2
$\begingroup$

Let $X$ be a set and suppose that $R$ is a binary relation on $X$. Suppose further that $S\subseteq X$ has the property that whenever $y\in S$ and $xRy$ ($x\in X$), then also $x\in S$. Does such an $S$ have a name? Of course, if $R$ happens to be a partial order (or even preorder), then $S$ is often called either a "down set", "lower set", or sometimes an "ideal" (usually one requires that the preorder be directed in the definition of ideal). But I am not aware of any terminology in the case that $R$ is assumed only to be a binary relation. If anyone can help, I would be very grateful. Thank you.

$\endgroup$
1
  • 2
    $\begingroup$ Transitive is sometimes used, especially when $R$ is set-membership $\in$. $\endgroup$ Commented Jan 8, 2014 at 2:32

3 Answers 3

3
$\begingroup$

I don't see any reason not to use the same terminology as one uses in the case of a partial order (or a pre-order), since a set $S$ is closed in your sense with respect a relation if and only if it is downward-closed with respect to the reflexive transitive closure of the relation, which is a partial pre-order.

Such sets $S$ are variously called downward closed, or down sets, or lower sets, or closed under predecessors. Sets like this are also precisely the open sets in the lower-cone topology, whose basic open sets consist of the predecessors of a single node (for the transitive closure of the relation).

$\endgroup$
1
  • $\begingroup$ I suspect there is some terminology in algebraic logic (in particular the study of cylindric and relational algebras) which has been given to such sets. Steven Givant and Ralph McKenzie are names that come to mind who might know this terminology. It might even be something like leftward closed, but I am guessing. Gerhard "Or Perhaps Prefix Closed Works?" Paseman, 2014.01.07 $\endgroup$ Commented Jan 8, 2014 at 0:06
2
$\begingroup$

Since $R$ could be a $\le$, or $\ge$, or all kinds of other possibilities, I kind of agree that "lower set" is not ideal (excuse the pun). "Leftward closed set" seems somewhat more reasonable, since no matter what $R$ is, if you write $xRy$ then you are writing $x$ on the left.

$\endgroup$
1
  • $\begingroup$ Or just left–closed, to keep the wording slightly simpler. $\endgroup$ Commented Jan 8, 2014 at 11:00
1
$\begingroup$

For the property $x \in S$ and $x \mathrel{R} y$ imply $y \in S$, I would say that $S$ is saturated by $R$. Thus, in your case, saturated by $R^{-1}$ might be an option.

$\endgroup$
2
  • $\begingroup$ Do you have a typo with $x$ and $s$? $\endgroup$ Commented Jan 8, 2014 at 12:10
  • $\begingroup$ @-david-hamkins Ooops... Thanks, corrected... $\endgroup$
    – J.-E. Pin
    Commented Jan 8, 2014 at 13:08

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .