5
$\begingroup$

Suppose you are given a convex polyhedron $\Delta$ in $\mathbb{R}^n$ (i.e. a convex hull of finitely many points in $\mathbb{Z}^n$) and consider a finite dimensional vector space $V$ over $\mathbb{C}$ defined by $$V=\left\{f(x)=\sum_{\alpha\in\mathbb{Z}^n}a_\alpha x^\alpha\in\mathbb{C}[x_1^{\pm},\cdots,x_n^\pm] \mid a_\alpha=0 \text{ if } \alpha\notin\Delta \right\}.$$ Suppose further that $\dim \Delta\leq n-1.$ We say $f\in V$ is Newton non-degenerate if $$\left\{x\in(\mathbb{C}^{*})^n\mid\frac{\partial f}{\partial x_1}(x)=\cdots=\frac{\partial f}{\partial x_n}(x)=0\right\}=\emptyset.$$ My question is

Is the set $$\{f\in V|f\text{ is Newton non-degenerate}\}$$ Zariski open in $V$?.

It is widely believed that this result was first affirmatively proved by Kouchnirenko (Theorem 6.1., Invent. Math., 1976; link), but it seems he just proved there is a Zariski open subset of $V$ consisting only of Newton non-degenerate polynomials. It might be possible that my understanding of French is not good enough, but at least he does not prove the Zariski openness of the set $\{f\in V\mid f\text{ is Newton non-degenerate}\}$.

Could you please give me another reference stating that the set $\{f\in V|f\text{ is Newton non-degenerate}\}$ is Zariski open? Or could you suggest the outline of the proof?

Thank you very much in advance

$\endgroup$
1
$\begingroup$

The set described in the question is not Zariski open. Let $n=2$ and $\Delta = \mathrm{ConvexHull}((2,0), (0,2))$. So we are considereding polynomials $a x^2+bxy+cy^2$. Then your set is $$\{ b^2 \neq 4 ac \} \cup \{ (t,0,0),\ (0,0,t) : t \neq 0 \}.$$ This contains the Zariski open set $b^2 \neq 4ac$, but is not Zariski open.

You might wonder if this could be fixed by restricting to the open set where the $a_{\alpha}$ are nonzero (for $\alpha \in \Delta$), but no. Take $n=3$ and $\Delta = \mathrm{ConvexHull}((2,0,0),(0,2,0), (0,0,2))$. So we are considering quadratic polynomials $(x_1, x_2, x_3) A (x_1,x_2,x_3)^T$ with $A$ some symmetric matrix. The resulting conic is smooth iff $\det A \neq 0$. It is smooth when restricted to $x_1 x_2 x_3 \neq 0$ both on the open set $\det A \neq 0$, but it is also zero if the conic is a pair of lines crossing at a point where one of the coordinates vanishes.

$\endgroup$
2
  • $\begingroup$ Thank you very much! You are right, and thanks to your example, I could notice that my definition of Newton non-degeneracy was wrong. $\endgroup$ – Templeman Feb 19 '17 at 18:22
  • $\begingroup$ Thank you very much! You are right, and thanks to your example, I could notice that my definition of Newton non-degeneracy was wrong. I would like to modify the definition as follows. Under the same setting, $f(x)$ is said to be Newton non-degenerate if for any face $\Delta^\prime <\Delta,$ the $\Delta^\prime$ principal part $$f_{\Delta^\prime}(x)=\sum_{\alpha\in\Delta^\prime}a_\alpha x^\alpha$$ satisfies $$\{x\in(\mathbb{C}^*)^n|\frac{\partial f_{\Delta^\prime}}{\partial x_1}(x)=\cdots =\frac{\partial f_{\Delta^\prime}}{\partial x_n}(x)=0\}=\phi.$$ Isn't it still Zariski open? $\endgroup$ – Templeman Feb 19 '17 at 18:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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