Sheaves in Geometry and Logic: A First Introduction to Topos Theory (Universitext) [Englisch] [Taschenbuch]

Topos theory now has applications in fields such as music theory, quantum gravity, artificial intelligence, and computer science. It has been viewed by some as being excessively abstract and difficult to learn, and this is certainly true if one attempts to learn it from the research literature. The use of this book to learn topos theory certainly puts this view to rest, as the authors have given the readers an introduction to topos theory that is crystal clear and nicely motivated from an historical point of view. Indeed the prologue to the book gives the reader a deep appreciation of the origins of the subject, and could even serve as an introduction to a class on algebraic geometry.

An understanding of sheaf theory and category theory will definitely help when attempting to learn topos theory, but the book could be read without such a background. Readers who want to read the chapters on logic and geometric morphisms will need a background in mathematical logic and set theory in order to appreciate them. Topos theory has recently been used in research in quantum gravity. A reader interested in understanding how topos theory is used in this research should concentrate on the chapter on properties of elementary topoi, the one on basic categories of topoi, and the chapter on localic topoi.

The authors introduce topos theory as a tool for unifying topology with algebraic geometry and as one for unifying logic and set theory. The latter application is interesting, especially for readers (such as this reviewer), who approach the book from the standpoint of the former. Indeed, the authors discuss a fascinating use of topos theory by Paul Cohen in his proof of the independence of the Continuum Hypothesis in Zermelo-Fraenkel set theory.

The prologue for this book is excellent, and should be read for the many insights and motivations for the subject of topos theory. The elementary category theory needed is then outlined in the next section. A "topos" is essentially a category that allows the construction of pullbacks, products, and so on, with the philosophy being that objects are to be viewed not only as things but as also having maps (functors) between them. In the section on categories of functors, this viewpoint becomes very transparent due to the many examples of categories that are also topoi are discussed. These examples are presented first so as to motivate the general definition of topos later on. Some of these categories are very familiar, such as the category of sets, the category of all representations of a fixed group, presheaves, and sheaves. Of particular interest in this section is the discussion of the propositional calculus, and its representation as a Boolean algebra. Replacing the propositional calculus with the (Heyting) intuitionistic propositional calculus results in a different representation by a Heyting algebra. From the standpoint of ordinary topology, the Heyting algebra is significant in that the algebra of open sets is not Boolean, i.e. the complement (or "negation") of an open set is closed and not open in general Instead it follows the rules of a Heyting algebra. This type of logic appears again when considering the subobjects in the sheaf category, which have a "negation" which belong to a Heyting algebra. Thus topos theory is one that follows more than not the Brouwer intuitionistic philosophy of mathematics. Recently, research in quantum gravity has indicated the need for this approach, and so readers interested in this research will find the needed background in this part of the book.

After a straightforward overview of how sheaf theory fits into the topos-theoretic framework, the authors also discuss the role of the Grothendieck topology in sheaf theory. This involves thinking of an open neighborhood of a point in a space as more than just a monomorphism of that neighborhood into the space (all the open neighborhoods thus furnishing a "covering" of the space). This need was motivated by certain constructions in algebraic geometry and Galois theory, as the authors explain in fair detail. A covering of a space by open sets is replaced by a new covering by maps that are not monomorphisms. Starting with a category that allows pullbacks, an indexed family of maps to an object of this category is considered. If for each object in this category one uses a rule to select a certain set of such families, called the coverings of the object under this rule, then ordinary sheaf theory can be used on these coverings. If one desires to drop the requirement that the category have pullbacks, this can be done by introducing a category that comes with such "covering families." This is the origin of the Grothendieck topologies, wherein the indexed families are replaced by the sieves that they generate. A Grothendieck topology on a category is thus a function that assigns to each object in the category a collection of sieves on the object (this function must have certain properties which are discussed by the authors). Several examples of categories with the Grothendieck topologies are discussed, one of these being a complete Heyting algebra. Another example discussed comes from algebraic topology, via its use of the Zariski topology for algebraic varieties. The discussion of this example is brilliant, and in fact could be viewed as a standalone discussion of algebraic geometry.

When considering the notion of the Grothendieck topology, the authors define the notion of a `site', which is essentially a (small) category along with a Grothendieck topology on the category. They then show how to define sheaves on a site, which then form a category. A `Grothendieck topos' is then a category which is equivalent to the category of sheaves on some site. The authors then show, interestingly, that a complete Heyting algebra can be realized as a subobject lattice in a Grothendieck topos.

This book is written in the best Mac Lane style, very clear and very well organized. It also benefits from Moerdijk's extensive work organizing the theory of Grothendieck toposes by elementary means. The reader should have basic graduate knowledge of algebra and topology. The book is long because it gives very explicit descriptions of many advanced topics--you can learn a great deal from this book that, before it was published, you could only learn by knowing researchers in the field.

This is a really good book to learn sheaves, it has the "tree generations" of sheaves; with topological spaces, with Grothendieck topologies, and with Lawvere-Tirney topologies. Also a great book in topoi theory, from the basics up to the Cohen topos.