Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $V$ be a $\Bbbk$-variety such that $\Bbbk^\times$ (as an algebraic group) acts algebraically on $V$. Given any $f\in\Bbbk[V]$, let us call $f$ homogeneous of degree $d$ if for all $v\in V$ and all $\lambda\in\Bbbk^\times$, we have $f(\lambda.v)=\lambda^d f(v)$.

My question is: Does this define a grading on $\Bbbk[V]$?

I was convinced that it is true, but I am running into difficulties. Let us first assume $\Bbbk=\mathbb{C}$, the ground field should not be an obstruction. The linear span of $\Bbbk^\times f$ decomposes since $\Bbbk^\times$ is reductive, but I don't see how to turn this into a grading on all of $\Bbbk[V]$.

If it is true, I would really like to see a proof - it should use as little machinery as possible.

share|improve this question
2  
If $k$ is algebraically closed then the action of $k^*$ identifies with an action of the torus $G_{m,k}$. This is a diagonalisable group scheme and therefore any action of $G_{m,k}$ on a $k$-algebra corresponds to a $\bf Z$-grading (it does not have to be finite dimensional over $k$). See for instance SGA 3, I, 4.7.3. –  Damian Rössler Aug 15 '12 at 11:00
1  
@Jesko: Please see my final comment to your earlier question related to this matter; Damian's SGA3 reference is precisely Ben's computation, carried out over any ring. Note also that if you already believed that "the linear span of $k^{\times}f$ decomposes since $k^{\times}$ is reductive" (as you say above) then you are done, since it would imply that every element lies in a finite-dimensional $k^{\times}$-stable subspace and so the span of any two such would be similarly exhausted in this way (hence graded, etc.). So that case done rigorously contains all of the difficulties. –  user22479 Aug 15 '12 at 14:19
    
If you imagine that K∗ is like a circle, it will have $π_1=Z$. Then you have have Z "act on" V. This "action" could be grading. I'm am not sure if this is non-sense from a coincidence or their is something to this. I am not even sure it this can be made precise. –  Spice the Bird Aug 16 '12 at 8:01
add comment

1 Answer

up vote 7 down vote accepted

Turning the action map of varieties into a map of rings, we get a ring map $\phi$ from $k[V]$ to $k[V][t,t^{-1}] $, the coordinate ring with an extra invertible variable (the coordinate on $k^*$) adjoined. Now, for any function $\phi(f)=\sum_{i\in \mathbb{Z}}f_it^i$ for some $f_i$'s, almost all of which are 0. Note that $f=\sum f_i$, which we obtain by restricting the function to $t=1$. Using associativity, applying $\phi$ again to the $f_i$'s is the same as applying pull-back by the multiplication map to t. Thus, as functions on $V\times k^*\times k^*$ (letting $t,u$ be the two coordinates)

$$\sum_{i\in \mathbb{Z}}\phi(f_i)u^i=\sum_{i\in \mathbb{Z}} f_i t^i u^i$$

since the pull-back of the coordinate by multiplication is just the product of the coordinates . Thus, $\phi(f_i)=f_it^i$.

We can define the grading by letting $f$ be homogeneous of degree $i$ if $\phi(f)=ft^i$. We have already seen that every element can be written uniquely as a sum of such elements (the $f_i$'s), and this is multiplicative since $\phi$ is a ring homomorphism.

Alternatively, we can note that we have proven that the span of the $f_i$'s is an finite-dimensional invariant subspace containing $f$, so we can apply your argument. In general, essentially the same argument shows that the action of any affine algebraic group on the coordinate ring of any affine variety by pull-back is a locally finite action: any function is contained in a finite-dimensional invariant subspace.

share|improve this answer
1  
Wonderful! This is just perfect. –  Jesko Hüttenhain Aug 15 '12 at 11:47
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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