Total positivity, Schubert positivity, and geometric Satake (Q296664)

Let \(G\) be a simple, simply connected complex algebraic group split over the reals and denote by \(G^\vee\) its \textsl{Langland} dual group over the complex numbers: this is the complex connected reductive group whose root datum is dual to that of \(G\) and so, for instance, \(\mathrm{GL}_n({\mathbb C})\) is the Langland dual of itself and \({\mathbb P} \mathrm{GL}_n({\mathbb C})\) is the Langland dual of \(\mathrm{SL}_n({\mathbb C})\). The paper under review, to put it dramatically, may be viewed as the composition of a fictitious argue among three algebraic geometric notions of positivity, defined on the centralizer of a principal nilpotent \(X\subset G^\vee\), by showing that there is no argue at all. Indeed, the main result of this important paper, Theorem 7.1., shows that the three following notions of Schubert positivity all coincide: a) that emerging from the Peterson's work [``Quantum cohomology of \(G/P\)'', Lecture Course, MIT, Spring Term (1997)], b) Lusztig's total positivity [\textit{G. Lusztig}, ``Total positivity in reductive groups'', Prog. Math. 123, 531--568 (1994; Zbl 0845.20034)] and c) the Mirković-Vilonen positivity as shaped in their paper ``Geometric Langlands duality and representations of algebraic groups over commutative rings'' [\textit{I. Mirković} and \textit{K. Vilonen}, Ann. Math. (2) 166, No. 1, 95--143 (2007; Zbl 1138.22013)]. The latter is obtained by playing with the now called Mirković-Vilonen cycles in the affine Grassmannian \(\mathrm{Gr}_G\): the latter is, morally, is the quotient of trivializable \(G\)-bundle over the formal punctured disk \(\mathrm{Spec}({\mathbb C}((t)))\) disk and the trivializable \(G\)-bundles over the formal disc \(\mathrm{Spec}({\mathbb C}[[t]])\). Indeed, the positivity one was concerned at item a) is based on the Ginzburg and Peterson isomorphism \(H_*(\mathrm{Gr}_G)\cong O(X)\), where \(O(X)\) denotes the coordinate ring of \(X\). The aforementioned main Theorem 7.1., heavily relies on work by Peterson, whose theory is detailed accounted in Section 6 of the paper, after reviewing facts regarding the affine Grassmannian together with the geometric Satake isomorphism and the quantum cohomology of homogeneous varieties as in Section 5. To give the flavor of the Satake isomorphism in a concrete familiar example the reader may think the isomorphism between the homology of the ordinary Grassmannian \(G(k,n)\) with the \(k\)-the exterior power of the homology of the projecive space \({\mathbb P}^{n-1}\). The second, rather spectacular, result, Theorem 7.3, proves that the total positive part \(X_{>0}\) and the totally non-negative part \(X_{\geq 0}\) of \(X\) are homeomorphic to \({\mathbb R}^n_{>0}\) and to \({\mathbb R}^n_{\geq 0}\) respectively. Originally conjectured by Lam, who proved that it is true for groups of type A, the paper offers the proof for all \(G\), disregarding the Lie type. Unavoidably rather technical, this rich article is suited for readers acquainted with several modern and sophisticated notions in the algebra and combinatorics of homogeneous varieties together with the last hits in the realm of representation theory of linear algebraic groups. It subdivides itself into thirteen sections and one appendix, that preceds an abundant reference list. A few initial preliminaries and notation listed in Section 2 are followed by Section 3, aimed to recall some basics on the notion of totally positivity. The statements of the main results and of a conjecture (numbered 7.2) concerning the totally positive part of \(X\), are collected in Section 7. Conjecture 7.2 is proven for group of Lie type \(A\) and \(C\) in the appendix, where it is shown that the quantum Schubert positivity implies the affine Schubert positivity in type \(C\). The proofs of the main theorems stated in Section 7 are spread over Sections from 8 to 12. The focus of Section 13 is on the proof of a technical Proposition, numbered 3.3, establishing a sufficient condition for the total positivity of an element \(y\) of the set of real points of the unipotent radical \(U_+\), via a case by case analysis according the Lie type of the group \(G\).