In most case, the definition of a variety over a field $k$ at least requires that being "of finite type" and being "separated". It has no question for me that being of finite type, since we always like finite.

I donot know the reason why we require being "separated" to a variety?

There is a reason since a scheme over a field $k$ being separated will have the property that the intersection of two affine open sets is still affine open set.

Are there any other acceptant reasons?

Thanks a lot.


Separatedness is the analogue in the Zariski topology of Hausdorfness in the cohmology of a manifold or complex analytic space. Without it, strange behaviour can occur. E.g. if $Y$ is not separated, then two morphisms $f,g: X \to Y$ could coicide on a dense open subset of $X$ while not coinciding on all of $X$. Hence "analytic continuation'' is not valid for morphisms into $Y$.

Many arguments in geometry proceed by analytic continuation/Zariski density arguments, so it is natural to place oneself in a context where those arguments can be applied without reservation.


Besides what Matt E has explained, if you include the separatedness in the definition of scheme i.e. as a separated (having closed diagonal) prescheme, the analogy will be more clear in this respect that a variety is in fact, a separated prevariety where prevariety by this definition is an irreducible, reduced prescheme of finite type over an algebraically closed field $k.$ This approach is closer to the way, Mumford in Red book of varieties and schemes has adopted to define the notion of variety.

  • 2
    $\begingroup$ Right, but nowadays we do not include separatedness in the definition of a scheme: that is preschemes no longer exist. To the extent that I am qualified to have opinions on such matters, I do think the terminology "prescheme" was poor, because -- e.g. compare with "presheaf" -- it suggests that the first thing we should do is apply some completion functor to get a scheme. But in moduli theory one does (unfortunately!) sometimes encounter nonseparated schemes/stacks, and I'm pretty sure that "scheme-ifying it" is not the thing to do. $\endgroup$ – Pete L. Clark May 12 '12 at 19:10
  • $\begingroup$ @Pete L.Clark: I admit that, prescheme is old fashioned these days, but I just wanted to note that it fits nicely to the hierarchy of AG definitions. $\endgroup$ – Ehsan M. Kermani May 12 '12 at 19:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.