Slope stability and exceptional divisors of high genus (Q957893)

In this very nice paper, the authors study slope stability for smooth projective surfaces. Slope stability for a couple \((M,L)\) where \(M\) is a projective manifold and \(L\) an ample line bundle, was introduced in [J. Differ. Geom. 72, No. 3, 429--466 (2006; Zbl 1125.53057)] by \textit{J. Ross} and \textit{R. Thomas}. This slope stability condition was modeled from the vector bundle case and can be seen as a particular case of \(K\)-stability induced by certain specific test configurations. It is known that the existence of a Kähler metric with constant scalar curvature in \(2\pi c_1(L)\) implies the \(K\)-stability of the manifold, which in turn implies the slope stability of the manifold. One of the main result is that a surface containing an exceptional divisor with arithmetic genus \(\geq 2\) is slope unstable for a certain polarization. In particular this shows that the induced Kähler class does not contain any Kähler metric with constant scalar curvature. The proof relies on the direct computation of the slope of a divisor \(D\) in terms of the following three intersection numbers: \(D\cdot L\), \(D\cdot D\) and \(D\cdot K_M\). Another striking result is that slope stability of surfaces can be tested with only divisors (instead of subschemes). This is done by deforming the destabilizing subscheme into a subscheme which is union of a divisor and a zero-dimensional piece. Then, in a second step, the authors show that zero-dimensional subschemes can never destabilize a surface. The authors also investigate the case of minimal surfaces. For surfaces with non-negative Kodaira dimension, a destabilizing divisor must have negative-self intersection and arithmetic genus at least two. Note that the proof requires naturally to play with the adjoint divisor \(K_M+rD\). As a by product, the authors obtain the slope stability of \((X,L)\) where \(X\) is a surface and \(L\) is a polarisation such that \(2(K_X \cdot L)L - (L^2)K_X\) is positive. It is known from the work of J. Song and B. Weinkove that under this condition, the Mabuchi functional is proper. Thus, it could be true that for minimal surfaces, slope stability is equivalent to \(K\)-stability. Finally, the authors show that a nef divisor \(D\) never destabilizes \((M,L)\). The idea is to find a blowdown of \(M\rightarrow M'\) such that \(D\) is the pullback of a divisor \(D'\) and \(K_{M'}+2D'\) is nef. Then one applies the previous results. A beautiful application is that \(\mathbb{CP}^2\) blown up at 2 point is slope stable for the anticanonical polarisation. Thus slope stability is a strictly weaker condition than \(K\)-stability (this manifold is not \(K\)-stable and does not carry a Kähler-Einstein metric but a Kähler-Ricci soliton). The paper provides many examples of possible applications and is very well written.