Threefolds with big and nef anticanonical bundles. II (Q657295)

Let \(X\) be a smooth complex projective variety whose anticanonical bundle \(-K_X\) is big and nef but not ample (\(X\) is then called an \textit{almost Fano threefold}). Then there exists a birational contraction \(\psi: X \to X'\), given by some multiple \(-mK_X\). In the case when the Picard number \(\rho(X) = 2\) and \(\psi\) is divisorial, there exists a complete list of such \(X\), due to first part [the authors, Math. Ann. 333, No. 3, 569--631 (2005; Zbl 1081.14054)]. In the paper under review, the authors continue their research, assuming now that \(\rho(X) = 2\) and \(\psi\) is small, i.e., \(\psi\) has only \(1\)-dimensional fibers. Namely, the authors also carry out a classification of the latter \(X\), assuming that the complementary extremal contraction on \(X\) (corresponding to the second extremal ray of the Mori cone \(\overline{\text{NE}}(X)\subset\mathbb{R}^2\)) is not birational. The study of possible types of \(X\) relies on the two main ingredients (cf. Section 2 in the text). The first one is the \textit{\(D\)-flop} \(\chi: X \dashrightarrow X^{+}\) w.\,r.\,t. the contraction \(\psi\) and (any) effective divisor \(D\) which is not \(\psi\)-nef. One should notice that \(X^{+}\) is also an almost Fano threefold, with \(\rho(X^{+})=2\), \((-K_X)^3 = (-K_{X^{+}})^3\), and \(h^0(\mathcal{O}_X(D)) = h^0(\mathcal{O}_{X^{+}}(\psi_*(D)))\). The second (ingenious) ingredient is the fact that \(X' = \psi(X)\) admits a smoothing, for \(X'\) is a Fano threefold with terminal Gorenstein singularities (see \textit{Y. Namikawa} [J. Algebr. Geom. 6, No. 2, 307--324 (1997; Zbl 0906.14019)]), i.e., there exists a flat family \(\mathcal{X} \to \Delta\) over a unit disc, where the fiber \(\mathcal{X}_0 = X'\) and the fibers \(\mathcal{X}_t\), \(t\neq 0\), are smooth Fano threefolds. It follows from [\textit{P. Jahnke} and \textit{I. Radloff}, Math. Z. 269, No. 3, 1129--1136 (2011; Zbl 1248.14016)] that also \(\text{Pic}(\mathcal{X}_t) = \text{Pic}(\mathcal{X}_0)\) (as \(\mathbb{Z}\)-lattices). The proof then goes along straightforward (yet quite lengthy) case-by-case analysis of possible \(X,~X^{+}\), etc., using the classification of smooth Fano threefolds (see \textit{V. A. Iskovskikh} and \textit{Yu. G. Prokhorov} [Encycl. Math. Sci. 47, 1--245 (1999; Zbl 0912.14013)]) and hyperelliptic and trigonal Fano threefolds with canonical Gorenstein singularities (see \textit{V. V. Przhiyalkovskij}, \textit{I. A. Cheltsov} and \textit{K. A. Shramov} [Izv. Math. 69, No. 2, 365--421 (2005); translation from Izv. Ross. Akad. Nauk, Ser. Mat. 69, No. 2, 145--204 (2005; Zbl 1081.14059)]).