24
$\begingroup$

What is the automorphisms group of the weighted projective space $\mathbb{P}(a_{0},...,a_{n})$ ? Consider the simplest case of a weighted projective plane, take for instance $\mathbb{P}(2,3,4)$; any automorphism has to fix the two singular points. Consider a smooth point $p\in\mathbb{P}(2,3,4)$. What is the subgroup of the automorphisms of $\mathbb{P}(2,3,4)$ fixing $p$ ?

$\endgroup$
2
  • $\begingroup$ Will anyone answer in general what an automorphism group for a weighted projective space looks like? $\endgroup$ Sep 23, 2012 at 20:48
  • $\begingroup$ It is a toric variety, so you could try Cox' paper on the homogenous coordinate ring of a toric variety.. $\endgroup$
    – J.C. Ottem
    Sep 24, 2012 at 16:19

3 Answers 3

20
$\begingroup$

The automorphism group is the quotient of the automorphism group of the corresponding graded algebra by 1-dimensional torus acting by rescaling. In the particular case of $P(2,3,4)$ the graded algebra is $A = k[x_2,x_3,x_4]$ with $\deg x_i = i$. Note that any automorphism should take $x_2 \to a x_2$, $x_3 \to b x_3$ (since those are only elements of $A$ of degree 2 and 3) and $x_4 \mapsto c x_4 + d x_2^2$. So, the group can be written as $((k^*)^3 \ltimes k) / k^*$, where $\ltimes$ stands for the semidirect product.

$\endgroup$
5
  • 2
    $\begingroup$ I am a bit confused about this. Consider $\mathbb P(1,2)$. This should be the same as $\mathbb P^1$. So the automorphism group is $\mathbb PGL(2)$. But with the method you describe, wouldn't you get a different answer? $\endgroup$
    – mrw
    Jun 12, 2011 at 19:08
  • 4
    $\begingroup$ I consider the weighted projective space as the quotient stack of a vector space (minus zero) by the torus action. This is not the same as a quotient in the category of schemes, and so the automorphism group is different. In your example, the quotient stack is $P^1$ but with an orbifold $Z/2Z$ point. The automorphism group should preserve the point. If you compare the subgroup of $PGL(2)$ preserving a point to the group analogous to the one in my answer you will see they are the same. $\endgroup$
    – Sasha
    Jun 12, 2011 at 19:53
  • $\begingroup$ ok. Thanks for the explanation. I also wanted to mention that the OP asked a slight different. He is asking what are the automorphisms that fix a smooth fixed point p. $\endgroup$
    – mrw
    Jun 13, 2011 at 13:52
  • $\begingroup$ The weighted projective plane $\mathbb{P}(2,3,4)$ is ismorphic to $\mathbb{P}(1,2,3)$. Following your method I find different automorphisms for $\mathbb{P}(1,2,3)$ and $\mathbb{P}(2,3,4)$. Where am I wrong ? $\endgroup$
    – Puzzled
    Jul 7, 2011 at 15:03
  • 5
    $\begingroup$ They are not isomorphic. $\endgroup$
    – Sasha
    Jul 7, 2011 at 16:17
6
$\begingroup$

The group Aut(WPS) is studied in "Classes d'idéaux et Groupe de Picard des Fibrés projectifs tordus " §8.2.,8.3.

K-Theory 2 (1989),559-578 (A. AL-AMRANI)

A. Al-Amrani, May 16, 2013.

$\endgroup$
4
  • 2
    $\begingroup$ Here is a short description of Aut(WPS), in English ! Aut(WPS) is completely computed in the article above.If the WPS is Proj(S), where S is the corresponding (weighted) A-graded algebra (here A is assumed a field), then consider G = Aut. of S as grad.-alg.. An explicit description of G is given, with an A* (units of A) weighted operation on. It is shown that the natural morphism from G to Aut(WPS) induces an isomorphism between G/A* and Aut(WPS). Remark : there is a weighted projective linear description of G/A* ( a generalization of PGL(n,A)) . A.Al-Amrani. $\endgroup$
    – Al-Amrani
    May 17, 2013 at 7:15
  • $\begingroup$ About last Sasha'comment above, P(2,3,4) and P(1,2,3) are isomorphic by reduction of the weights ( proved first by Charles Delorme in "Espaces projectifs anisotropes" ). The computation of Aut(PWS) I mentioned myself is done after reduction of the weights ( which is shown to be the best possible ). $\endgroup$
    – Al-Amrani
    May 22, 2013 at 14:22
  • $\begingroup$ In the paper, did you prove the result regarding $Aut(WPS)$ only for a WPS over a field? It seems otherwise the isomorphism $f^* \mathcal{O}_{WP}(k) \to \mathcal{O}_{WP}(k)$ wouldn't be quite correct. Namely, I think it would have to be $f^* \mathcal{O}_{WP}(k) \to \mathcal{O}_{WP}(k) \otimes \mathcal{L}$ where $\mathcal{L}$ is the pullback of a line bundle on the base. $\endgroup$
    – user
    Aug 11, 2019 at 18:42
  • $\begingroup$ @user7090 :Yes, in the paper as published above, Aut (WPS) is studied over a field A. The original preprint (Strasbourg IRMA June 1983) contains a paragraph more (§8.4) giving a marticial interpretation of G/A*. It ends with an ?isomorphism? : W(X) --->Aut(WPS) over a scheme X . It is mentioned that for X=Spec(A), A factoriel ring, the morphism is an isomorphism, since Cl(WPS over A) = Z, is generated by O(1) over WPS.In the same way as over a field. If you are interested,I can send you a copy of my hole preprint. $\endgroup$
    – Al-Amrani
    Aug 15, 2019 at 16:00
2
$\begingroup$

Just to complement Al-Amrani's answer (and to self-promote a little bit): the automorphism 2-group of the weighted weighted projective stack $\mathbb{P}_S(a_1,\ldots,a_n)$ over an arbitrary base scheme $S$ has been computed, and its structure elaborated, in https://doi.org/10.1016/j.jalgebra.2017.05.002

If I am not misunderstanding something, Al-Amrani's result can be interpreted as saying that the $\pi_0$ of the 2-group of automorphisms of the stack is isomorphic to the group of automorphisms of its coarse moduli space (namely, the corresponding weighted projective space).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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