A nonboundary nef divisor on \(\overline{M}_{0,12}\) (Q1954186)

Let \(\overline{M}_{0,n}\) be the moduli space of stable curves of genus \(0\) with \(n\) marked points. It has a stratification where the codimension \(k\) strata consist of irreducible components of the locus of curves with at least \(k\) nodes. In particular, the \(1\)-dimensional strata are called \(F\)-curves. The \(F\)-conjecture predicts that a divisor on \(\overline{M}_{0,n}\) is nef if and only if it has nonnegative intersection with every \(F\)-curve. It has been proven for \(n\leq 7\) by Keel and McKernan, and is still open for \(n\geq 8\). Call a divisor on \(\overline{M}_{0,n}\) \(F\)-nef, if it intersects every \(F\)-curve nonnegatively. \textit{A. Gibney, S. Keel} and \textit{I. Morrison} [J. Am. Math. Soc. 15, No. 2, 273--294 (2002; Zbl 0993.14009)] stated a relevant conjecture (sometimes called Fulton's Conjecture) that every \(F\)-nef divisor on \(\overline{M}_{0,n}\) is numerically equivalent to an effective sum of boundary divisors. It has been proven for \(n\leq 7\) by \textit{P. L. Larsen} [J. Lond. Math. Soc., II. Ser. 85, No. 1, 1--21 (2012; Zbl 1251.14015)], and if true in general, it would imply the above \(F\)-conjecture by induction. In the paper under review, the author comes up with a very interesting counterexample to Fulton's Conjecture on \(\overline{M}_{0,12}\). More precisely, using the combinatorial data of the \((11,5,2)\) biplane \(\mathcal P\), the author constructs explicitly an \(F\)-nef divisor \(D_{\mathcal P}\) on \(\overline{M}_{0,12}\), which is not an effective sum of boundary divisors. Pulling back \(D_{\mathcal P}\) to \(\overline{M}_{0,n}\) provides similar counterexamples for any \(n > 12\). In addition, the author shows that \(D_{\mathcal P}\) is an extremal nef divisor and can be written as an effective sum of boundary divisors with a hypertree divisor defined by \textit{A.-M. Castravet} and \textit{J. Tevelev} [J. Reine Angew. Math. 675, 121--180 (2013; Zbl 1276.14040)].