F.Kato has a statement said that a log smooth curve $f:(X, M,\alpha)\to (k,N,\beta)$ where $k$ is an algebraic closed field and $N$ is some fine log structure on $k$, is equvalent to pointed node curve in a paper "Log sooth deformation and moduli of log smooth curves". I am quite confused with the log structure $N$ on k, when one associate a log smooth curve to a pointed node curve.

Let's consider the following example, $X$ has only one double point $x$, and only one marked point $s$, and $x\neq s$. Let $A=k[t_1, t_2]/(t_1t_2)$, $U=Spec (A)$ be an etale neighborhood of $x$ , with $x= (0,0)$, we see that the log structure over $U$ is defined by the following $\alpha:M_U:=k^*\oplus N^2 \to A $ by $\alpha(k_1,n_1,n_2)= k_1t_1^{n_1}t_2^{n_2}$, and the log structure $N:=k^*\oplus \mathbb{N}$ and the map $\beta:N\to k $ defined by $\beta(k_1,0)=k_1$, and $\beta(k_1,n)=0$ for $n>0$. The map $f$ on the monoid is defined by the following $f^0:N\to M_U$, $f^0(k_1,n)=(k_1,n,n)$.

On the other hand Let $B=k[T]$ , $V=Spec(B)$, be an etale neighborhood of $s$, with $s=(0)$. Then the log structure on V is $M_V:=k^*\oplus \mathbb{N}$ with $\alpha(k_1,n)=k_1T^n$.

Now F. Kato claimed one extend $M_U$, $M_V$ to $X$, and define $M:=M_U\oplus_{\mathcal{O}^{\times}_{X}} M_V$, then $(X, M, \alpha)$ is log smooth over $(k, N, \beta)$. My question is how to define the map $f^0 : N\to M$ on the monoid? Over $U$, it is already given above. But what about over V, it seem that it is impossible to define such a map, for one need to satisfy the following relation: $ \alpha\circ f^0= i\circ \beta$, where $i: k\to k[T]$ is the natural map. And one fine that $\alpha\circ f^0(k_1, n)=i(0)=0 $ for $n>0$, however $0\notin Imge( \alpha)$, so it is impossible to define $f^0$.


1 Answer 1


You haven't really explained your notation, but it seems that you are covering the log curve $X$ with two neighborhoods $U$ and $V$, where $U$ contains a node, and $V$ contains a log point. I think the problem with your definition of $M_V$ is that you forgot about the nontrivial log structure on the base log scheme. If I'm not mistaken, the neighborhood a log point on the curve should have log structure associated to $k^* \times \mathbb{N}^2$ instead of $k^* \times \mathbb{N}$, with $\alpha(k_1,m,n) = k_1 0^m T^n$ and $f^0(k_1,m) = (k_1,m,0)$.

  • $\begingroup$ @Carnahan, you are right, I made a mistake about the log structue near the marked point. Thank you for your help. $\endgroup$
    – Lan
    May 18, 2014 at 16:28

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.