I have an ideal $I$ generated by quadratic and cubic homogeneous polynomials in $10$ variables.

Macaulay2 tells me that $I$ defines an irreducible variety $X$ of dimension $5$ and degree $10$ in $\mathbb{P}^9$, and that $I$ is not radical.

When I ask for the primary decomposition of $I$ Macaulay2 gives me two ideals $I_0,I_1$. He says that $I_0$ is radical and defines an irreducible variety of dimension $5$ and degree $10$, while $I_1$ is not radical and defines a variety of dimension $-1$. Indeed, when I ask for the radical of $I_1$ Macaulay2 gives me the ideal generated by all the homogeneous coordinates of $\mathbb{P}^9$ thus corresponding to the empty set. This last fact is really confusing me.

What is then the meaning of $I_1$ in the primary decomposition of $I$?

Thank you very much in advance.

Here are the ideals I am considering

P9 = QQ[a00,a01,a02,a03,a11,a12,a13,a22,a23,a33]

F0 = a00*a12-a01*a02+a01*a13-a03*a11

F1 = a00*a23+a01*a33-a02*a03-a03*a13

F2 = a01*a22-a02*a12+a11*a23-a12*a13

F3 = a02*a23-a03*a22+a12*a33-a13*a23

F4 = a00*a22-a02^2-a11*a33+a13^2

J = ideal(F0,F1,F2,F3,F4)

M = matrix{{a00,a01,a02,a03},{a01,a11,a12,a13},{a02,a12,a22,a23},{a03,a13,a23,a33}}

I = J + minors(3,M)

L = primaryDecomposition(I)

X = variety L_0

Y = variety L_1


  • $\begingroup$ Just to have it available: faculty.math.illinois.edu/Macaulay2/doc/Macaulay2-1.12/share/… . $\endgroup$ – LSpice Oct 31 '20 at 19:38
  • $\begingroup$ In terms of the definition, it seems to me that we would expect $I = I_0 \cap I_1$ (which I think we can't judge from the information given) and $\sqrt{I_0} \ne \sqrt{I_1}$ (which is the case in your computation). I think that there is no expectation that $I_1$ itself be radical. $\endgroup$ – LSpice Oct 31 '20 at 19:42
  • $\begingroup$ I posted the ideals I am considering. What is really unclear to me is the difference between the variety defined by $I$ and the variety defined by $L_0$. $\endgroup$ – user125056 Oct 31 '20 at 19:59
  • $\begingroup$ When you post code, please make sure to format it as code (either by manually adding 4 spaces at the beginning of each line, or by using the format-as-code button in the GUI). Otherwise, Markdown gets at it and makes a mess. I have edited accordingly. $\endgroup$ – LSpice Oct 31 '20 at 20:07
  • 1
    $\begingroup$ I am sorry. Next time I will do it. $\endgroup$ – user125056 Oct 31 '20 at 20:58

Commutative algebra is NOT the same as algebraic geometry, especially projective algebraic geometry.

The variety in $\mathbb{P}^9$ defined by $I$ and the variety in $\mathbb{P}^9$ defined by $I_0$ are the same variety.

If you were to work in $\mathbb{A}^{10}$, then $I$ and $I_0$ would define different affine schemes; the scheme defined by $I$ has some extra fatness at the origin, but since the origin is not part of projective space, you don't see this difference in $\mathbb{P}^9$.

EDIT: The relevant bit of Hartshorne is Exercise II.5.10

  • $\begingroup$ Thank you very much for the answer. I suspected that the varieties defined by $I$ and $I_0$ had to coincide. Anyway I am still confused about $I$ not being radical. Can we have a non radical ideal $I$ such that $I$ and $r(I)$ define the same scheme? This seems to be the case. Could you elaborate a bit more on the facr that $I$ and $I_0$ define the same scheme in $\mathbb{P}^9$? $\endgroup$ – user125056 Oct 31 '20 at 21:00
  • 1
    $\begingroup$ Your question is, I think, exactly @AlexanderWoo’s point: a non-radical ideal and its radical will always define the same variety, but never the same scheme. It may help to simplify your picture to $I = (x^2)$; then the variety determined by $I$ is the single point $\{0\}$ in affine space, but the scheme determined by $I$ is a doubled point $\{0\}$. You can see the difference, for example, by finding that the tangent space to the scheme at its sole point is 1-dimensional. $\endgroup$ – LSpice Nov 1 '20 at 1:04
  • 1
    $\begingroup$ No I think that Alexander Woo is claiming that the ideal $I$ and its radical define the same scheme in $\mathbb{P}^9$ but different schemes in $\mathbb{A}^{10}$. $\endgroup$ – user125056 Nov 1 '20 at 9:36
  • $\begingroup$ @user125056, you are right. Sorry! $\endgroup$ – LSpice Nov 1 '20 at 14:34
  • $\begingroup$ @user125056: You should look at Exercise II.5.10 in Hartshorne for more details. $\endgroup$ – Alexander Woo Nov 3 '20 at 4:49

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.