3
$\begingroup$

For points $p_1=[1,0,0,0], p_2=[0,1,0,0], p_3=[0,0,1,0]$, $p_4=[0,0,0,1]$ and $p_5=[1,1,1,1]$
in the projective space $\mathbb P^3$, Let $l_{ij}$ be the line through $p_i, p_j$. Let

$$C=l_{12} \cup l_{23} \cup l_{34} \cup l_{45} \cup l_{51}.$$

Can one construct an explicit family of curve $\pi : \mathcal C \rightarrow B$ such that $\pi^{-1}(0)=C$ and $\pi^{-1}(t)$ is a smooth elliptic curve in $\mathbb P^3$ for $0<|t|<\epsilon$, where $\epsilon$ is a sufficiently small positive number?

More concretely, I would like to know the defining equations of $\pi^{-1}(t)$ in $\mathbb P^3$ for each $t$.

$\endgroup$
11
  • 2
    $\begingroup$ Is it your homework? $\endgroup$
    – Sasha
    Commented Apr 9, 2020 at 8:42
  • 1
    $\begingroup$ No, it is not. It is related with my research( I simplified a little bit) Is it an easy question that can be on a homework? Then please give me an answer! $\endgroup$
    – Basics
    Commented Apr 9, 2020 at 8:45
  • 3
    $\begingroup$ Not the union, the intersection! Just deform $Q'$ to a general quadric. Note that we are in $\Bbb{P}^3$, in case you didn't notice. $\endgroup$
    – abx
    Commented Apr 9, 2020 at 9:34
  • 1
    $\begingroup$ Now I have edited the question so that the smoothed elliptic curve is not a complete intersection. $\endgroup$
    – Basics
    Commented Apr 9, 2020 at 10:02
  • 2
    $\begingroup$ Try to write down equations for an elliptic curve of degree 5 in $\Bbb{P}^3$, you'll understand why there is no easy answer. $\endgroup$
    – abx
    Commented Apr 10, 2020 at 4:32

1 Answer 1

4
$\begingroup$

The answer to your first question is yes, but the defining equations seem harder to get (as @abx points out).

Let me explain why the answer to your question is yes.

Let $p_1 = [1,0,0,0,0], p_2 = [0,1,0,0,0], \ldots , p_5=[0,0,0,0,1]$ be five points in $\mathbb{P}^4$. Consider the line $L_{i,j}$ joining $p_i$ to $p_j$. The curve: $$C_0 = L_{1,2} \cup L_{2,3} \cup L_{3,4} \cup L_{4,5} \cup L_{5,1}$$ is the classical pentagonal configuration (it was studied in details by Ellinsgrud-Laksov for instance).

For $a \in \mathbb{C}$, consider the following quadrics in $\mathbb{P}^4$: $$Q_0 : ax_0^2 + a^2x_2x_3 -x_1x_4 = 0,$$ $$ Q_1 : ax_1^2 + a^2x_3x_4 - x_2x_0 =0,$$ $$ Q_2 : ax_2^2 + a^2x_4x_0 -x_3x_1 = 0,$$ $$Q_3 : ax_3^2 + a^2x_0x_1 - x_4x_2 = 0, $$ $$Q_4 : ax_4^2 + a^2x_1x_2 - x_0x_3 = 0.$$

For $a=0$, the intersection of this five quadrics is the penthagonal configuration $C_0$ and for $a$ generic, you can prove (with some computations) that intersection of these quadrics is scheme-theoretically a normal elliptic quintic in $\mathbb{P}^4$.

Hence you get a smoothing of $C_0$ in $\mathbb{P}^4$ by taking $X$ the intersection of the five quadrics $\{Q_i\}_{i=0\ldots 4}$ in $\mathbb{P}^4 \times \mathbb{A}^1$ and $\pi : X \longrightarrow \mathbb{A}^1$ the natural projection.

Now, your question is related to a smoothing of the penthagonal configurations in $\mathbb{P}^3$. Geometrically the smoothing is esay to obtain. Let $P$ be a generic point in $\mathbb{P}^4$. Let:

$$p_P : \mathbb{P}^4 \longrightarrow \mathbb{P}^3$$ be the projection from $P$. If $P$ is outside the secant-threefold of $C_0$, then $p_P$ realizes an isomorphism from $C_0$ to $p_P(C_0)$, the latter being a pentagonal configuration in $\mathbb{P}^3$.

The projection map $p_P : C_a \longrightarrow p_P(C_a)$ is then an isomorphism for generic $a \in \mathbb{A}^1$ and the curve $p_P(C_a)$ is a quintic elliptic curve in $\mathbb{P}^3$.

One obtains the smoothing you are longing for in the following way: let $\mathcal{C}$ be the projection of $X$ in $\mathbb{P}^3 \times \mathbb{A}^1$. The projection map $\pi : \mathcal{C} \longrightarrow \mathbb{A}^1$ staisfies $\pi^{-1}(0) = P_p(C_0)$ is a pentagonal configuration and for $a \in \mathbb{A}^1$, the curve $\pi^{-1}(a)$ is a smooth elliptic quintic curve in $\mathbb{P}^3$.

The second part of your question is more difficult (at least in $\mathbb{P}^3$). It is easy in $\mathbb{P}^4$ because it is easily proved that $\mathcal{J}_C(2)$ is globally generated, where $C$ is a normal elliptic quintinc. On the other hand, it is easily shown that a quintic elliptic curve in $\mathbb{P}^3$ is contained in at most one quadric surface (exactly one when it is the nodal/cuspidal projection of a normal elliptic curve in $\mathbb{P}^4$, and probably none when it is generic). Hence the equations of the smoothing can't be quadratic, whch makes the situation more difficult. I don't know if cubic equations are sufficient...

EDIT : Since I found it a fun "homework" I kept on working on it, and I think one can explicit the equations with Macaulay2. In fact, getting the equations of an elliptic curve in $\mathbb{P}^3$ is not the problem. A smooth elliptic quintic in $\mathbb{P}^3$ is scheme-theoretically defined by 5 cubics and these cubics can be found in a few seconds by Macaulay2 (using elimination theory). The hard part is to get the equations for the deformation to the pentagonal configuation. So I changed a bit my inital pentagonal configuration in $\mathbb{P}^4$.

Let $q_1 = [1,0,0,0,0], q_2 = [0,1,0,0,0], q_3=[0,0,1,0,0] , q_4=[0,0,0,1,0], q_5=[1,1,1,1,-1]$. Consider the line $D_{i,j}$ joining $q_i$ to $q_j$. The curve: $$C_0 = D_{1,2} \cup D_{2,3} \cup D_{3,4} \cup D_{4,5} \cup D_{5,1}$$ is again a pentagonal configuration.

For $[a,b] \in \mathbb{P}^1$, consider the following quadrics in $\mathbb{P}^4$: $$T_0 : ab(x_0+x_4)^2 + a^2(x_2+x_4)(x_3+x_4) -2b^2(x_1+x_4)x_4 = 0,$$ $$ T_1 : ab(x_1+x_4)^2 + 2a^2(x_3+x_4)x_4 - b^2(x_2+x_4)(x_0+x_4) =0,$$ $$ T_2 : ab(x_2+x_4)^2 + 2a^2x_4(x_0+x_4) -b^2(x_3+x_4)(x_1+x_4) = 0,$$ $$T_3 : ab(x_3+x_4)^2 + a^2(x_0+x_4)(x_1+x_4) -2b^2x_4(x_2+x_4) = 0, $$ $$T_4 : 4abx_4^2 + a^2(x_1+x_4)(x_2+x_4) -b^2(x_0+x_4)(x_3+x_4) = 0.$$

Let $X$ be the scheme defined by the intersection of the $T_i$ in $\mathbb{P}^4 \times \mathbb{P}^1$. The projection map $\pi : X \longrightarrow \mathbb{P}^1$ gives you (generically) a smoothing of $C_0$. Again, we can project everything into $\mathbb{P}^3$ from a generic point to get a smoothing to an elliptic curve of the pentagonal configuration you started with.

As a matter of fact, the point $[0,0,0,0,1]$ is generic enough to get that smoothing. If $p : \mathbb{P}^4 \times \mathbb{P}^1 \longrightarrow \mathbb{P}^3 \times \mathbb{P}^1$ is the projection from $[0,0,0,0,1]$, the equations of $p(X)$ are given by the ideal obtained from $I = (Q_0,Q_1,Q_2,Q_3,Q_4)$ and eliminating $x_4$. This can be done using Macaulay2. There are only $42$ equations($42$, oviously!), but they are a bit long. I can't print them all on mathoverflow, but at least, I can give you the first one (Macaualay2 does not siplify much the coefficients) $$2a^3bx_0^3+ab^3x_0^3-2a^4x_0^2x_1-3a^2b^2x_0^2x_1-ab^3x_0^2x_1+a^3bx_0x_1^2+3a^2b^2x_0x_1^2-a^3bx_1^3-2ab^3x_1^3+3a^2b^2x_0^2x_2-ab^3x_0^2x_2-2a^3bx_0x_1x_2+2b^4x_0x_1x_2+a^3bx_1^2x_2+2ab^3x_1^2x_2+a^3bx_0x_2^2-3a^2b^2x_0x_2^2-2b^4x_0x_2^2+a^3bx_1x_2^2+2ab^3x_1x_2^2-a^3bx_2^3-2ab^3x_2^3-2a^3bx_0^2x_3-ab^3x_0^2x_3+2a^4x_0x_1x_3+2ab^3x_0x_1x_3+a^3bx_1^2x_3-3a^2b^2x_1^2x_3-2b^4x_1^2x_3+2a^4x_0x_2x_3+2ab^3x_0x_2x_3-2a^3bx_1x_2x_3+2b^4x_1x_2x_3+a^3bx_2^2x_3+3a^2b^2x_2^2x_3-2a^3bx_0x_3^2-ab^3x_0x_3^2+3a^2b^2x_1x_3^2-ab^3x_1x_3^2-2a^4x_2x_3^2-3a^2b^2x_2x_3^2-ab^3x_2x_3^2+2a^3bx_3^3+ab^3x_3^3$$

If you are really interested by the specific equations, you can contact me in PM and I will send them to you.

$\endgroup$
5
  • $\begingroup$ Why do you say that a smooth quintic elliptic curve in $\mathbb{P}^3$ lies on a smooth quadric surface? I guess it does not (what its bidegree would be?}. Instead, it lies on a smooth cubic surface. As such, it can be thought of as a curve of type $3\ell - e_1 - e_2 - e_3 - e_4$. In the same linear system you can find the penthagon of lines $e_1, \ell - e_1 - e_2, e_2, \ell - e_2 - e_3, \ell - e_1 - e_4$. Of course, one is the deformation of the other. $\endgroup$
    – Sasha
    Commented Apr 11, 2020 at 4:29
  • $\begingroup$ @Sasha : you are right for the quadric, it is true only if the projection is from a point $P \in Sec(C) $ (so that the curve in $\mathbb{P}^3$ is not smooth but nodal/cuspidal), I will correct that. As for the second part of your comment, I am not sure I understand the notations and the argument. In any case, I don't think that the smoothing in $\mathbb{P}^3$ of a pentagonal configuration to an elliptic curve can be realized only with cubic equations. Though, if it were true, it would certainly be intresting, it doesn't appear in Ellingsrud-Laksov. $\endgroup$
    – Libli
    Commented Apr 11, 2020 at 10:16
  • $\begingroup$ I am using the standard notation for the blowup of a $\mathbb{P}^2$ at 6 points ($\ell$ is the line class, $e_i$ are the exceptional divisors). I don't claim that the deformation can be realized by cubic equations; instead, I say that one can deform a curve within a linear system on the fixed cubic surface. $\endgroup$
    – Sasha
    Commented Apr 11, 2020 at 11:38
  • $\begingroup$ @Sasha : I don't see how you find a smooth elliptic quintic in the linear system $3 \ell -e_1-e_2-e_3-e_4$ $\endgroup$
    – Libli
    Commented Apr 11, 2020 at 16:13
  • $\begingroup$ A general curve in this linear system is the strict transform of a plane cubic passing through 4 out of 6 blowup points, a general such curve is smooth, hence so is its strict transform. Its degree is 5 because $(3\ell - e_1 - e_2 - e_3 - e_4)(3\ell - e_1 - e_2 - e_3 - e_4 - e_5 - e_6) = 9 - 4 = 5$. $\endgroup$
    – Sasha
    Commented Apr 11, 2020 at 18:57

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .