Forcing and classifying topoi

From MaRDI portal





The notion of classifying topos goes back, in certain special cases, to Grothendieck, who for instance showed how the topos of continuous actions of a Galois group Gal(\=k/k) classifies separable closures of the field k. And to Hakim, who proved e.g. that the Zariski topos classifies local rings [\textit{M. Hakim}, Topos annelés et schémas relatifs (1972; Zbl 0246.14004)]. A systematic theory of classifying toposes could only be formulated after the advent of categorical logic, in particular sheaf semantics. The fragment of (infinitary first order intuitionistic) logic preserved by left exact left adjoint functors between toposes was called geometric logic, due to the geometric significance of these functors, and it was shown in the mid 70's that every theory formulated inside this fragment has a classifying topos, or a generic topos model (Joyal, Reyes, Benabou, Tierney,...). The present work gives a good exposition of how the study of geometric theories and their classifying toposes provide a unified method for forcing constructions and independence results. Crucial in any specific such construction is to find out something more about the generic model than its genericity - in particular to find out some of its non-geometric properties. The author gives a substantial list of such known extra information, and its use. We mention two striking examples (both due to Joyal): 1) The topos of simplicial sets classifies the notion of interval. This leads to results for the geometric realization functor [cf. \textit{P. Johnstone}, Proc. Lond. Math. Soc., III. Ser. 38, 237-271 (1979; Zbl 0402.18006)]. 2) A topos in which the internal real unit interval is not compact: This can be obtained by constructing the generic covering of the external real unit interval by an ideal of open subsets, each of which has Lebesgue- measure \(<frac{1}{2}\), and thus admitting no finite subcover. The non- geometric extra property to be proved here is that the internal and external reals are the same. Most of the examples deal, like this, with intuitionistic analysis, or with set theory (continuum hypothesis, Suslin hypothesis,...).











This page was built for publication: Forcing and classifying topoi

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3320376)