Canonical rings of surfaces whose canonical system has base points (Q3147674)

\textit{F. Enriques}, in Chapter VIII of his famous book [Le Superficie Algebriche, Zanichelli Editore (1949; Zbl 0036.37102)], posed many questions about minimal surfaces of general type with \(p_g=4, K^2=7\) and their moduli. \textit{C. Ciliberto} [Duke Math. J. 48, 121--157 (1981; Zbl 0468.14011)], was the first to describe an irreducible Zariski open set of the moduli space of minimal surfaces with \(p_g=4, K^2=7\), and constituted by surfaces with birational canonical morphism whose image \(\Sigma\) has only ordinary singularities. Later on, the second author [in: Algebraic geometry, Proc. int. Conf., Bucharest/Rom. 1982, Lect. Notes Math. 1056, 68--111 (1984; Zbl 0557.14019)] and \textit{F. Zucconi} [Osaka J. Math. 34, No. 2, 411--428 (1997; Zbl 0923.14023)] analyzed two of the possible cases. A very precise description of all the surfaces with \(p_g=4, K^2=7\) was given by the first author [Surfaces with \(K^2=7\) and \(p_g=4\). Mem. Am. Math. Soc. 721 (2001; Zbl 0985.14016)], where it was shown that the moduli space \(M_{K^2=7,p_g=4}\) has three irreducible components of respective dimensions \(36,36,38\) (two of them consist of surfaces with non birational canonical map). The challenging open problem for \(K^2=7\) is to completely understand the structure of the moduli space, i.e. to determine the incidence correspondence of the several strata which are described in the classification. (An approach is to look at the deformations of the canonical rings for surfaces with \(p_g=4, K^2=7\), as proposed by \textit{M. Reid} [Problem 12, Part 2, Problemes ouverts, Séminaire de Mathématiques Supérieures, Univ. Montréal (1980) and Proc. of Alg. Geometry mini Symposium, Tokyo Univ. Dec., 1--22 (1989)] where the method was clearly exposed.) \textit{F. Catanese} [loc. cit.] introduced the method of quasigeneric canonical projections, which allows in principle to describe the canonical rings of surfaces of general type. The method works under the assumption that the surface admits a morphism to a \(3\)-dimensional projective space which is a projection of the canonical map, and is birational to its image \(\Sigma\). In this paper, the authors are motivated by the following question: what happens when the canonical system has base points, in particular in the case of surfaces with \(p_g=4, K^2=7\)? The first aim of the paper is to introduce a general method to calculate the canonical ring of minimal surfaces of general type whose canonical system has base points but gives a birational canonical map. Then the method is applied to compute the canonical rings of surfaces \(S\) with \(p_g=4, K^2=7\), whose canonical system has only one base point and the canonical map is a birational map onto a sextic in \({\mathbb P}^3\). Using the computer algebra program Macaulay 2, the authors give three different descriptions of the above canonical rings. With these presentations, the authors are able to deform explicitly the canonical ring of a minimal surface with \(p_g=4, K^2=7\) and with birational canonical map onto a sextic in \({\mathbb P}^3\) to the canonical ring of a minimal surface with the same invariants but with base point free canonical system. The existence of such deformations were plausible since in the above cited monograph it is proved that surfaces with \(p_g=4, K^2=7\) such that the canonical system gives a birational map onto a sextic in \({\mathbb P}^3\) form an irreducible family of dimension \(35\) in their moduli space \(M_{K^2=7,p_g=4}\) and therefore they cannot dominate an irreducible component of the moduli space which, in any point, has dimension at least \(10\chi - 2K^2=36\). To reach their aim, the authors put the canonical ring of a minimal surface \(S\) with \(p_g=4, K^2=7\) and with birational canonical map onto a sextic in an antisymmetric and extrasymmetric format: i.e. they get that the canonical ring \(R(S)\) of \(S\) is represented as a quotient of the weighted graded ring \({\mathbb C}[y_{0},y_{1},y_{2},y_{3},w_{0},w_{1},u]\), where \(\deg(y_{i},w_{i},u)=(1,2,3)\), by an ideal which is generated by the \(4 \times 4\)-pfaffians of an antisymmetric and extrasymmetric \(5 \times 5\) matrix. Then they are able to construct a \(1\)-parameter flat family of rings \(R_{t}\), where \(R_{t}\), for \(t\neq 0\), corresponds to a surface with \(p_g=4, K^2=7\) and \(| K| \) base point free, while for \(t=0\) they obtain \(R(S)\). Finally they compare the above \(1\)-family of deformations with the one predicted by Enriques showing that it is a completely different type of degeneration. The paper ends with two Macaulay \(2\) scripts (without output) that they need to make explicit calculations. For the entire collection see [Zbl 0989.00069].