Let $\phi: A \to B$ an inclusion of commutative rings, $I \subset A$ an ideal of $A$, $J \subset B$ an ideal of $B$.
There are two natural operations on the ideals of $A$ and $B$, contraction $J^c := J \cap A = \phi^{-1}(J)$ and extension $I^e:= \phi(I) \cdot B$.
Trivially we have relations $(J^c)^e \subset J$ and $I \subset (I^e)^c$. Natural question is when these inclusions are even equalities?
The 2nd case has rather satisfying answer: Namely there is a rather nice class of ring extensions for which $I = (I^e)^c$ if $\phi:A \to B$ always holds: the faithfully flat extensions, see eg here.
#EDIT: But attention, as Martin Brandenburg indicated in a comment beeing faithfully flat is not equivalent to have equality $I = (I^e)^c$ for all ideals of $A$, but still it's a "nice" class of ring extensions which carry this feature. This also motivates:
Question: What about first case, under which non trivial conditions on $\phi$ holds $(J^c)^e = J$?
Are there some interesting class(es) of ring extensions which have this property in similar spirit faithfully flat extensions share the property forcing $I = (I^e)^c$.
For exactly, there are for example the "boring" case of quotients $B=A/I$. Another trivial example as Lukas Heger remarked in comments are field extensions. So beeing injective is not an exclusive criterion.
But can one say about which common features such classes of ring extensions would consequently share which imply this property in nearly similar vein/analogy as in case of faithfully flat extensions towards $I = (I^e)^c$, even though faithfully flat is stronger then having this property?
EDIT#:( As Lukas Heger remarked this part is wrong; I don't want to delete it completely as this at least inducates what I roughly looking for when I'm asking for their common features; these should play the role analogous to that faithfully flat extensions play with respect to condition $I = (I^e)^c$ ) [Asking differently, can it for example be showed that if $\phi$ is injective, the inclusion $(J^c)^e \subset J$ can never be an equality if $\phi$ is not an iso? (This would also confirm the intuition that quotient maps $A \to A/I$ can never be faithfully flat except $I =0$; In that sense that quotient maps are "opposite" to faithfully flat maps). Philosophy: This would give dual notion of faithfully flat maps. This question generalizes this one. ]