Collapsed toposes and cartesian closed varieties (Q910487)

In a previous paper [Algebra Univers. 21, 198-212 (1985; Zbl 0599.18003 )] the author gave syntactic conditions on an algebraic theory T which characterized when the category T-\underbar{Alg}, of T-algebras, is a topos. The article under review provides the answer to the naturally arising related question of determining when T-\underbar{Alg} is cartesian closed. The author proves that every cartesian closed variety of algebras arises from a topos via a certain ``collapsing'' construction (the two-valued collapse) which associates to a topos its full subcategory of well-supported objects. As was done in the earlier paper, a syntactic characterization in terms of the operations of the theory T is also given. If T is a commutative hyperaffine theory, then T- \underbar{Alg} will be cartesian closed, but it is not a topos. The article also discusses how to recover a topos from its two-valued collapse and shows that the two-valued collapse of a Grothendieck topos \({\mathcal E}\) is a variety of algebras iff \({\mathcal E}\) has a projective set of generators.