User loginNavigation |
The Two Dualities of Computation: Negative and Fractional Types
The Two Dualities of Computation: Negative and Fractional Types
Abstract Every functional programmer knows about sum and product types, a+b and a*b respectively. Negative and fractional types, a-b and a/b respectively, are much less known and their computational interpretation is unfamiliar and often complicated. We show that in a programming model in which information is preserved (such as the model introduced in our recent paper on Information Effects), these types have particularly natural computational interpretations. Intuitively, values of negative types are values that flow “backwards†to satisfy demands and values of fractional types are values that impose constraints on their context. The combination of these negative and fractional types enables greater flexibility in programming by breaking global invariants into local ones that can be autonomously satisfied by a subcomputation. Theoretically, these types give rise to two function spaces and to two notions of continuations, suggesting that the previously observed duality of computation conflated two orthogonal notions: an additive duality that corresponds to backtracking and a multiplicative duality that corresponds to constraint propagation. By Greg Buchholz at 2014-05-21 16:59 | LtU Forum | previous forum topic | next forum topic | other blogs | 7393 reads
|
Browse archives
Active forum topics |
Recent comments
16 weeks 5 days ago
16 weeks 6 days ago
16 weeks 6 days ago
39 weeks 6 hours ago
43 weeks 2 days ago
44 weeks 6 days ago
44 weeks 6 days ago
47 weeks 4 days ago
1 year 14 hours ago
1 year 16 hours ago