Generalized metric space

In mathematics, specifically in category theory, a generalized metric space is a metric space but without some properties, such as symmetry.[1] Precisely, it is a category enriched over , the one-point compactification of . The notion was introduced in 1973 by William Lawvere who noticed that a metric space can be viewed as a particular kind of category.

The categorical point of view is useful since by Yoneda's lemma, a generalized metric space can be embedded into a much larger category in which, for instance, one can construct the Cauchy completion of the space.

Discussion

We can view as a symmetric monoidal category as follows:[2] objects are points in , the set of morphisms between objects is

and composition is given by the sum

The tensor operation is .

This category structure is equivalent to one obtained by viewing the poset as a category in the usual way. The above definition is analogous to the following example: let be the Boolean algebra generated by some subsets of a finite set and with and with . Then is a symmetric monoidal category.

Now, let be a metric space. Then it can be viewed as a category enriched over as follows. The objects are the points of and we let . The composition for is a morphism in

and its well-definedness is exactly the triangle inequality.

Notes

  1. ↑ Namely, the property that distinct elements have non-zero distance between them and the property that the distance between two elements is always finite.
  2. ↑ Lawvere (1973), p. 145, § 1.

References

  • Bonsangue, M. M.; Van Breugel, F.; Rutten, J. J. M. M. (1998). "Generalized metric spaces: Completion, topology, and powerdomains via the Yoneda embedding". Theoretical Computer Science. 193 (1–2): 1–51. doi:10.1016/S0304-3975(97)00042-X. hdl:1887/4083537.

Further reading