Besse conjecture with positive isotropic curvature (Q2674962)

A Riemannian metric \(g\) on a smooth closed manifold \(M\) is said to be Einstein if the Ricci curvature \(\mathrm{Ric}_g\) is a constant multiple of the metric itself. The Einstein equations arise as the Euler-Lagrange equations for the Einstein-Hilbert functional \[ \mathcal{S}(g) = \int_M {\mathrm{Scal}_g\, \mathrm{dvol}_g} \] restricted to metrics of unit volume. Now, as in [\textit{A. L. Besse}, Einstein manifolds. Berlin etc.: Springer-Verlag (1987; Zbl 0613.53001), Section 4] restricts the functional \(\mathcal{S}\) to metrics of unit volume and constant scalar curvature. The Euler-Lagrange equations of the restricted functional relates the Ricci curvature of a metric \(g\) with constant scalar curvature with the Hessian of an eigenfunction \(f\) of the Laplacian with eigenvalue \(\frac{1}{n-1}\mathrm{Scal}_g\). In the paper under review this system is referred to as the Critical Point Equation (CPE). In Besse's exposition [loc. cit.; Remark 4.48] it is expected that solutions to the CPE on a closed manifold must be Einstein. The claim is trivially true for \(\mathrm{Scal}_g\leq 0\) since the eigenfunction \(f\) must then vanish. Therefore the conjecture reduces to the claim that any non-trivial solution to the CPE is a round sphere \(\mathbb{S}^{n}\) with \(f\) the restriction of a linear function on \(\mathbb{R}^{n+1}\). This remains open in general and it is only proved under additional assumptions, including prior work of the authors [\textit{G. Yun} et al., Taiwanese J. Math. 18, No. 5, 1439--1458 (2014; Zbl 1357.58017); Taiwanese J. Math. 20, No. 3, 699--703 (2016; Zbl 1357.58018)] under the assumption of harmonic curvature. In the paper under review, the authors prove that the conjecture holds when the metric is assumed to have positive isotropic curvature. The proof is divided into two main steps. One first uses the CPE and the positive isotropic curvature assumption to show that the vector field \(\mathrm{Ric}(\nabla f, \cdot)\) is parallel to \(\nabla f\). This is done in Section 5 and the curvature condition enters to guarantee positivity of the curvature term in a Bochner-Weitzenböck formula. In Sections 3 and 4 then the authors show that the only solution of CPE with \(\mathrm{Ric}(\nabla f, \cdot)\) parallel to \(\nabla f\) is the round sphere. This is done by studying the level sets of the eigenfunction \(f\).