1
$\begingroup$

I was reading S. Mukai's- An Introduction to Invariants and Moduli, and I came across Proposition 6.16 on page 195 (see the screenshot below)

enter image description here

It says that "every invariant rational function can be expressed as ratio of semiinvariants of weight $\chi$."

Should it be $n\chi$, for some $n\in\mathbb{N}$?

The reason I'm confused is because- say I consider the weight space determined by $\chi$, i.e., all the semiinvariants of weight $\chi$. It has the structure of a $\mathbb{K}$-vector space. Say I choose two of the basis element $p,q\in\mathfrak{B}$. Now consider the invariant rational function $\frac{p^2}{q^2}$, which is ratio of semiinvariants of weight $2\chi$.

I don't understand how can I write $\frac{p^2}{q^2}$ as ratio of semiinvariants of weight $\chi$?

Also, does the Definition 6.15 about "stable with respect to $\chi$" coincide with the King's stability condition (see here) in the context of quiver representations?

$\endgroup$
3
  • 1
    $\begingroup$ I think you are right. There might not even exist semiinvariants of weight $\chi$. $\endgroup$ Commented Dec 1, 2023 at 9:07
  • $\begingroup$ @FriedrichKnop But say there exists semiinvariants of weight $\chi$, then is the statement of the proposition true? Or should it still be $n\chi$? And, in this case is the $n$ fixed or does $n$ depend on the invariant rational function that I choose? $\endgroup$
    – It'sMe
    Commented Dec 1, 2023 at 19:02
  • 1
    $\begingroup$ It should still be $n\chi$ because there might not be enough semiinvariants of weight $\chi$. Possibly the simplest example is $X=\mathbf A^2$ and $G=\mathbf G_m$ acting by $t\cdot(x,y)=(tx,t^2y)$. Take $\chi(t)=t$. Then all points are $\chi$-stable except for the origin and $f=x^2/y$ is a counterexample. So there is a glitch in Mukai's book. The last sentence of the Proposition is not affected, though. $\endgroup$ Commented Dec 4, 2023 at 14:11

0

You must log in to answer this question.