The existence of algebraic surfaces with preassigned Chern numbers (Q909737)

It is well-known that there is a unique minimal smooth model in each birational class of complex algebraic surfaces of general type. The Chern numbers \(c^ 2_ 1(S)\) and \(c_ 2(S)\) are two fundamental discrete invariants of a smooth \(surface\quad S\). By Noether's formula \(c^ 2_ 1(S)+c_ 2(S)=12\chi ({\mathcal O}_ S)\) and by the equality \(c^ 2_ 1(S)=K^ 2_ S\), where \(K_ S\) is the canonical divisor of S, the Chern numbers \((c^ 2_ 1(S),c_ 2(S))\) are equivalent to the discrete invariants \((K^ 2_ S,\chi ({\mathcal O}_ s))\). By abuse of language, sometimes \((K^ 2_ S,\chi ({\mathcal O}_ S))\) are referred to as the Chern numbers of the \(surface\quad S\). The Chern numbers of a minimal surface of general type are restricted by the inequalities \(2\chi\) (\({\mathcal O}_ S)-6\leq K^ 2_ S\leq 9\chi ({\mathcal O}_ S)\) and \(K^ 2_ S({\mathcal O}_ S)>0\). It is natural to ask the following question of existence: For any pair of positive integers (a,b) satisfying the inequality 2a-6\(\leq b\leq 9a\), can we find a minimal surface S of general type with \(K^ 2_ S=b\), \(\chi\) (\({\mathcal O}_ S)=a?\) The main result of this article is the following statement: Theorem. For every pair of positive integers (a,b) satisfying 2a-6\(\leq b\leq 9a-347\), there exists a minimal surface S of general type with \(K^ 2_ S=b\), \(\chi\) (\({\mathcal O}_ S)=a.\) Thus there remains only the border region to be filled in. Historically it is \textit{U. Persson} [cf. Compos. Math. 43, 3-58 (1981; Zbl 0479.14018)] who first raised this question and established in 1981 the existence of minimal surfaces with Chern numbers satisfying \(2\chi\) (\({\mathcal O}_ S)-6\leq K^ 2_ S\leq 8\chi ({\mathcal O}_ S)\); there remains only the case \(8a<b\leq 9a\). The strategy of Persson is simple but effective. First he established a lattice of invariants, then by imposing singularities he perturbed these surfaces to fill in the gaps. In this paper we follow the same strategy, but instead of inducing singularities by coverings, we impose singularities by quotients. To this end, first we make a cyclic base extension, then take the quotient with respect to a cyclic group of automorphisms so that the quotient surface has Hirzebruch-Jung singularities. After desingularization, the Chern numbers are disturbed. By manipulating the cyclic covering of the base curve and the twist of the fibres, every needed minimal surface can be thus constructed.