The universal property of a localization characterizes the functors out of it, but applications also require control over its categorical structure and its mapping animae. We address these two problems in turn. We first determine when terminal objects and cartesian products survive localization. We then study localizations admitting adjoints, whose fully faithful adjoints identify the localized category with a category of local objects. Finally, a calculus of fractions gives explicit formulas for mapping animae and pointwise Kan extensions.

We fix a universe of small categories \(\Cat\). Given a category \(C\) and a subcategory \(W\hookrightarrow C\), we say localization of \(C\) at \(W\) for localization at the collection \(\Map([1],W)\) of morphisms in \(W\). We usually denote it by \(L(C,W)\), and by \(L(C)\) when \(W\) is understood. The notation \(C[W^{-1}]\) from Definition 3.3.6 emphasizes the universal target in which the morphisms of \(W\) have become invertible. The notation \(L(C,W)\) instead keeps both inputs visible and is convenient when treating localization as a construction functorial in the pair \((C,W)\); the two notations refer to the same category.

The first two sections extract structural consequences from the universal property of localization and from the theory of Bousfield localizations in Definition 4.4.9. The final section uses Yoneda density and pointwise Kan extensions from the preceding chapters to pass from the universal property to explicit calculations.

Sections