Wondertopes (Q6899274)

Positive geometries, introduced in [\textit{N. Arkani-Hamed} et al., J. High Energy Phys. 2017, No. 11, Paper No. 39, 124 p. (2017; Zbl 1383.81273)], are pairs \((X, X_{\geq 0})\) of a complex projective variety and a semialgebraic subset of its real points, equipped with a unique meromorphic canonical form determined recursively by residues along boundary components. Since a positive geometry \((X, X_{\geq 0})\) naturally defines a pair of a variety and a divisor, it is natural to ask whether it admits a log resolution to another positive geometry \((\widetilde{X}, \widetilde{X}_{\geq 0})\) where \(\widetilde{X}\) is smooth and the boundary is a simple normal crossing divisor. This paper answers the question affirmatively for the prototypical case of a polytope \(\mathcal{P}\) in projective space \(\mathbb{P}V\) by introducing a new family of positive geometries called \textit{wondertopes}.\N\NThe construction proceeds as follows. Given a polytope \(\mathcal{P} \subset \mathbb{P}V\) and a building set \(\mathcal{B}\) of linear subspaces satisfying three conditions -- (1) each \(F \in \mathcal{B}\) meets \(\mathcal{P}\) in a face, (2) \(\mathcal{B}\) contains all facet hyperplanes, and (3) \(\mathcal{B}\) is a building set in the sense of De Concini-Procesi -- the authors form the wonderful compactification \(X_{\mathcal{B}}\) by sequentially blowing up the elements of \(\mathcal{B}\), and define the wondertope \(\widetilde{\mathcal{P}}_{\mathcal{B}}\) as the Euclidean closure of the preimage of the interior of \(\mathcal{P}\). The main result (Theorem 1.4) states that \((X_{\mathcal{B}}, \widetilde{\mathcal{P}}_{\mathcal{B}})\) is a positive geometry whose canonical form equals the pullback \(\pi_{\mathcal{B}}^{*}\Omega(\mathbb{P}V, \mathcal{P})\). Since the boundary of \(X_{\mathcal{B}}\) is simple normal crossing by \textit{C. De Concini} and \textit{C. Procesi} [Sel. Math., New Ser. 1, No. 3, 459--494 (1995; Zbl 0842.14038)], this provides the desired log resolution. The proof first establishes a ``fundamental computation'' for blowing up a single face (Theorem 3.1) and then proceeds by induction on the building set. A notable corollary recovers the result of \textit{N. Arkani-Hamed} et al. [SIGMA, Symmetry Integrability Geom. Methods Appl. 17, Paper 092, 41 p. (2021; Zbl 1484.13049)] that \((\overline{M}_{0,n+1}, (\overline{M}_{0,n+1})_{\geq 0})\) is a positive geometry with the Parke-Taylor form, by specializing to the braid arrangement with an appropriate building set. Several non-examples demonstrate that each of the three conditions on \(\mathcal{B}\) is necessary. In an appendix, the authors develop combinatorial analogues of the geometric constructions in the language of matroids and prove a product decomposition for nested set complexes of building sets (Theorem 1.7).