Rigidity of four-dimensional gradient shrinking Ricci solitons (Q6137562)

The paper under review deals with shrinking gradient Ricci solitons, that is, complete Riemannian manifolds \((M,g)\) satisfying \[ \mathrm{Ric}(g)+\nabla^{2}f=\lambda g, \] where \(\lambda>0\) and \(f\in C^{\infty}(M)\). Such metrics are important as generalisations of Einstein metrics (which are recovered by setting \(f\) to be constant) as well as in the theory of Ricci flow, where they appear as singularity models. In dimensions \(2\) and \(3\), there is a complete classification of such metrics and a rather small list of possible geometries. Using the metrics on this list as a model, \textit{P. Petersen} and \textit{W. Wylie} [Pac. J. Math. 241, No. 2, 329--345 (2009; Zbl 1176.53048)] defined a rigid gradient soliton to be one which is isometric to a quotient of \(N\times_{\Gamma}\mathbb{R}^{k}\) where \(N\) is an Einstein manifold and \(\Gamma \) is a group acting freely on \(N\) and by orthogonal transformations on \(\mathbb{R}^{k}\). While it is known that there exist non-rigid shrinking gradient solitons in dimensions greater than three, it is interesting to try to characterise rigid solitons through their geometric properties. For example, in the compact case, \textit{M. Eminenti} et al. [Manuscr. Math. 127, No. 3, 345--367 (2008; Zbl 1160.53031)] have shown that a gradient shrinking soliton is rigid if and only if its scalar curvature is constant. For the non-compact case, Cao has conjectured that the same result must also hold (IX Workshop on Differential Geometry 2019 in Maceió, Brazil). In dimensions \(4\), through work of \textit{M. Fernández-López} and \textit{E. García-Río} [Proc. Am. Math. Soc. 144, No. 1, 369--378 (2016; Zbl 1327.53057)] the scalar curvature \(S\) can only take the values \(0, 2\lambda, 3\lambda,\) or \(4\lambda\). In the cases where \(S=0\) or \(S=4\lambda\) it is known that the metric must be Einstein. In the case where \(S=3\lambda\) it is known that the metric is isometric to a finite quotient of \(\mathbb{S}^{3}\times \mathbb{R}\). The main result of the paper deals with the outstanding \(S=2\lambda\) case and the authors prove the following result. Let \((M,g,f)\) be a four-dimensional complete non-compact shrinking gradient Ricci soliton satisfying \[ \mathrm{Ric}(g)+\nabla^{2}f=\lambda g. \] If the scalar curvature is constant and equal to \(2\lambda\) then \((M,g)\) must be isometric to a finite quotient of \(S^{2}\times \mathbb{R}^{2}\). Combined with the results above, this confirms Cao's conjecture in dimension \(4\). The proof uses some inequalities involving the quantity \(\Delta_{f} (\mathrm{tr}(\mathrm{Ric^{3}}))\) to demonstrate that the Ricci tensor must be of rank \(2\). From this fact it follows that the soliton must be isometric to \(\mathbb{S}^{2}\times \mathbb{R}\).