6
$\begingroup$

My question is on which situation stable map space $\overline{M_{g,n}}(X,\beta)$

can have non-reduced structure.

There is many example of stack structure from automorphism of stable curve

but I don't know any example of non-reduced structure

And more explicitly, when (g,n) = (0,0), is it still posible that

$\overline{M_{0,0}}(X,\beta)$ can have non-reduced stack structure?

$\endgroup$
3
  • 1
    $\begingroup$ If $X$ is a surface of degree $d\geq 4$ in $\mathbb{P}^3$, then for every stable map from a genus $0$ (unpointed) curve into the smooth locus of $X$, the stack $\overline{\mathcal{M}}_{0,0}(X,\beta)$ is nonreduced at the corresponding point. $\endgroup$ Apr 10, 2015 at 11:09
  • $\begingroup$ Thank you very much!! can you explane me the reason why? $\endgroup$ Apr 10, 2015 at 13:31
  • $\begingroup$ What I wrote was wrong: there are such examples, e.g., Shioda's inseparably uniruled hypersurfaces, but the general such example does not work. Let me give a better example below. $\endgroup$ Apr 11, 2015 at 12:06

2 Answers 2

3
$\begingroup$

There are examples of the following form. Let $C\subset X$ be a contractible curve in a threefold which is isomorphic to $\mathbb{P}^1$ and has normal bundle $\mathcal{O}\oplus\mathcal{O}(-2)$. Such examples were studied by Miles Reid and have very explicit local models. In such a situation, the inclusion $C\to X$ is an isolated point in $\overline{M}_0(X,[C])$ which has non-reduced structure. The infinitesimal deformation given by the non-zero section of $N_{C/X}$ is obstructed. The length of the non-reduced structure is an invariant that Mile Reid calls width.

Explicitly, let $V$ be the affine hypersurface in $\mathbb{C}^4$ given by $x^2+y^2+z^2+w^{2k}=0$ where $k\geq 2$. Let $X\to V$ be the small resolution of this singularity. The exceptional set of this resolution is $C$. Then $\overline{M}_0(X,[C])\cong Spec(\mathbb{C}[t]/(t^k))$ .

$\endgroup$
1
  • $\begingroup$ Thank you so much! I'll check this line by line as soon as Possible! $\endgroup$ Apr 12, 2015 at 6:25
2
$\begingroup$

The suggestion in my comment was wrong -- I was incorrectly (and absurdly) extrapolating from Shioda's examples of inseparably uniruled surfaces (which are beautiful, and which I would love to generalize). Here is the simplest example I know that works in characteristic $0$.

Let $k$ be an algebraically closed field of characteristic $\neq 2,3$, and work in the category of $k$-schemes. Let $\mathbb{P}_k^4$ have homogeneous coordinates $[Z_0,Z_1,Z_2,Z_3,Z_4]$. Let $F$ be the degree $4$ (essentially) Fermat polynomial, $$ F(Z_0,Z_1,Z_2,Z_3,Z_4) = Z_0^4 + Z_1^4 + Z_2^4 + Z_3^4 + Z_4^4. $$ Denote by $X\subset \mathbb{P}^4_k$ the zero scheme of $F$. Since $\text{char}(k)\neq 2$, this is a smooth quartic hypersurface of dimension $n=3$. Now let $U$ be the open subset of $X\times X$ that is the complement of the diagonal $\Delta_X$. Let $F\subset U$ be the closed subscheme parameterizing ordered pairs of distinct points $([S_0,S_1,S_2,S_3,S_4],[T_0,T_1,T_2,T_3,T_4])$ such that the homogeneous polynomial in coordinates $[S,T]$, $$ F(TS_0-ST_0,TS_1,-ST_1,TS_3-ST_3,TS_4-ST_4) = T^4(S_0^4+S_1^4+S_2^4+S_3^4) - $$ $$ 4ST^3(S_0^3T_0+S_1^3T_1+S_2^3T_2+S_3^3T_3+S_4^3T_4) + $$ $$ 6S^2T^2(S_0^2T_0^2 +S_1^2T_1^2 + S_2^2T_2^2 + S_3^2T_3^2 + S_4^2T_4^2) - $$ $$ 4S^3T(S_0T_0^3+S_1T_1^3+S_2T_2^3+S_3T_3^3+S_4T_4^3) + $$ $$ S^4(T_0^4+T_1^4+T_2^4+T_3^4+T_4^4). $$ Since the characteristic is not $2$ or $3$, it is straightforward to see that $F$ has codimension $3$ in $U$. It follows that the space of lines, $\overline{\mathcal{M}}_{0,0}(X,1)$ in the stable map notation, is a projective curve.

Now let $i$ be a root of $t^4+1$ in $k$, and consider the hyperplane section $\text{Zero}(Z_4-iZ_3)\cap X$. Using homogeneous coordinates $[Z_0,Z_1,Z_2,Z_3]$ on $\text{Zero}(Z_4-iZ_3)$, the intersection with $X$ is the zero scheme of $Z_0^4+Z_1^4+Z_2^4$. In particular, there is no dependence on $Z_3$. This means that the hyperplane section is a cone with vertex $[Z_0,Z_1,Z_2,Z_3] = [0,0,0,1]$.
Thus the plane curve $$ B= \{[Z_0,Z_1,Z_2] \in \mathbb{P}^2_k |Z_0^4+Z_1^4+Z_2^4 = 0\}, $$ is an irreducible component of the space of lines: for every $[S_0,S_1,S_2]$ in $B$, associate the line $L=\text{span}([S_0,S_1,S_2,0,0],[0,0,0,1,i]\}$. But it is straightforward to compute that the normal bundle of $L$ in $X$ is isomorphic to $\mathcal{O}_L(-2)\oplus \mathcal{O}_L(1)$. Thus, the Zariski tangent space at every point of $B$ is $2$-dimensional. Therefore the space of lines is everywhere reduced along $B$.

I believe this example is an exercise in Debarre's textbook.

$\endgroup$
1
  • $\begingroup$ Thank you so much! I'll check this line by line as soon as Possible! Can you tell me the name of the Debarre's textbook? $\endgroup$ Apr 12, 2015 at 6:25

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.