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

Given a commutative, $\mathbb N$-graded ring, one can associate to it a scheme via the $Proj$ construction.

What happens if one tries to copy this procedure but instead of $\mathbb N$ with another indexing gadget (say commutative monoid) ?

Some thoughts about this: Considering projective varieties is roughly the same as studying affine varieties equivariant under the multiplicative group. So I would guess that replacing $\mathbb N$ by something else corresponds to replacing the multiplicative group by something else.

share|improve this question
An action of the multiplicative group on an affine variety induces a decomposition of its coordinate ring into irreps which are "polynomially defined," and this is where the N-grading comes from abstractly. So I guess one expects an action of a more general group to induce a grading by the corresponding polynomial irreps. –  Qiaochu Yuan Nov 29 '10 at 15:10
I think that using different gradings as you suggest corresponds to entering the realm of geometric invariant theory. Note that an $\mathbb{Z}$ grading on a ring is the same as an action of the multiplicative group $\mathbb{G}_m$. –  Daniel Loughran Nov 29 '10 at 15:12
Proj isn't just the grading. You also have to decide what corresponds to the the irrelevant ideal. –  arsmath Nov 29 '10 at 15:44
add comment

2 Answers

up vote 11 down vote accepted

Weighted projective spaces $\mathbb{P}(a_1,\ldots,a_n)$ are examples where a grading other than the standard grading is used. In general you can study gradings coming from any finitely generated abelian group, and this grading gives rise to a torus action on the ring. The Proj you speak of is then a GIT-quotient of $Spec R$ by this torus action (if you are familiar with GIT, the choice of linearization of the action corresponds to a choice of irrelevant ideal). This GIT-quotient construction is completely analogous to the usual construction $$ \mbox{Proj} k[x_0,\ldots,x_n]=\left(\mbox{Spec} k[x_0,\ldots,x_n]-V(x_0,\ldots,x_n)\right) // \mathbb{G}_m $$ The best reference I can give for this stuff is Chapter 1 of the book "Cox rings" by Arzhantsev, Derenthal, Hausen and Antonio Laface. Other nice references are

The Homogeneous Coordinate Ring of a Toric Variety by Cox, which deals with Toric varieties

Lectures on invariant theory by Dolgachev, which is a nice introduction to quotients in algebraic geometry.

share|improve this answer
Too bad I can't vote my answer down while voting yours up! –  Allen Knutson Nov 29 '10 at 21:55
add comment

First let's revisit the usual case. The $\mathbb N$-grading on $R$ says that $Spec\ R$ is a cone, and there is a map $R \to R_0$. We can rip out $Spec\ R_0$ from $Spec\ R$, take the quotient, and get a reasonable space $Proj\ R$.

If we grade instead by say $({\mathbb N})^k$, then there are $k$ analogous quotients of $R$, and $k$ closed subschemes to rip out before we divide by $({\mathbb G}_m)^k$. This comes up if e.g. $R = Fun(G/N)$, with the torus acting on the right (through the maximal unipotent group $N$), and the corresponding quotient is $G/B$.

Some of the things I would look up here are "Cox coordinate ring of a toric variety" and "Mori dream space", but I don't know the references well enough to suggest a best place.

share|improve this answer
add comment

Your Answer


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.