Fiber fans and toric quotients (Q878068): Difference between revisions

From MaRDI portal
Added link to MaRDI item.
RedirectionBot (talk | contribs)
Removed claim: author (P16): Item:Q596657
Property / author
 
Property / author: Diane MacLagan / rank
Normal rank
 

Revision as of 19:51, 19 February 2024

scientific article
Language Label Description Also known as
English
Fiber fans and toric quotients
scientific article

    Statements

    Fiber fans and toric quotients (English)
    0 references
    0 references
    26 April 2007
    0 references
    Let \(\pi:\mathbb Z^n\to\mathbb Z^d\) be a surjective, linear map and \(P\subseteq\mathbb R^n\) a lattice polyhedron. For each \(u\in\pi(P)\), one obtains a slice \(P_u:=\pi^{-1}(u)\cap P\) -- but they share only finitely many normal fans. In [\textit{L.~J. Billera} and \textit{B.~Sturmfels}, Ann. Math. (2) 135, No. 3, 527--549 (1992; Zbl 0762.52003)], this situation led to the notion of the fiber polytope being the Minkowski sum average of these slices; its normal fan \(\mathcal N(P,\pi)\) is the coarsest common refinement of the normal fans of the above slices. Since the fiber polytope construction requires compactness of \(P\), the authors of the present paper prefer dealing with the fan \(\mathcal N(P,\pi)\) directly and call it the fiber fan of \((P,\pi)\). On the other hand, the coarsest common refinement of all images \(\pi(F)\) of faces \(F\leq P\) yields a polyhedral subdivision \(\mathcal P=\mathcal P(P,\pi)\) of \(\pi(P)\subseteq\mathbb R^d\), hence a fan \(\widetilde{\mathcal P}:=\overline{\mathbb R_{\geq 0}\cdot (\mathcal P,1)}\subseteq\mathbb R^{d+1}\). If \(P\) is a polyhedral cone, then \(\mathcal P\) was already a fan, and the authors explain the Gale duality \(\mathcal P(P,\pi)=\mathcal N(P^\vee,\pi^\vee)\) with \(\pi^\vee:(\mathbb Z^n)^*\to(\ker\pi)^*\cong\mathbb Z^{n-d}\) and \(P^\vee\subseteq(\mathbb R^n)^*\) being the dual cone. For arbitrary \(P\), the authors obtain \(\widetilde{\mathcal P}(P,\pi)=\mathcal N(\widetilde{P}^\vee,\widetilde{\pi}^\vee)\) with \(\widetilde{P}:=\overline{\mathbb R_{\geq 0}\cdot (P,1)}\) and \(\widetilde{\pi}:\mathbb Z^{n+1}\to\mathbb Z^{d+1}\). Geometrically, \((P,\pi)\) translates into a \(d\)-dimensional subtorus action on the \(n\)-dimensional toric variety \(X_P\). In this context, \(\mathcal P\) is known to be the GIT chamber system. On the other hand, generalizing the compact case treated in [\textit{M.~M. Kapranov, B.~Sturmfels} and \textit{A.~V. Zelevinsky}, Math. Ann. 290, No. 4, 643--655 (1991; Zbl 0762.14023)], the authors show that the toric variety associated to \(\mathcal N(P,\pi)\) equals the normalization of the so-called toric Chow quotient \(X_P^{\text{ch}} (\mathbb C^*)^d\). The latter is defined as the canonical component of the inverse limes of the GIT quotients.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    toric variety
    0 references
    GIT chambers
    0 references
    Chow quotient
    0 references