退出

Is $V(X+Y-Z)$ a toric variety?

代数几何 Math StackExchange 2 票 1 回答 49 浏览 提问者: Cecilia 2026-08-18 21:16
algebraic-geometry affine-varieties toric-varieties

问题内容

I'm reading CLS's Toric Varieties right now, and something is confusing me.

By Theorem 1.1.17, an affine variety $V$ is toric iff $I(V)$ is toric, i.e. prime and generated by binomials.

Now the variety $V = V(X+Y-Z) \subset \mathbb{C}^3$ seems to me to be toric simply because it's isomorphic to $\mathbb{C}^2$. Via the isomorphism $\phi: \mathbb{C}^2 \rightarrow V, (x,y) \mapsto (x,y,x+y)$, I think one should get the open torus (?) $T_V = \phi((\mathbb{C}^*)^2) = V\setminus V(XY) = \{(x,y,x+y)\ |\ x,y\in\mathbb{C}^*\} \subset V$ with the inherited action $(\mathbb{C}^*)^2 \curvearrowright T_V$ and everything working fine.

On the other hand, the ideal $I(V) = (X+Y-Z) \subset \mathbb{C}[X,Y,Z]$ of $V$ is clearly prime, but I just cannot see how one would be able to generate it using only binomials.

Where is my mistake here?

回答 (1)

User 1 票 2026-08-18 21:55 原文

Theorem 1.1.17 asserts the existence of an embedding for which the defining ideal is toric, not that the ideal of every embedding of a toric variety must be toric. Consider the map $\phi:\mathbb C^2\longrightarrow V$ where $(x,y)\longmapsto (x,y,x+y)$, which is an isomorphism. Therefore, the open subset $T=\phi\left( (\mathbb C^*)^2 \right)=\left\{ (x,y,x+y)\mid x,y\neq 0 \right\}$ is isomorphic to the torus $(\mathbb C^*)^2$. Its action extends to all of $V$ by the formula $(\lambda,\mu)\cdot(x,y,x+y)=(\lambda x,\mu y,\lambda x+\mu y)$, and hence $V$ is indeed an affine toric variety

However, our embedding $V\subset\mathbb C^3$ is not toric with respect to the coordinates $X,Y,Z$. On the open torus, the restrictions of the coordinate functions are $X|_T=x$, $Y|_T=y$, $Z|_T=x+y$. The functions $x$ and $y$ are characters of the torus, whereas $x+y$ is a sum of two characters rather than a character. In a toric embedding, the coordinate functions must be given by characters, that is, by monomials, so in these coordinates one obtains the trinomial $X+Y-Z$ rather than a binomial relation. It is worth noting that the ideal $I(V)=(X+Y-Z)$ in these coordinates indeed contains no nonzero binomials. If $X^aY^bZ^c-\lambda X^{a'}Y^{b'}Z^{c'}\in I(V)$, then after substituting $Z=X+Y$ one would obtain $\lambda X^{a'}Y^{b'}(X+Y)^{c'}$. Since $X$, $Y$, and $X+Y$ are pairwise nonassociate irreducible elements of the ring $\mathbb C[X,Y]$, uniqueness of factorization implies that all the exponents coincide and $\lambda=1$. Hence such a binomial is zero


To obtain a toric ideal, it suffices to choose different coordinates $u=X$, $v=Y$, $w=Z-X$. Then the equation $X+Y-Z=0$ becomes $v-w=0$. Hence $V\simeq V(v-w)\subset\mathbb C^3$, where $(v-w)$ is a prime binomial toric ideal. The corresponding monomial parametrization is $(s,t)\longmapsto(s,t,t)$