Monoidal closed categories, weak topoi and generalized logics (Q810504)

This paper seeks to develop a categorical foundation for fuzzy set theory by having a category whose internal logic, referred to by the author as a generalized logic, is based on \([0,1]\)-Lukasiewicz logic in the sense that the logical connectives of this logic appear as truth arrows in the category. This is accomplished via the notion of a weak topos. A weak topos is a monoidal closed category (however in general not cartesian closed) and admits a weak subobject classifier, which classifies extremal subobjects, although the classifying maps need not be unique. Before outlining the theory of weak topoi, the example of the category MET1 of \([0,1]\)-valued metric spaces with contraction mappings is explored. The reviewer is surprised that the author seems unaware of the seminal paper of \textit{F. W. Lawvere}, ``Metric spaces, generalized logic and closed categories'' [Rend. Sem. Mat. Fis. Milano 43, 135-166 (1973; Zbl 0335.18006)]. Lawvere's observation that metric spaces can be studied using the theory of enriched categories leads to many useful constructions. Thus, the fact that MET1 is a monoidal, closed category follows directly from this earlier work. Using MET1 as a motivating example, the paper under review proceeds to develop the axioms of a weak topos and investigates numerous consequences and strengthenings of these axioms. Then, as a further example generalizing both MET1 and sheaves on a frame, the author considers the category of complete M-valued sets, where M is what is referred to as a ``locally involutory monoid''. (There is an unfortunate typographical omission in 4.3 I2 in the definition of such a monoid.) There may well be some connections between this latter example and the work of \textit{U. Berni-Canani}, \textit{F. Borceux}, and \textit{R. Succi-Cruciani} [J. Pure Appl. Algebra 62, 123-136 (1989; Zbl 0685.18006)], where they build a fibration over Q-sets (that is Q-valued sets where Q is an idempotent quantale) whose fibers are toposes. In fact, a closer scrutiny of enriched category theory as a tool in these investigations may prove useful. In his concluding remarks, the author's supposition that the structures of metric spaces and sheaves have been heretofore unrelated is not accurate, as there is the aforementioned work of Lawvere on interpreting metric spaces as enriched categories, and sheaves can also be interpreted as enriched categories following the work of \textit{R. F. C. Walters} [J. Pure Appl. Algebra 24, 95-102 (1982; Zbl 0497.18016)].