The concept of restriction category is an element-free formalisation of the idea of a category of partial maps, where the domain of a map $f\colon X\to Y$ is encoded in a specified idempotent endomorphism $\overline{f}$ of $X$.
Given a category $C$, a restriction structure consists of the assignment to each morphism $f\colon X\to Y$ a morphism $\overline{f}\colon X\to X$ satisfying the following conditions:
where $s(f) = \text{dom}(f)$ is the domain of $f$. A restriction category is a category with a restriction structure.
Note that a restriction structure is a structure in the technical sense of the word, not a property. Note that it follows from the above definition that $\overline{f}$ is idempotent for composition: $\overline{f}\circ \overline{f} = \overline{f}$, and that the operation $f \mapsto \overline{f}$ is also idempotent: $\overline{\overline{f}} = \overline{f}$ (among other properties).
Every category admits the trivial restriction structure, with $\overline{f} = id_{s(f)}$.
Conversely the wide subcategory consisting off all the objects together with the morphisms $f$ satsfying $\overline{f}=id_{s(f)}$—the total morphisms—has a trivial induced restriction structure.
The category $PF$ of sets and partial functions is the prototypical example, where for a partial function $X \supseteq U \stackrel{f}{\to} Y$, the partial endomorphism $\overline{f}$ is the partially-defined identity function $X \supseteq U \hookrightarrow X$. Many other examples are listed in (Cockett–Lack 2002)
Robin Cockett, Steve Lack, Restriction categories I: categories of partial maps, Theoretical Computer Science 270 (2002) pp 223–259 (author pdf)
Robin Cockett, Steve Lack, Restriction categories II: partial map classification (web)
Robin Cockett, Steve Lack, Restriction categories III: colimits, partial limits, and extensivity (arXiv:math/0610500)
Last revised on November 3, 2021 at 07:15:31. See the history of this page for a list of all contributions to it.