MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I would like to understand some trivialities about log-structures. Given a log-scheme $(X,M_X)$ the log-structure $M_X$ is defined via push-out. Are there stupid examples in which this push-out is actually a direct sum $\mathbb{G}_m\oplus P$ for some monoid $P$?

Furthermore assume we have a ring $R$, an element $t\in R$ defining a log-structure on $Spec(R)$ via $\mathbb{N}\rightarrow t^n$. Consider $X=Spec(R[x,y]/(xy-t))$ with the standard log-structure on the two brances $x$ and $y$. Let $p:Y\rightarrow X$ be the blow-up of $X$ at $(x,y,t)$ and put a log-structure on $Y$ over the preimage $\tilde{x},\tilde{y}$ of the two branches $x,y$ but not on the exceptional divisor. Namely locally on $Y$ we take $\mathbb{N}\rightarrow \mathcal{O}_Y$ where we send $1$ to $\tilde{x}$ (resp. to $\tilde{y}$). Call this log-structure $M_Y$. Is it true that $p_{*}M_Y$ splits as a direct sum? If so how do the summands look like?

share|cite|improve this question
up vote 1 down vote accepted

For the first question, given a sharp monoid $P$ (i.e., the group of units of $P$ is the zero group), there is a log structure on any scheme $X$ given by $M_X={\mathcal O}_X^*\oplus P$ with the structure map given by $(f,p)\mapsto f$ if $p=0$ and $(f,p)\mapsto 0$ if $p\not =0$. This comes from the prelog structure $P\rightarrow {\mathcal O}_X$ given by $p\mapsto 1$ if $p=0$, $p\mapsto 0$ if $p\not=0$. I wouldn't call this a stupid log structure: it in fact can be quite useful. For example, the standard log point is defined as the log structure on the scheme $Spec(k)$ with $P$ the natural numbers as above.

For the second question, I'm not quite sure what direct sum decomposition you are expecting, but I can't think of a natural direct sum decomposition here. If you are thinking that $p_*M_Y$ will decompose as a direct sum of the sheaf ${\mathcal O}_X^*$ and a sheaf supported on the branches $x$ and $y$, this is not the case. To take a simpler example, taking the divisorial log structure associated to $0 \in \mathbb{A}^1$, the log structure does not decompose as $O_{A^1}\oplus {\mathbb N}_0$. Indeed, there is no section of $M_{\mathbb{A}^1}$ with support at $0$. In the example you describe, there will be an exact sequence $$ 1 \rightarrow {\mathcal O}_Y^*\rightarrow M_Y \rightarrow {\mathbb{N}}_x \oplus {\mathbb{N}}_y\rightarrow 0 $$ where the subscripts $x$ and $y$ denote which branch the constant sheaf with coefficients ${\mathbb N}$ is sitting on. I believe when pushing down, one obtains an exact sequence $$ 1\rightarrow {\mathcal O}_X^*\rightarrow p_*M_Y\rightarrow {\mathcal F} \rightarrow 0 $$ where ${\mathcal F}$ is the sheaf supported on $D$, the vanishing locus of $t$, which is obtained by taking the constant sheaf ${\mathbb N}$ on $D-\{0\}$ and extending by zero across $0$. This is because there are no functions on the inverse image of a open neighbourhood of $0$ which do not vanish off the two branches, but do vanish one or both of the branches.

share|cite|improve this answer
@Mark Gross by direct sum I mean $p_{*}M_Y=p_{*}M_{\tilde{x}}\oplus p_{*}M_{\tilde{y}}$ where the factor $M_{\tilde{x}}$ (resp. M_{\tilde{y}}) takes care of the log-structure given by $t,\tilde{x}$ (resp. $t,\tilde{y}$). I would like to know if this direct sum exists as $p_{*}\mathcal{O}_{X}^{*}$-log-structure and if somehow the log-structure of each piece $M_{\tilde{x}}$ and $M_{\tilde{y}}$ is trivial. – ketth Jun 24 '13 at 8:02
I'm having trouble with mathjax since the change to Math Overflow 2.0, so I don't know if this will be readable, but here goes. It seems to me that $p_*M_Y$ – Mark Gross Jun 25 '13 at 18:08
...sorry, nothing seems to be working correctly. One more try: $p_*M_Y=F_1\oplus_{O_X^*} F_2$, where F_1 and F_2 are funny log structures whose ghost sheaf ($F_i/O_X^*$) is a copy of ${\mathbb N}$ on one of the two branches, but extended by zero over the origin. I'm still not sure if this is what you are looking for. I think the particular push-forward log structure you described in your original question is not a particularly well-behaved one, e.g., it is not fine. – Mark Gross Jun 25 '13 at 18:12

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.