4
$\begingroup$

If $f:X\to S$ is a universal homeomorphism, is $f':X\times_S X\to X$ always a nil-immersion? This seems to be easy, yet possibly I miss something. Should I give references to this fact in a paper?

$\endgroup$
4
  • $\begingroup$ what is a universal homomorphism? $\endgroup$ Commented Jul 16, 2010 at 8:50
  • 1
    $\begingroup$ Do you mean "homeomorphism"? $\endgroup$
    – S. Carnahan
    Commented Jul 16, 2010 at 11:49
  • $\begingroup$ Sorry! Certainly, I wanted to say "homeomorphism". $\endgroup$ Commented Jul 16, 2010 at 12:27
  • 3
    $\begingroup$ Mikhail, your terminology seems non-standard. I agree it's obvious that the map becomes an isomorphism on underlying reduced schemes, but it need not be an immersion and hence should not be called a nil-immersion (which I take to mean "closed immersion defined by quasi-coherent ideal sheaf whose local sections are nilpotent"). One already sees this for a non-trivial finite purely inseparable extension of fields (or any other finite flat radiciel covering of degree > 1). $\endgroup$
    – BCnrd
    Commented Jul 16, 2010 at 12:54

1 Answer 1

5
$\begingroup$

In general, the answer is no. As Brian Conrad points out in the comments above, purely inseparable field extensions do not have this property. However (and maybe this is what you were getting at?), it is true that if $X\to S$ is a universal homeomorphism, then the diagonal $X\to X\times_SX$ is a nilimmersion.

$\endgroup$
0

You must log in to answer this question.

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