No label defined (Q4786575)

This paper, which is based on the paper by \textit{S. Tanno} [Trans. Am. Math. Soc. 314, 349-379 (1989; Zbl 0677.53043)] deals with certain differential equations on a contact Riemannian manifold whose almost CR structure is not integrable, in general. Let \((M,\eta)\) be a contact manifold, i.e., a real \((2n+1)\)-dimensional smooth manifold endowed with a 1-form \(\eta\) such that \(\Psi=\eta \wedge (d\eta)^n\) is a volume form on \(M\). There is a unique tangent vector field \(\xi \in \chi(M)\) so that \(\eta(\xi)=1\) and \(\xi\rfloor d\eta=0\) (the characteristic direction of \((M,\eta)\)). It is well known that there exist a Riemannian metric \(g\) and a (1,1)-tensor field \(\phi\) on \(M\) such that \(g(X,\xi)=\eta(X)\), \(\phi^2=-I+\eta \otimes \xi\), and \(g(X,\phi X)=(d\eta)(X,Y)\), for any \(X, Y \in \chi(M)\). Such \(g\) is referred to as associated to \(\eta\) and \({\mathcal M}(\eta)\) will denote the set of all associated Riemannian metrics; clearly, \((\phi, \xi, \eta, g)\) is a contact metric structure on \(M\). Observe that each \(g \in {\mathcal M}(\eta)\) has the same volume form \(\Psi\). S. Tanno considered the Dirichlet energy \[ E(g)=\int_M \|{\mathcal L}_{\xi}g\|^2 dv_g, \quad g \in {\mathcal M}(\eta) \] and showed that an associated metric \(g \in {\mathcal M}(\eta)\) is a critical point of \(E\) if and only if \(T(g)=0\), \(T\) being the so called Tanno operator \[ T(g)=\nabla_{\xi}{\mathcal L}_{\xi}g-2({\mathcal L}_{\xi}g)\dot \phi. \] The first main result of the paper under review is an extension of Tanno's result and reads as follows. Let \((M,\eta)\) be a \((2n+1)\)-dimensional compact contact manifold and \(f:R \to R\) a real analytic function. Set \[ E_f(g)=\int_M L(g)\Psi, \quad L(g)=f(\|{\mathcal L}_{\xi}g\|^2) \quad g \in {\mathcal M}(\eta). \] Let \(F(\rho)=f''(\rho)\rho+1/2f'(\rho)\) and set \(Z(F)=\{ x \in R: F(x)=0 \}\). Then either (i) \(Z(F)=R\), and then the Lagrangian \(L(g)\) is proportional to \(\|{\mathcal L}_{\xi}g\|\), or (ii) \(Z(F) \subset R\) and then the associated metric \(g \in {\mathcal M}(\eta)\) is a critical point of \(E_f\) if and only if \(\xi(\|{\mathcal L}_{\xi}g\|^2)=0\) everywhere on \(M\) and \(T(g)=0\) at all points \(x \in M\) where \(f'(\|{\mathcal L}_{\xi}g\|_x^2)\neq 0\). As for the first variation, it is quite interesting the following. Let \((M,\eta)\) be a compact contact manifold. An associated metric \(g \in {\mathcal M}(\eta)\) is a critical point of \(E_f(g)=\int_M f(\rho) \Psi\) if and only if \(f''(\rho) \xi (\rho) {\mathcal L}_{\xi}g+f'(\rho)T(g)=0\). The authors show that the sublaplacian \(\Delta_H\), also introduced by Tanno, is subelliptic of order 1/2, hence hypoelliptic and has a discrete spectrum going to \(+ \infty\). They build a Lorentzian metric \(G_\eta\) on the total space of a certain principal \(S^1\)-bundle \(\pi:F(M) \to M\) over a contact manifold \((M,\eta)\), so that \(G_\eta\) is the Fefferman metric provided that the almost CR structure on \(M\) is integrable. Finally, it is shown that a smooth mapping \(f:M \to N\) from a contact Riemannian manifold \((M,\eta)\) in a Riemannian manifold \((N,g')\) satisfies the partial differential equation system \(\Delta_H f^i + 4g^{\alpha \bar \beta}\xi_{\alpha}(f^j)\xi_{\bar \beta}(f^k)({\Gamma'}^i_{jk}\circ f)=0\) if and only if the vertical lift \(f \circ \pi\) of \(f\) is a harmonic map with respect to the generalized Fefferman metric \(G_\eta\). Summarizing, this is a very nice and well written paper that should be read by anyone interested in contact manifolds.