5
$\begingroup$

There is a famous example of Macaulay which shows that there are prime ideals of height two in $\mathbb C[X_1,X_2,X_3]$ having at least $l$ generators for any $l\ge 3$.

In Macaulay's words, the example is constructed as follows: "Consider $l(l-1)/2$ straight lines through the origin $O$ in $3$-dimensional space, not lying on any cone of order $l-2$. Draw a cone of order $l$ and a surface (not a cone) of order $l$ through the $l(l-1)/2$ lines so as to intersect again in an irreducible curve of order $l(l+1)/2$ with $l(l-1)/2$ tangents at $O$. Then no basis of the prime module determined by this curve can have less than $l$ members, where $l$ is a number which can be chosen as high as we please."

I'd be grateful if someone can "translate" the above in an algebraic language.

(In this paper one can find more details and a proof, but the language is also geometric.)

$\endgroup$
3
  • $\begingroup$ Posted previously on M.SE at math.stackexchange.com/questions/1285065/… $\endgroup$
    – user26857
    May 18, 2015 at 7:05
  • $\begingroup$ Does anybody have a precise definition of "cone" and "cone of order $l$"? looks like it's of very old use, but I couldn't find a definition. $\endgroup$
    – YCor
    May 18, 2015 at 8:12
  • $\begingroup$ @YCor Maybe the linked paper of Abhyankar could shed some light, but definitely I'm not the one who can spot it. $\endgroup$
    – user26857
    May 18, 2015 at 11:17

1 Answer 1

4
$\begingroup$

There seems to be some terminology drift here. I would say that "order" would be called degree in modern terminology, for example.

Here is the way I see it, and please someone correct me if I am wrong.

Take $l(l-1)/2$ lines through the origin in $\mathbb C^3$, i.e. ideals $I_i\subset\mathbb C[X_1,X_2,X_3]$ of the form $\langle X_2-\alpha_iX_1,X_3-\beta_i X_1\rangle $ with $1\leq i\leq l(l-1)/2$ and $\alpha_i$ and $\beta_i$ generic (so that no degree $l-2$ homogeneous polynomial in $X_1,X_2,X_3$ lies in all of the above ideals $I_i$).

Consider a polynomial $F_1$ which is homogeneous of degree $l$ and is contained in each $I_i$. Containment in $I_i$ is a codimension one condition, and there is a dimension $(l+1)(l+2)/2$ space of degree $l$ polynomials, so such $F_1$ exists. I assume that such $F_1$ is chosen generically (this is the "cone").

Consider a polynomial $F_2$ which does not have to be homogeneous, but rather is a sum of polynomials of degrees $\geq l$ which is contained in all $I_i$ (and presumably is generic with respect to this property).

Then the ideal in question is the radical of $\langle F_1,F_2\rangle$.

I don't claim to know how the actual argument proceeds, but this looks like the algebraic translation you are after.

$\endgroup$
1
  • $\begingroup$ "Order" means "order of vanishing at the origin". "Cone" means stable under linear maps $(X_i\to\lambda X_i)$. $\endgroup$ May 18, 2015 at 17:04

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.