Skip to main content
edited tags
Link
YCor
  • 63.9k
  • 5
  • 187
  • 286
Source Link
Yikun Qiao
  • 930
  • 6
  • 11

What is bad when stabilizers are non-reductive in moduli stacks?

Here is J. Alper's definition of good moduli spaces. Definition of good moduli spaces

Consider in characteristic zero. Then we see that the classifying stack of any non-reductive group $H$ does not have a good moduli space. In particular the natural morphism $p:[*/H]\to *$ is not a good moduli space. This is because $p_*$ sends a $H$-representation $V$ to $V^H$, which is not right exact.

Good moduli spaces follow from good categorical quotients in GIT, where non-reductive groups do not play a role. However, there are actions with obvious nice quotients, for example the trivial action $H\curvearrowright *$ and the left multiplication $H\curvearrowright H$. One of them, $[*/H]$, does not have a good moduli space, and the other, $[H/H]$, does have one.

My questions are as follows.

  1. Do people define other terminologies so that $[*/H]\to*$ has a name?
  2. Why reductivity of stabilizers of (closed) points so desirable?