Hom stacks (Q813274)

For an affine noetherian scheme \(S\) over an excellent Dedekind domain, let \({\mathcal X}\) and \({\mathcal Y}\) be two algebraic stacks over \(S\). Then the Hom 2-functor \({\mathcal HOM}({\mathcal X},{\mathcal Y})\) is a contravariant 2-functor from the category of affine noetherian schemes over \(S\) to the 2-category of groupoids defined by \({\mathcal HOM}({\mathcal X},{\mathcal Y})(T): \text{Hom}_T({\mathcal X} \times_ST,{\mathcal Y}\times_ST)\), where the right hand side stands for the groupoid of 1-morphisms. In the paper under review, the author proves the following representability theorem for the functor \({\mathcal HOM}({\mathcal X},{\mathcal Y})\): If \(X\) is proper and flat over \(S\) and \(Y\) is of finite representation over \(S\), then the 2-functor \({\mathcal HOM} ({\mathcal X},{\mathcal Y})\) is even an Artin stack. i.e., an algebraic stack for which the diagonal is representable and locally of finite type. This theorem somehow extends a recent result of \textit{M. Olsson} [Duke Math. J. 134, No. 1, 139--164 (2006; Zbl 1114.14002)], which states that if \(X\) is a proper flat algebraic space and \(Y\) is a separated algebraic space of finite type, then the corresponding functor \({\mathcal H}om(X,Y)\) is representable by an algebraic space. Moreover, if \(X\) and \(Y\) are quasi-projective schemes, then \({\mathcal HOM}(X,Y)\) is also a quasi-projective scheme. However, Olsson's method of proof involves Hilbert schemes and his special ``Quot functor'' [\textit{M. Olsson} and \textit{J. Starr}, Commun. Algebra 31, No. 8, 4069--4096 (2003; Zbl 1071.14002)], both of which are not applicable to the present situation of Artin stacks. Therefore the author proves his main theorem in a different way. Actually, using his recent previous results on deformations of algebraic stacks [\textit{M. Aoki}, Compos. Math. 141, No. 1, 19--34 (2005; Zbl 1071.14004)], he manages to verify \textit{M. Artin}'s representability criterion [Invent. Math. 27, 165--189 (1974; Zbl 0317.14001)] directly under the given assumptions. This involves a deeper study of the deformation theory of 1-morphisms of Artin stacks, a generalization of L. Illusie's classical work on the cotangent complex in algebraic deformation theory, and a verification of some related Schlessinger conditions. Then, after a brief discussion of the possible quasi-separatedness of the Artin stack \({\mathcal HOM}({\mathcal X},{\mathcal Y})\), the author gives a concrete application of his main result to the Picard 2-functor \({\mathcal P}ic_X\) of an algebraic stack \(X\) over \(S\) [à la \textit{G. Laumon} and \textit{L. Moret-Bailly}, ``Champs algébriques (Ergebnisse der Mathematik und ihrer Grenzgebiete, 3. Folge 39, Berlin: Springer) (2000; Zbl 0945.14005)]. More precisely, the author's second main theorem establishes the fact that the Picard stack \({\mathcal P}ic_{\mathcal X}\) is an Artin stack if \({\mathcal X}\) is proper and flat over \(S\). This may be seen as a generalization of M. Artin's classical result on algebraic spaces.