Fano manifolds with nef tangent bundles are weakly almost Kähler-Einstein (Q1642920)

Let \(X\) be a complex projective manifold of dimension \(n\) such that the anticanonical class \(c_1(T_X)\) is ample and the tangent bundle \(T_X\) is nef. A well-known conjecture of Campana and Peternell claims that in this situation the manifold \(X\) is rational homogeneous, i.e. is a quotient \(G/P\) of a linear algebraic group by a parabolic subgroup. It is known that rational homogenous manifolds always carry a Kähler-Einstein metric, so from an analytic point of view it is tempting to conjecture that Fano manifolds with nef tangent bundle are Kähler-Einstein. In this paper the author proves a weak version of this conjecture: denote by \(\text{PC}^{1,1}(X) \cap c_1(X)\) the set of positive currents of bidegree \((1,1)\) in the anticanonical class. Then there exists a family of smoothing operators \(J_\varepsilon\) mapping elements \(T \in \text{PC}^{1,1}(X) \cap c_1(X)\) to a smooth closed positive definite \((1,1)\)-form \(J_\varepsilon(T) \in c_1(X)\) such that \(J_\varepsilon(T)\) converges weakly to \(T\) for \(\varepsilon \to 0\). Moreover for \(\varepsilon \in ]0,1]\) there exists a Kähler metric \(\omega_\varepsilon\) on \(X\) such that \(\text{Ricci}(\omega_\varepsilon)=J_\varepsilon(\omega_\varepsilon)\). The construction of \(J_\varepsilon\) and the ``weakly almost Kähler-Einstein'' metric \(\omega_\varepsilon\) is not explicit and follows from the Schauder fixed point theorem. \newline The family of metrics \((\omega_\varepsilon)_{\varepsilon \in ]0,1]}\) is defined to be well-behaved in \(L^p\)-norm if \[ \int_X (\rho_{n, \varepsilon}-\rho_{1, \varepsilon}) \omega_\varepsilon^n \] converges to \(0\) for \(\varepsilon \to 0\), where \(\rho_{n, \varepsilon}(x)\) (resp. \(\rho_{1, \varepsilon}(x)\)) are the highest (resp. lowest) eigenvalue of \(\text{Ricci}(\omega_\varepsilon)\) at a point \(x \in X\). For \(p \geq 1\) this property implies the semistability of \(T_X\), for \(p \geq 2\) it implies the Chern class inequality \[ [n c_1(X)^2 - (2n+2) c_2(X)] \cdot c_1(X)^{n-2} \leq 0. \] The author raises a number of questions that could lead to further progress on the Campana-Peternell conjecture.