A simply connected surface of general type with \(p_g=0\) and \(K^2=2\) (Q2464701): Difference between revisions

Surfaces of general type with geometric genus \(p_g=0\) (and thus also irregularity \(q=0\)) have been studied by algebraic geometers for a very long time and plenty of examples have been constructed, but at present a classification seems still out of reach. A very interesting and hard question concerning these surfaces is the construction of simply connected examples, which are of great interest also in the study of differentiable four-manifolds. The first such example was constructed by \textit{R. Barlow} [Invent. Math. 79, 293--301 (1985; Zbl 0561.14015)] and up to now was the only known simply connected surface of general type with \(p_g=0\). The classical methods for constructing surfaces with \(p_g=0\) are, using the terminology of Miles Reid, the Campedelli method, which consists in taking the minimal model of the desingularization of a double cover of the plane branched on a very special, highly singular, curve, and the Godeaux method, which consists in taking the quotient of a canonical surface by the free action of a finite group \(G\). By construction, a Godeaux type example is never simply connected, since its fundamental group surjects onto \(G\). Barlow's surface, which has \(K^2=1\), is obtained by a variation of the Godeaux construction, in which the group \(G\) has some isolated fixed points. However, neither method seems to be useful in producing new simply connected examples and it has long been an open question whether there exist simply connected surfaces of general type with \(p_g=0\) and \(K^2>1\). The paper under review contains the construction of a simply connected minimal surface of general type with \(p_g=0\) and \(K^2=2\), and also of two simply connected minimal surfaces of general type with \(p_g=0\) and \(K^2=1\). The examples are constructed by a new method, namely as smoothings of singular rational surfaces. The construction of the main example (\(K^2=2\)) goes as follows: one takes a certain pencil of cubics in the plane and blows up its base locus, obtaining an elliptic rational surface with a particular configuration of singular fibers. Then one blows up further and obtains a rational surface containing 5 disjoints chains of rational curves, which can be blown down to get a singular rational surface \(X\). Every singularity of \(X\) is of class \(T\), namely it admits a local \(\mathbb{Q}\)-Gorenstein smoothing. Using deformation theory, the authors prove that there is indeed a global \(\mathbb{Q}\)-Gorenstein smoothing of \(X\), namely a \(1\)-parameter family \({\mathcal X}\to\Delta\) of projective surfaces such that: -- the central fiber is \(X\); -- the general fibre \(X_t\) is smooth and projective; -- the relative canonical divisor \(K_{X/\Delta}\) is \(\mathbb{Q}\)-Cartier. Then it is not difficult to show that the general fibre \(X_t\) is a minimal surface of general type with \(K^2=2\) and \(p_g=0\). Finally, one shows that \(X_t\) is simply connected by using standard arguments on Milnor fibers. One should understand that the construction is not only very technical but really subtle and ingenious.