Analysis of the Kohn Laplacian on quadratic CR manifolds. (Q1413975)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Analysis of the Kohn Laplacian on quadratic CR manifolds. |
scientific article |
Statements
Analysis of the Kohn Laplacian on quadratic CR manifolds. (English)
0 references
17 November 2003
0 references
Let \(V\) be an \(n\)-dimensional vector space, \(W\) an \(m\)-dimensional real vector space, \(W^{\mathbb{C}}\) its complexification, \(\Phi: V\times V\to W^{\mathbb{C}}\) a Hermitian map, and \(S= \{(z, t+ iu)\in V\times W^{\mathbb{C}}: u= \Phi(z,z)\}\) the associated quadratic manifold; \(S\) is a CR manifold of CR dimension \(n\) and real codimension \(m\). The form \(\Phi\) can be identified with the Levi form on \(S\) (vector valued), and a central role in what follows depend on the signatures of the scalar valued-forms \(\Phi^\lambda(z, z')= \lambda(\Phi(z, z'))\), \(\lambda\in W^*\) (where \(W^*\) is the dual of \(W\)). Let \(n^+(\lambda)\) (respectively \(n^-(\lambda)\)) the number of positive, respectively negative eigenvalues of \(\Phi^\lambda\). Consider now the Kohn Laplacian \(\square^{(a)}_b= \partial_b \overline\partial^*_b\overline \partial_b\) acting on \((0,q)\) forms on \(S\), where \(\overline\partial^*_b\) is the adjoint of \(\overline\partial_b\) (with respect to the Lebesgue \(dz\,dt\) on \(S\) and to a fixed Hermitian inner product on \(V\)). The main results of the authors are: 1) The Kohn Laplacian \(\square^{(a)}_b\) is solvable iff there is no \(\lambda\in W^*\) such that \(n^+(\lambda)= q\) and \(n^-(\lambda)= m-q\) (when \(\square^{(a)}_b\) is not solvable it is not even locally solvable); 2) \(\square^{(a)}_b\) is hypoelliptic iff there is no \(\lambda\in W^*\setminus\{0\}\) such that \(n^+(\lambda)\leq q\) and \(n^-(\lambda)\leq n- q\). As a corollary, it results that \(\square^{(a)}_b\) is solvable for any \(q\) if the forms \(\Phi^\lambda\) are degenerate for all \(\lambda\). These results are obtained by identifying \(S\) with a step 2-nilpotent Lie group \(G_\Phi\) in the following manner: Let \(S\) be defined by \(\text{Im\,}w= \Phi(z, z)\), \(z\in V\), \(u\in W^{\mathbb{C}}\). For \((z',w')\in S\) consider the complex-affine transformation of \(V\times W^{\mathbb{C}}\) given by \(\tau_{(z',w')}(z, w)= (z+z', w+ w'+ zi\Phi(z,z'))\) which maps \(S\) onto itself. As \(\tau_{(z',w')} \tau_{(z'', w'')}= \tau_{z'+ z'',w'+ w''+ zi\Phi(z',z')}\) and identifying \(\tau_{(z',w')}\) with \((z', w')\in S\), one has a Lie group structure on \(S\); pulled back on \(V\times W\) the group multiplication takes the form \((z,t)\cdot (z',t)= (z+ z', t+ t', 2\text{\,Im\,}\Phi(z, z'))\). This group is denoted by \(G_\Phi\) and \(g_\Phi\) its Lie algebra. The study of the Kohn Laplacian on \(G_\Phi\), in particular the anlysis of the image of \(\square^{(a)}_b\) as a system of harmonic oscillators, the Fourier analysis, on \(G_\Phi\) (in particular a Plancherel formula), allows the authors to construct a tempered fundamental solution \(K= K_q\) for \(\square ^{(q)}_b\), first in the case when the cone \(\Omega_q= \{\lambda, n^+(\lambda)= q, n^-(\lambda)= n- q\}\) is empty (and the authors consider separately the case \(\nu= n\) and \(\nu< n\) (here \(0\leq\nu(\lambda)= \text{rank\,}\Phi^\lambda\leq n\) and \(\nu= \max_{\lambda\in W^*}v(\lambda))\). For \(\nu= n\) and \(\Omega_q= \emptyset\), the explicitely defined fundamental solution is a well-defined temperature distribution on \(G_\Phi\), and \(K_q\) is a global homogeneous fundamental solution of \(\square^{(q)}_b\). When \(\Omega_q\) is non-empty, minor modifications permit to construct a ``relative'' fundamental solution \(L_{rel}\), i.e. satisfying \(\square^{(a)}_b K_{rel}= \delta_0\otimes I-{\mathcal C}_q\) generalize the Gindikin formula for the Cauchy-Szegő kernel [\textit{S. G. Gindikin}, Usp. Mat. Nauk 19, No. 4(118), 3--92 (1964; Zbl 0144.08101)] and is given explicitly. The case when \(\Phi^\lambda\) is degenerate for all \(\lambda\) is studied first when \(\nu< n-1\) and when \(\nu= n-1\). Let us mention that the authors prove that the orthogonal projection of \(L^2(\mathbb{R}^n)\otimes \Lambda_q\) onto the null space of \(\square^{(a)}_b\) maps a form \(\omega\) into \(\omega*{\mathcal C}_q\). Concerning the hypoellipticity \(\square^{(a)}_{\mathcal C}\), the authors show that four sets of conditions are equivalent to the hypoellipticity of \(\square^{(q)}_b\), the condition ``for every \(\lambda\neq 0\), \(\Phi\lambda\) has at least \(\max(q+ 1,n- q+1)\) eigenvalues with the same sign, or at least \(\min(q+ 1, n-q+1)\) pairs of eigenvalues with opposite signs'' is a natural generalization of the \(Y(q)\) condition to quadratic manifolds of higher codimension.
0 references
Kohn Laplacian
0 references
local solvability
0 references
nilpotent Lie group
0 references
tangential Cauchy-Riemann complex
0 references
hypoellipticity
0 references
0 references
0 references
0 references