Savage surfaces (Q6056484)

Let \(G\) be the topological fundamental group of a given nonsingular complex projective surface. The main result of the present paper is the surprising assertion that the Chern slopes \(c_1^2/c_2\) of minimal nonsingular complex projective surfaces \(S\) having topological fundamental group isomorphic to \(G\) are dense in the interval \([1,3]\). Given a real number \(r\in [2,3]\), in [\textit{X. Roulleau} and \textit{G. Urzúa}, Ann. Math. (2) 182, No. 1, 287--306 (2015; Zbl 1346.14097)] it was shown that there exists a sequence of simply connected surfaces \(\{X_p\}\) such that \(c_1^2(X_p)/c_2(X_p)\) approaches \(r\) as \(p\to \infty\), and in the present paper this result is extended for \(r\in [1,3]\), using special configurations of curves in \(\mathbb P^2\). Note that a similar density result for \(r\in [1/5, 2]\) was proven by \textit{U. Persson} [Compos. Math. 43, 3--58 (1981; Zbl 0479.14018)] with a different type of surfaces. The sequences of surfaces \(\{X_p\}\) are fundamental for the proof of the result in the present paper, as follows. Having a minimal surface \(Y\) with fundamental group \(G\) and \(r\in [1,3]\), one takes a sequence \(\{X_p\}\) as above of simply connected surfaces. Then a new sequence of surfaces \(\{S_p\}\) with \(\pi_1(S_p)\simeq G\) and such that \(c_1^2(S_p)/c_2(S_p)\) approaches \(r\) as \(p\to \infty\) is constructed by finding for each \(p\) a suitable surface inside \(X_p\times Y\). The construction of the surfaces \(S_p\) is quite subtle. They are obtained as the intersection of two general sections of a lef line bundle \(M\) on \(X_p\times Y\), which in turn is obtained from a lef line bundle \(\Gamma_p\) in \(X_p\) and a very ample divisor \(B\) in \(Y\) (for the notion of lef line bundle see [\textit{M. A. A. de Cataldo} and \textit{L. Migliorini}, Ann. Sci. Éc. Norm. Supér. (4) 35, No. 5, 759--772 (2002; Zbl 1021.14004)]). The authors explain also that their method of construction does not work for Persson's sequences of simply connected surfaces, on one hand because it is not clear how to find the suitable lef line bundles and on the other hand because for low values of the slope there are constraints on the kind of fundamental groups that can occur. The paper finishes with two conjectures on density of the Chern slopes in \([1,3]\). One is similar to the result in the present paper but asking for canonical class ample. The other states that given a Brody hyperbolic non singular surface \(Y\), then the Chern slopes of hyperbolic non singular surfaces with topological fundamental group isomorphic to the fundamental group of \(Y\) are dense in \([1,3]\).