Elements of rigid geometry. Vol. I. Construction and geometric study of rigid spaces (Q710335)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Elements of rigid geometry. Vol. I. Construction and geometric study of rigid spaces |
scientific article |
Statements
Elements of rigid geometry. Vol. I. Construction and geometric study of rigid spaces (English)
0 references
19 October 2010
0 references
The book under review deals with ``rigid geometry'', i.e., analytic geometry over the \(p\)-adics or, more generally, over a non-archimedean valued field~\(K\). Let us recall that the first definition of rigid spaces was given by \textit{J.~Tate} in 1961 (see [Invent. Math. 12, 257--289 (1971; Zbl 0212.25601)]) in an analytic way, using maximal spectra of rings of convergent power series (called ``affinoid spaces''). Later, \textit{M.~Raynaud} [Bull. Soc. Math. Fr., Suppl., Mém. 39-40, 319--327 (1974; Zbl 0299.14003)] explained that any rigid space~\(X\) can be obtained as some kind of ``generic fiber'' of a formal scheme~\(\mathfrak{X}\) over~\(R\) (the valuation ring of~\(K\)) and that such a model is unique up to admissible (i.e., centered in the special fiber) blow-ups. This approach made the machinery of formal schemes available in the setting of rigid geometry, leading to slick proofs of many results. It was further documented in papers by \textit{S.~Bosch} and \textit{W.~Lütkebohmert} [Math. Ann. 295, No. 2, 291--317 (1993; Zbl 0808.14017); Math. Ann. 296, No.3, 403--429 (1993; Zbl 0808.14018)] and \textit{S.~Bosch, W.~Lütkebohmert} and \textit{M.~Raynaud} [Math. Ann. 302, No.1, 1--29 (1995; Zbl 0839.14013); Invent. Math. 119, No. 2, 361--398 (1995; Zbl 0839.14014)]. In A.~Abbes' book, Raynaud's approach to rigid geometry via formal schemes is presented in a systematic way and extended to relative situations. The first chapter mainly contains results of commutative algebra that will be used in the sequel. Let us mention that some part of Grothendieck's terminology is modified to make it suitable also in the non-noetherian case (for example, adic rings are required to have an ideal of definition of finite type). A.~Abbes introduces the notion of ``idyllic ring'' on which his theory is based: an adic ring is called idyllic if it is either noetherian or topologically finitely presented over a valuation ring of height~1. Those rings are proved to satisfy nice properties such as the Artin-Rees condition. The second chapter deals with formal geometry. An idyllic formal scheme is defined to be an adic formal scheme locally isomorphic to Spf(\(A\)), where~\(A\) is an idyllic ring. (Once again, Grothendieck's definition of adic formal scheme needs to be slightly modified in the non-noetherian case.) Over such a space, the classical results are shown to hold: the structure sheaf is coherent, the pushforward of a coherent sheaf by a proper morphism is still coherent, there are comparison results between the formal and algebraic theory including algebraicity results (Grothendieck's existence theorem),\dots. The third chapter introduces an essential tool that will be used in the definition of rigid spaces: an admissible blow-up of an adic formal scheme~\(\mathfrak{X}\) is the blow-up of an open finite type ideal in~\(\mathfrak{X}\). Next comes the notion of a rigid point of an idyllic formal scheme~\(\mathfrak{X}\): it is an affine subscheme of~\(\mathfrak{X}\) whose ring is a 1-valuative order (a local idyllic domain of dimension~1 whose topology is not discrete, e.g. \(\mathrm{Spec}\mathbb{Z}_{p})\)). Any finite type morphism \(f : \mathfrak{X} \to \mathfrak{Y}\) induces a map \(\langle \mathfrak{X} \rangle \to \langle \mathfrak{Y} \rangle\) between the sets of rigid points. When~\(f\) is an admissible blow-up, it is a bijection. Another useful concept is the notion of rigid closure. If~\(\mathfrak{X}\) is an idyllic formal scheme with a coherent ideal of definition~\(\mathcal{I}\), the rigid closure of an \(\mathcal{O}_{\mathfrak{X}}\)-module~\(\mathcal{F}\) is \[ \mathcal{H}^0_{\mathrm{rig}}(\mathcal{F}) =\varinjlim\mathcal{H}om(\mathcal{I}^{n},\mathcal{F}). \] One should think of it as the module of sections on the generic fiber. Any adic morphism \(f : \mathfrak{X} \to \mathfrak{Y}\) between idyllic formal schemes induces a map \[ \mathcal{H}^0_{\mathrm{rig}}(\mathcal{F}) \to f_{*}(\mathcal{H}^{0}_{\mathrm{rig}}(f^*\mathcal{F})). \] When~\(f\) is an admissible blow-up of finite presentation and~\(\mathcal{F}\) is coherent, it is an isomorphism. This statement corresponds to Tate's acyclicity theorem in rigid geometry, claiming in particular that the presheaf of sections on affinoid domains is a sheaf. Indeed, as explained in Raynaud's paper, the operation of blowing-up an open ideal in a formal scheme corresponds to passing to an affinoid subcover inside its generic fiber. The notion of rigid space is introduced in chapter 4. Let us consider the category of quasi-compact idyllic formal schemes with locally finitely presented morphisms. The localization~\textbf{R} of this category by admissible blow-ups is the category of coherent rigid spaces, christened as Raynaud's category. The localization functor is denoted \(\mathfrak{X} \mapsto \mathfrak{X}^{\mathrm{rig}}\). Some properties of the morphisms in~\textbf{R} (being an immersion, finite or proper, for instance) are defined by requiring that a model satisfies the corresponding property. The basic properties of those morphisms are established. Let us add a few words about relative discs and annuli, which are the examples given in the book. Let~\(\mathcal{S}\) be an idyllic formal scheme with an invertible ideal of definition~\(\mathcal{I}\). Let~\(\mathcal{D}\) be the affine formal line over~\(\mathcal{S}\) with parameter~\(T\) and~\(\mathcal{D}'\) the blow-up of the ideal \((T)+\mathcal{I}\). The formal disc~\(\mathcal{D}_{\mathcal{I}}\) of radius~\(\mathcal{I}\) (resp. annulus~\(\mathcal{C}_{\mathcal{I}}\) of thickness~\(\mathcal{I}\)) over~\(\mathcal{S}\) is defined to be the maximal open subset of~\(\mathcal{D}'\) where \((T)+\mathcal{I} = \mathcal{I}\) (resp. \((T)+\mathcal{I} = (T)\)). Their images by the localization functor are the closed disc of radius~\(\mathcal{I}\) and the closed annulus of thickness~\(\mathcal{I}\) over~\(\mathcal{S}^{\mathrm{rig}}\). More explicit descriptions are given when~\(\mathcal{S}\) is the formal spectrum of an idyllic ring. The next important step is to define a topology on rigid spaces. An admissible covering of a coherent rigid space~\(X\) is a family \((X_{i} \to X)_{i\in I}\) of open immersions for which there exists a finite subset~\(J\) of~\(I\) such that the induced map \(\bigsqcup_{i\in J} \langle X_{j} \rangle \to \langle X \rangle\) is surjective. One defines in this way a topology on the category~\(\mathrm{\mathbf Ad}_{/X}\) of open immersions \(U \to X\). The topos of sheaves of sets on this site is denoted~\(X_{\mathrm{ad}}\) and called the admissible topos of~\(X\). Then a careful study of the topos~\(\mathfrak{X}^{\mathrm{rig}}_{\mathrm{ad}}\), for~\(\mathfrak{X}\) a quasi-compact idyllic formal scheme, is carried out. It is described as a fibered topos over the category of admissible blow-ups, where the fiber over an object \((\mathfrak{X}',\varphi)\) is the Zariski topos of~\(\mathfrak{X}'\) and over a morphism~\(f\) the functor~\(f^*\). From this result, one can deduce that the points of the topos~\(\mathfrak{X}^{\mathrm{rig}}_{\mathrm{ad}}\) are exactly those of the Zariski-Riemann space of~\(\mathfrak{X}\), i.e. \(\varprojlim |\mathfrak{X}'|\), where~\(|\mathfrak{X}'|\) denotes the underlying space of~\(\mathfrak{X}'\) and the limit is taken over all admissible blow-ups of~\(\mathfrak{X}\). This is reminiscent of \textit{R.~Huber}'s results [Math. Z. 212, No.3, 455--477 (1993; Zbl 0788.13010)]. Next, using the construction~\(\mathcal{H}^0_{\mathrm{rig}}\) described above, A.~Abbes associates to any sheaf~\(\mathcal{F}\) on~\(\mathfrak{X}\) a sheaf~\(\mathcal{F}^{\mathrm{rig}}\) on~\(\mathfrak{X}^{\mathrm{rig}}_{\mathrm{ad}}\). From this, he gets a ringed topos \((\mathfrak{X}^{\mathrm{rig}}_{\mathrm{ad}},\mathcal{O}_{\mathfrak{X}}^{\mathrm{rig}})\) and then a ringed topos \((X_{\mathrm{ad}},\mathcal{O}_{X})\) for any coherent rigid space~\(X\). He is then able to prove the results that one expects: noetherianity of the stalks the local rings at rigid points, coherence of the structure sheaf, coherence of the \(R^q f_{*} F\) when \(f : X \to Y\) is proper and \(F\)~coherent. Some other results need more hypotheses. A coherent rigid space~\(X\) is called affinoid if it is of the form Spf(\(A\)), where~\(A\) is an idyllic ring with a monogeneous ideal of definition. In this case, he manages to show that any coherent sheaf~\(F\) is generated by its global sections and that \(H^q(X_{\mathrm{ad}},F)=0\) for any \(q\geq 1\). Chapter 5 is devoted to the study of flatness along the ideas of \textit{M. Raynaud} and \textit{L. Gruson} [Invent. Math. 13, 1--89 (1971; Zbl 0227.14010)]. It is first carried out in the setting of idyllic formal schemes using a technique of ``dévissage''. Next, the author turns to the rigid case. He proves several results, including the fact that a flat morphim between coherent rigid spaces has a flat model and that flatness for coherent sheaves is stable by base change. This chapter ends with results on faithfully flat descent for coherent modules (see also \textit{S.~Bosch} and \textit{U.~Görtz} [J. Reine Angew. Math. 495, 119--134 (1998; Zbl 0884.14009)]). Chapter 6 deals with the differential properties of coherent rigid spaces. The notions of smooth, étale, etc. morphisms are introduced and shown to satisfy the expected properties (the jacobian criterion for instance). In the last chapter, general (i.e., non-necessarily quasi-compact) quasi-separated rigid spaces are introduced. They are defined as sheaves on the big admissible topos~\(\tilde{\mathbf R}\) associated to~\textbf{R} that admit an admissible covering by objects of~\textbf{R}. More concretely, they correspond to ind-representable ind-objets of~\textbf{R} with open immersions as transition maps. The relative open unit disc is given as an example. The book ends with a study of analytification. Let~\((S,T)\) be an idyllic pair with~\(S\) coherent (the basic example being \(S=\mathrm{Spec}(A)\) and \(T=V(I)\) where~\(A\) is an idyllic ring and~\(I\) an ideal of definition of finite type). To any scheme~\(V\) of finite type over \(U=S-T\), one can functorially associate a quasi-separated rigid space~\(V^{\mathrm{an}}\) over~\(\mathcal{S}^{\mathrm{rig}}\), where~\(\mathcal{S}\) denotes the formal completion of~\(S\) along~\(T\). This GAGA functor preserves several properties of morphisms and induces a flat morphism of toposes. For a proper morphism \(f: W\to V\) and a coherent sheaf~\(F\) on~\(W\), the natural morphism \[ (R^q f_{*} F)^{\mathrm{an}} \to R^qf_{*}^{\mathrm{an}}(F^{\mathrm{an}}) \] is also proved to be an isomorphism, as one expects. To finish, let us mention that the author announces that a second volume dealing with the theory of étale cohomology will follow.
0 references
rigid geometry
0 references
formal schemes
0 references
flatness
0 references
rigid topos
0 references