Log concavity of the Grothendieck class of \(\overline{\mathcal{M}}_{0,n}\) (Q6952886)

In this paper, the authors study the moduli space \(\overline{\mathcal{M}}_{0,n}\) of stable genus-\(0\) curves with \(n\) marked points, for \(n \ge 3\). Their main result concerns an asymptotic log-concavity property for the Poincaré polynomials of these spaces. They also observe that these polynomials are \(\gamma\)-positive. Taken together, these results provide evidence for a broader conjecture claiming that the Poincaré polynomials of \(\overline{\mathcal{M}}_{0,n}\) have only real zeros.\N\NMore preciesly, the authors work with the class of \(\overline{\mathcal{M}}_{0,n}\) in the Grothendieck group of varieties \(K(\mathrm{Var}_{\mathbb{C}})\). It is known that the class of \(\overline{\mathcal{M}}_{0,n}\) can be expressed as a polynomial with integer coefficients in the Lefschetz-Tate class \(\mathbb{L} = [\mathbb{A}^1]\), such a class is denoted by\N\[\N[\overline{\mathcal{M}}_{0,n}] = a_{n,0} + a_{n,1}\mathbb{L} + \cdots + a_{n,n-3}\mathbb{L}^{\,n-3}.\N\]\NThe Poincaré polynomial is obtained by specializing \(\mathbb{L}\) to \(t^2\). Notice that \(\mathcal{M}_{0,n}\) has only even cohomology, and that the integers \(a_{n,k}\) may be interpreted as the ranks of the cohomology groups of \(\overline{\mathcal{M}}_{0,n}\). Log-concavity of the associated polynomial is thus the statement that\N\[\Na_{n,i}^2 \;\ge\; a_{n,i-1}\,a_{n,i+1} \qquad\text{for all } i \ge 1 \text{ and all } n \ge 3.\N\]\NThe stronger notion of ultra-log-concavity requires the inequality\N\[\N\frac{a_{n,i}}{\binom{n-3}{i}}^{\!2} \;\ge\; \frac{a_{n,i-1}}{\binom{n-3}{i-1}} \cdot \frac{a_{n,i+1}}{\binom{n-3}{i+1}} \qquad\text{for all } i \ge 1,\; n \ge 3.\N\]\NThe main theorem is that the coefficients of the Grothendieck class of \(\overline{\mathcal{M}}_{0,n}\) -- equivalently, the Betti numbers \(a_{n,k} = \mathrm{rk}\, H^{2k}(\overline{\mathcal{M}}_{0,n},\mathbb{Q})\) -- satisfy an asymptotic log-concavity property. Precisley, we have:\N\N\textbf{Theorem.} For every \(i \ge 1\) there exists an \(N\) such that for all \(n \ge N\),\N\[\N\bigg(\frac{a_{n,i}}{\binom{n-3}{i}}\bigg)^{\!2} \;\ge\; \frac{a_{n,i-1}}{\binom{n-3}{i-1}} \cdot \frac{a_{n,i+1}}{\binom{n-3}{i+1}}.\N\]\NMoreover, we have that the class \([\overline{\mathcal{M}}_{0,n}]\) in the Grothendieck group \(K(\mathrm{Var}_k)\), for any algebraically closed field \(k\), is determined by the following recursive formula. We have \([\overline{\mathcal{M}}_{0,3}] = 1\). For \(n > 3\), the classes satisfy\N\[\N[\overline{\mathcal{M}}_{0,n}] = [\overline{\mathcal{M}}_{0,n-1}](1 + \mathbb{L}) + \mathbb{L}\sum_{i=3}^{n-2} \binom{n-2}{\,i-1\,}\,[\overline{\mathcal{M}}_{0,i}]\,[\overline{\mathcal{M}}_{0,n+1-i}].\N\]\NRecall that a univariate polynomial\N\[\Nf(t) = \sum_i a_i t^i\N\]\Nis called \textit{symmetric} with \textit{center} \(d/2\) if \(a_{d-i} = a_i\) for all \(i\). Any such polynomial \(f(t) \in \mathbb{Z}[t]\) with center \(d/2\) can be uniquely expanded in the form\N\[\Nf(t) = \sum_{i=0}^{\lfloor d/2 \rfloor} \gamma_i \, t^i (t+1)^{\,d - 2i},\N\]\Nfor uniquely determined integers \(\gamma_i\). A symmetric polynomial \(f\) is said to be \textit{\(\gamma\)-positive} if all of the coefficients \(\gamma_i\) are nonnegative.\N\NOur interest in this notion stems from the following well-known fact that every real-rooted symmetric polynomial with nonnegative coefficients is \(\gamma\)-positive.\N\NThus, \(\gamma\)-positivity provides indirect evidence for real-rootedness.\N\N\textbf{Theorem.} For every \(n \ge 3\), the polynomial \(P_n(t) \in \mathbb{Z}[t]\) defined by\N\[\N[\overline{\mathcal{M}}_{0,n}] = P_n(\mathbb{L})\N\]\Nis \(\gamma\)-positive.