} [Am. J. Math. 116, 479--499 (1994; Zbl 0802.35050)]. If \(S_q(bD)\) is expressed using the defining function \(f\), then \(S_q(bD)\) is given by an expression involving the first and second derivatives of \(f\). In particular, if \(bD\) is the graph of a function \(u:\Omega\rightarrow{\mathbb R}\) where \(\Omega\) is a domain in \({\mathbb R}^{2n+1}\), then \(u\) satisfies a fully nonlinear equation, which, in contrast to the real case, is not elliptic on \(s^{(q)}\)-pseudoconvex hypersurfaces. Instead, the characteristic form is nonnegative definite with one dimensional kernel. However, the missing null direction can be recovered by taking commutators of suitable vector fields spanning the orthogonal complement of the kernel. The authors use this property to establish a strong comparison principle, which they then use to prove various symmetry theorems for domains with constant curvatures and identification results for domains with comparable curvatures. These results are analogous to their real counterparts.
    0 references
    Levi form
    0 references
    \(s\)-pseudoconvex set
    0 references
    \(s\)-Levi curvature
    0 references
    fully nonlinear degenerate
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references