Let $ f: \mathbb{P}^n \dashrightarrow (\mathbb{P}^1)^n , [x_0:\dots :x_n] \mapsto ([x_0,x_1],[x_0,x_2], \dots ,[x_0,x_n])$ a birational map.

In particular, if $X$ is the blow-up of $\mathbb{P}^n$ at $r+n-1$ points and $X'$ is the blow-up of $(\mathbb{P}^1)^n$ at $r$ points. Consider $\phi : X \dashrightarrow X'$ the birational map that makes the diagram with the blow-up's and $f$ commute, is $\phi$ a Crepant birational map?

  • $\begingroup$ Maybe you don't believe me, but I still contend that $f$ cannot be a birational map. It is projective: factor as the graph of $f$ followed by the projection onto the second factor. $f$ being separated, means that the graph of $f$ is a closed immersion. Thus, $f$ is projective and therefore proper. Kontsevich proved that if $X$ and $Y$ admit a proper birational map, then the hodge numbers must be the same. In the previous question, I show that the hodge numbers are not the same: mathoverflow.net/questions/123700/crepant-birational-map/… $\endgroup$ Mar 6 '13 at 7:14
  • $\begingroup$ Even if you believe $f$ is a birational morphism, since blow-ups are projective, we have that $X$ has the same hodge polynomial of $\mathbb{P}^n$ and $X'$ has the same hodge polynomial of $(\mathbb{P}^1)^n$, but I have already computed these hodge polynomials in the last question. They are not equal. Therefore, there is no crepant resolution of $X \rightarrow X'$ by Batyrev's or Kontsevich's theorem because $H(X,u,v) \neq H(X',u,v)$. $\endgroup$ Mar 6 '13 at 8:00
  • 1
    $\begingroup$ @Andrew: doesn't $f$ restrict to an isomorphism between the obvious open subsets isomorphic to $\mathbb{A}^n$? $\endgroup$
    – quim
    Mar 6 '13 at 8:22
  • $\begingroup$ yes, this is true, $f$ is birational, but it is not universally closed. I made an error. $\endgroup$ Mar 6 '13 at 8:38

Suppose for the moment that $n=2$. Then the birational map $f\colon \mathbb{P}^n\dashrightarrow(\mathbb{P}^1)^n$ described in the question, which is $f\colon [x:y:z] \dashrightarrow([x:y],[x:z])$ has two base-points, $[0:0:1]$, $[0:1:0]$ and the inverse is $f^{-1}\colon ([a:b],[c:d]) \dashrightarrow[ac:bc:ad]$ and has one base-point, namely $([0:1],[0:1])$. If $X\to \mathbb{P}^2$ is the blow-up of $[0:0:1]$, $[0:1:0]$ and $X'$ is the blow-up of $(\mathbb{P}^1)^2$ at $([0:1],[0:1])$, the corresponding map $X \dashrightarrow X'$ is an isomorphism, and thus a "crepant" birational map. If you want to blow-up more points on $X$ and $X'$, which corresponds via this isomorphism you again get an isomorphism. If you blow-up other points you get a birational map which is not an isomorphism and thus not crepant (between two smooth projective surfaces crepant=isomorphism).

For $n>2$, the birational map $f$ has more that only isolated base-points, so you need to blow-up more complicated stuff, then contract, and maybe blow-up again... and can get a sequence of simple birational maps (blow-ups or inverse of blow-ups) from $\mathbb{P}^n$ to $(\mathbb{P}^1)^n$. In any case, you cannot get morphisms if you only blow-up points of $\mathbb{P}^n$.

  • $\begingroup$ @Jérémy: just a LaTeX suggestion: if you use $\dashrightarrow$ it looks much better. Cheers! $\endgroup$ Mar 6 '13 at 11:53
  • $\begingroup$ of course, yes, segre embedding is a blow up of one point of $\mathbb{P}^2$. The exception divisors below should line up in the formula I presented. $\endgroup$ Mar 6 '13 at 12:03
  • $\begingroup$ @Sándor Kovács. Normally I use $\dasharrow$ and it works in LaTeX but not on this website, thanks for the hint. I fixed it. You can also use \ mapstochar\ dashrightarrow in fact. @Andrew: By Segre embedding you probably mean from $\mathbb{P}^1\times \mathbb{P}^1$ to the quadric in $\mathbb{P}^3$. But in this case it is the blow-up of two pts of $\mathbb{P}^2$, followed by the contraction of the line passing through the two points. $\endgroup$ Mar 6 '13 at 12:11
  • $\begingroup$ In fact $\mapstochar$ does not work on MO. $\endgroup$ Mar 6 '13 at 12:13

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.