This paper studies the structure of prime spectra in algebras acted upon by cocommutative Hopf algebras, revealing stratification methods and properties of stable ideals under such actions.
Contribution
It introduces a stratification of prime spectra for algebras with integral cocommutative Hopf algebra actions and proves stability properties of semiprime ideals in characteristic zero.
Findings
01
Prime spectrum stratification in terms of commutative subalgebras
02
Largest H-stable ideal in a semiprime ideal is semiprime
03
Results hold under integral action assumptions in characteristic zero
Abstract
Let H be a cocommutative Hopf algebra acting on an algebra A. Assuming the base field to be algebraically closed and the H-action on A to be integral, that is, it is given by a coaction of some Hopf subalgebra of the finite dual H∘ that is an integral domain, we stratify the prime spectrum \mboxSpecA in terms of the prime spectra of certain commutative algebras. For arbitrary H-actions in characteristic 0, we show that the largest H-stable ideal of A that is contained in a given semiprime ideal of A is semiprime as well.
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Full text
Actions of Cocommutative Hopf Algebras
Martin Lorenz
,
Bach Nguyen
and
Ramy Yammine
Department of Mathematics, Temple University,
Philadelphia, PA 19122
Abstract.
Let H be a cocommutative Hopf algebra acting on an algebra A. Assuming
the base field to be algebraically closed and the H-action on A to be integral, that is, it is
given by a coaction of some Hopf subalgebra of the finite dual H∘
that is an integral domain, we stratify the prime spectrum
SpecA in terms of the prime spectra of certain commutative algebras.
For arbitrary H-actions in characteristic [math], we show that
the largest H-stable ideal of A that is
contained in a given semiprime ideal of A is semiprime as well.
Key words and phrases:
Hopf algebra, action, quantum invariant theory, prime spectrum,
stratification, prime ideal, semiprime ideal, integral action, rational action, algebraic group,
Lie algebra, derivation
2010 Mathematics Subject Classification:
16T05, 16T20
Introduction
0.1.
Let H be a Hopf algebra over a field k and let A be an arbitrary associative k-algebra.
An action of H on A is given by a k-linear map
H⊗A→A, h⊗a↦h.a, that makes A into a
left H-module and satisfies the “measuring” conditions
h.(ab)=(h1.a)(h2.b) and h.1=⟨ε,h⟩1 for h∈H and a,b∈A.
Here, ⊗=⊗k , Δh=h1⊗h2 denotes
the comultiplication of H, and ε is the counit.
We will write H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@setlinewidth0.32pt\pgfsys@setdash0.0pt\pgfsys@roundcap\pgfsys@roundjoin\pgfsys@moveto-1.19998pt1.59998pt\pgfsys@curveto-1.09998pt0.99998pt0.0pt0.09999pt0.29999pt0.0pt\pgfsys@curveto0.0pt-0.09999pt-1.09998pt-0.99998pt-1.19998pt-1.59998pt\pgfsys@stroke\pgfsys@endscope\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA to indicate such an action.
Algebras equipped with an H-action are called left H-module algebras.
With algebra maps that are also H-module maps as morphisms, left H-module algebras
form a category, HAlg.
For example, an action of a group algebra kG on A amounts to the datum of a group homomorphism
G→AutA, the automorphism group of the algebra A.
For the enveloping algebra Ug of a Lie k-algebra g, an action Ug\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA
is given by a Lie homomorphism g→DerA, the Lie algebra of
all derivations of A. In both these prototypical cases, the acting Hopf algebra is cocommutative.
This article investigates the effect of a given action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA
of an arbitrary cocommutative Hopf algebra H on the prime and semiprime ideals of A,
partially generalizing prior work of the first author on rational actions of algebraic groups
[11], [12], [13]. The interesting article [18] by Skryabin covers related
territory, but the actual overlap with our work is insubstantial.
0.2.
For now, let H be arbitrary and let A∈HAlg.
An ideal I of A that is also an H-submodule of A is called an H-ideal.
The action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA then passes down to an H-action on the quotient algebra A/I.
The sum of all H-ideals that are
contained in an arbitrary ideal I, clearly the unique largest H-ideal of A
that is contained in I, will be referred to as
the H-core of I and denoted by I\sl:H. Explicitly,
[TABLE]
If A=0 and the product of any two
nonzero H-ideals of A is again nonzero, then A is said to be H-prime.
An H-ideal I of A is called H-prime if A/I
is H-prime. It is easy to see that H-cores of prime ideals are H-prime. Denoting
the collection of all H-primes of A by H-SpecA, we thus obtain a map
SpecA→H-SpecA , P↦P\sl:H.
The fibers
[TABLE]
are called the H-strata of SpecA.
The stratification SpecA=⨆I∈H-SpecASpecIA was pioneered by Goodearl and Letzter [7]
in the case of group actions or, equivalently, actions of group algebras.
It has proven to be a useful tool for investigating SpecA, especially
for rational actions of a connected affine algebraic group G over an algebraically closed
field k. In this case, one has a description of each stratum SpecIA in terms of the prime spectrum
of a suitable commutative algebra [12, Theorem 9].
Our principal goal is to generalize this result to the context of cocommutative Hopf algebras.
This is carried out in Section 3 of this article.
The first two sections serve to deploy
some generalities on actions of Hopf algebras that are needed for the proof, with Section 2
focusing on the cocommutative case.
0.3.
To state our main result, we make the following observations;
see Sections 1-3 for details.
Let H be cocommutative and k algebraically closed. Assume that
the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA is locally finite, that is, dimkH.a<∞ for all a∈A.
Then A becomes a right comodule algebra over the (commutative) finite dual H∘ of H:
[TABLE]
The action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA will be called integral if the image of the map
(2) is contained in A⊗O for some Hopf subalgebra
O⊆H∘ that is an integral domain. This condition serves as a replacement for connectedness
in the case of algebraic group actions. Assuming it to be satisfied,
it follows that each I∈H-SpecA is in fact a prime ideal of A. Consequently, the extended
center C(A/I)=ZQ(A/I) is a k-field, where Q(A/I) denotes the symmetric ring of quotients
of A/I. The action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA/I extends uniquely to an H-action on Q(A/I) and this action
stabilizes the center C(A/I).
Furthermore, O∈HAlg via the “hit” action ⇀ that is given by
\langle{h\text{\scriptsize\rightharpoonup}f,k}\rangle=\langle{f,kh}\rangle for f∈O and h,k∈H.
The actions ⇀ and H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureC(A/I) combine to an H-action on the tensor product; so
[TABLE]
The algebra CI is a commutative integral domain.
We let SpecHCI denote the subset of SpecCI consisting of all prime H-ideals of CI .
Theorem 1**.**
Let H be a cocommutative Hopf algebra over an algebraically closed field k and
let A be a k-algebra that is equipped with an integral action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA.
Then, for any I∈H-SpecA, there is a bijection
[TABLE]
having the following properties, for any P,P′∈SpecIA:
(i)
c(P)⊆c(P′)* if and only if P⊆P′, and*
2. (ii)
Fract(CI/c(P))≅C((A/P)⊗O).
0.4.
As a first example,
let G be an affine algebraic k-group and let O=O(G) be the algebra of polynomial functions on G.
Then O⊆H∘ with H=kG.
A rational G-action on A, by definition, is a locally finite action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA such that the image
of (2) is contained in A⊗O. If G is connected, then O is an integral domain
and so the action is integral. In this setting, Theorem 1 is covered by [12, Theorem 9].
Next, let g be a Lie k-algebra acting by derivations on A
and assume that every a∈A is contained
in some finite-dimensional g-stable subspace of A.
With H=Ug, we then have a locally finite action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA
and hence a map (2). If chark=0, then the convolution algebra H∗
is a power series algebra over k and hence
H∗ is a commutative domain; see §4.3 below. Since H∘ is a subalgebra of H∗,
we may take O=H∘ and so we have an integral action. Theorem 1 appears to be
new in this case.
0.5.
In Section 4, we show that if chark=0 and H is cocommutative, then the
core operator \,\text{\raisebox{-1.93747pt}{\bm{\cdot}}}\,\text{\sl:}H preserves semiprimess. Recall that
an ideal I of A is called semiprime if I is
an intersection of prime ideals or, equivalently,
A/I has no nonzero nilpotent ideals.
In Section 4, we turn to the question as to whether
I\sl:H is then semiprime as well.
This question may be reformulated in various alternative
ways (Lemma 15). In general, the answer is negative:
even for a cocommutative Hopf algebra H, semiprimeness may be lost upon
passage to the H-core. For instance, consider the group algebra
A=kG with its standard G-grading. For G finite, this grading amounts to an
action of the Hopf dual H=(kG)∗ on A. The only H-ideals of A
are [math] and A; so the H-core of every proper ideal of A is [math]. If G is abelian, then
H is cocommutative. If the operator \,\text{\raisebox{-1.93747pt}{\bm{\cdot}}}\,\text{\sl:}H
is to preserve semiprimeness for any such G, then we must require that
chark=0 by Maschke’s Theorem. It turns out that this is also sufficient in general:
Theorem 2**.**
Let A∈HAlg and assume that
H is cocommutative and chark=0.
Then I\sl:H is semiprime for every semiprime ideal I of A.
0.6.
In future work, we hope to pursue the general theme of this article for Hopf algebras
that are not necessarily cocommutative.
In particular, we plan to address “rationality” of prime ideals and explore the Dixmier-Mœglin
equivalence in the context of Hopf algebra actions, generalizing the work on group actions
in [11], [12].
Notations and conventions**.**
We work over a base field k and continue to write ⊗=⊗k.
Throughout, H is a Hopf k-algebra, cocommutative when so specified. The
counit and the antipode of H will be denoted by ε and S, respectively. Furthermore,
we fix the following notations for the remainder of this article:
A will be a left H-module algebra, with H-action h⊗a↦h.a(h∈H,a∈A);
AH={a∈A∣h.a=⟨ε,h⟩a for all h∈H} is
the subalgebra of H-invariants;
B=Homk(H,A) will denote the convolution algebra.
1. Generalities on actions
1.1. The convolution algebra B
We begin by introducing certain subalgebras and automorphisms of B as well as some H-operations
on B that will be used throughout this article.
First, the counit of H and the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA give rise to the following maps
ι,δ:A→B, which are easily
seen to be k-algebra embeddings:
[TABLE]
We will generally identify A with ιA and we will also identify the linear dual H∗
with the image of the algebra map
u∗:H∗↪B that is given by the unit u:k→A. In this way, we view A, H∗, and
A⊗H∗ as k-subalgebras of B, with A⊗H∗
consisting of all k-linear maps H→A that have finite rank.
Explicitly, a⊗f=(h↦a⟨f,h⟩), ιa=a⊗ε and u∗f=1⊗f
for a∈A, f∈H∗.
The algebra B becomes a left H-module algebra via the
⇀-action, which is defined by
[TABLE]
We will write (B,\text{\scriptsize\rightharpoonup}) when viewing B∈HAlg with (4).
The map \delta\colon A\to(B,\text{\scriptsize\rightharpoonup}) is a morphism in HAlg and A⊗H∗ is an H-module subalgebra
of (B,\text{\scriptsize\rightharpoonup}).
Our main focus later will be on the following alternative left H-module structure on B, which
takes into account the given action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA:
[TABLE]
If H is cocommutative, then B∈HAlg with
(5) as well; see Proposition 6 below.
In general, we may pass between (4) and (5) by means of the k-linear automorphisms
\Phi,\Psi\colon B\mathrel{\leavevmode\hbox to27.15pt{\vbox to8.2pt{\pgfpicture\makeatletter\hbox{\hskip 13.57596pt\lower-3.33301pt\hbox to0.0pt{\pgfsys@beginscope\pgfsys@invoke{ }\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\pgfsys@setlinewidth{0.4pt}\pgfsys@invoke{ }\nullfont\hbox to0.0pt{\pgfsys@beginscope\pgfsys@invoke{ }{{}}
{}{{}}{}
{}{{}}{}{}{}{}{{}}{}
{{}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{-10.24295pt}{0.0pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
{{}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{10.24295pt}{0.0pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
{
{}{}{}}{}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{
{}{}{}}{}{
{}{}{}}
{{{{{}}{
{}{}}{}{}{{}{}}}}}{}{{{{{}}{
{}{}}{}{}{{}{}}}}}{{}}{}{}{}{}{}{{{}{}}}{}{{}}{}{}{}{}{{}}\pgfsys@moveto{-6.70996pt}{0.0pt}\pgfsys@lineto{6.24997pt}{0.0pt}\pgfsys@stroke\pgfsys@invoke{ }{{}{{}}{}{}{{}}{{{}}{{{}}{\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{6.24997pt}{0.0pt}\pgfsys@invoke{ }\pgfsys@invoke{ \lxSVG@closescope }\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}{{}}}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{-3.8889pt}{0.7pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{{\sim}}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope{{
{}{}{}{}{}}}{}{}\hss}\pgfsys@discardpath\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope\hss}}\lxSVG@closescope\endpgfpicture}}}B that are defined by
[TABLE]
These maps are inverse to each other and they
satisfy the following “intertwining” formulas, for any a∈A, b∈B and h∈H:
[TABLE]
[TABLE]
The Ψ-identities follow from those for Φ=Ψ−1.
To verify (6), one checks that both sides send a given h∈H to
h_{1}.\big{(}a\,b(h_{2})\big{)}. For (7), observe that (Φb)(h)=(h⋅b)(1) and so
\Phi(h\bm{\cdot}\,b)(k)=(kh\bm{\cdot}\,b)(1)=(\Phi b)(kh)=(h\text{\scriptsize\rightharpoonup}(\Phi b))(k) for h,k∈H and b∈B.
Finally, note that Φ and Ψ fix the unit element 1B=u∗(ε) and they are right linear
for the subalgebra Homk(H,AH)⊆B.
In particular, Φ and Ψ restrict to the identity map on H∗ and are right H∗-linear.
1.2. Invariants and the locally finite part of (B,\text{\scriptsize\rightharpoonup})
The locally finite part of any H-module algebra A is defined by
[TABLE]
Here, “cofinite” is short for “having finite codimension.”
The locally finite part Afin is always
an H-module subalgebra of A containing the algebra of invariants, AH.
The following lemma determines the invariants and
the locally finite part of (B,\text{\scriptsize\rightharpoonup}).
{(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}}=\{b\in B\mid\operatorname{Ker}b\text{ contains some cofinite ideal of }H\}\cong A\otimes H^{\circ}.
Proof.
(a)
Note that b\in(B,\text{\scriptsize\rightharpoonup})^{H} if and only if b(kh)=⟨ε,h⟩b(k) for all h,k∈H. Equivalently,
b(h)=⟨ε,h⟩b(1) for all h∈H, which in turn states that b=ι(b(1)).
The assertion about H-invariants follows.
(b)
For any ideal I of H and any b∈B, the equality I\text{\scriptsize\rightharpoonup}b=0 is equivalent to I⊆Kerb;
so b\in{(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}} if and only if Kerb contains some cofinite ideal of H.
In particular, \iota A\subseteq{(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}} and the embedding u∗:H∗↪B
sends the finite dual H∘ to {(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}}.
The isomorphism between the subalgebra of B consisting of all finite-rank maps
and A⊗H∗ (§1.1) restricts to an
isomorphism of subalgebras, {(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}}\cong A\otimes H^{\circ}.
∎
The isomorphisms in Lemma 3 will be treated as identifications as in §1.1. In particular,
we will write A\otimes H^{\circ}={(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}}
and identify A and H∘ with the subalgebras ιA=A⊗ε
and u∗H∘=1⊗H∘, respectively.
As was mentioned, {(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}} is an H-module subalgebra of (B,\text{\scriptsize\rightharpoonup}); explicitly,
with b=a⊗f(a∈A,f∈H∘) formula (4) becomes
[TABLE]
Furthermore, {(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}}=A\otimes H^{\circ} is also stable under the H-operation (5),
which becomes the standard Hopf operation on tensor products:
[TABLE]
More generally, for any Hopf subalgebra O⊆H∘,
we will consider the subalgebra
[TABLE]
Each such AO is stable under both
⇀ and ⋅ by (8) and (9).
Lemma 4**.**
Let O⊆H∘ be a Hopf subalgebra.
The k-subspaces of AO
that are stable under right multiplication by O and under the
⇀-action (8) are exactly those of the form W⊗O,
where W is an arbitrary k-subspace of A.
Proof.
Certainly, each W⊗O is stable under right multiplication by O
and under (8).
For the converse, equip AO with the “trivial” right O-Hopf module structure:
the right O-module and right O-comodule structures are given by IdA⊗mO and
d=IdA⊗ΔO, where mO and ΔO are the multiplication and comultiplication of
O, respectively [14, Examples 10.3, 10.4].
Then AO↪AO⊗O↪Homk(H,AO) via d,
with (db)(h)=h\text{\scriptsize\rightharpoonup}b for b∈AO.
Now let V⊆AO be a k-subspace that is
stable under right multiplication by O and under (8). Then dV⊆Homk(H,V)∩(AO⊗O)=V⊗O and so V is a
O-Hopf submodule of AO. By the Structure Theorem for Hopf modules
[14, §10.1.2], V is generated as right O-module by the
subspace of O-coinvariants, VcoO={v∈V∣δv=v⊗1}.
Since VcoO⊆(AO)coO=A⊗OcoO=A⊗k,
it follows that V=W⊗O with W=VcoO.
∎
1.3. Coefficient Hopf algebras of locally finite actions
1.3.1.
Assume that the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA is locally finite, that is, A=Afin.
Then, since \delta\colon A\to(B,\text{\scriptsize\rightharpoonup}) is
a morphism in HAlg (§1.1), the image δA is contained in
{(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}}=A\otimes H^{\circ} and
A becomes a right H∘-comodule algebra via δ [14, Proposition 10.26].
Any Hopf subalgebra O⊆H∘ such that
δA is contained in the subalgebra A_{\mathcal{O}}=A\otimes\mathcal{O}\subseteq{(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}} will be called a
coefficient Hopf algebra for H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA.
We then have the following version of (2), which makes A a right O-comodule algebra:
[TABLE]
Any coefficient Hopf algebra for H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA will also
serve as a coefficient Hopf algebra for the H-action on quotients of A modulo H-ideals
and on H-module subalgebras of A, and the intersection of all coefficient Hopf subalgebras for H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA
is the unique smallest one.
Lemma 5**.**
Let H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA be locally finite and let O⊆H∘ be a coefficient
Hopf algebra. Then:
(a)
All subspaces of AO of the form
W⊗O, where W⊆A is an H-stable k-subspace, are stable under the maps
Φ,Ψ (§1.1).
2. (b)
If I is an ideal of A, then
I\sl:H=Ψ(I⊗O)∩A.
Proof.
(a)
If W⊆A is H-stable, then δW⊆W⊗O. Thus,
for any a∈W and f∈O,
right O-linearity of Φ and (6) now
give Φ(a⊗f)=δ(a)f=a0⊗a1f∈W⊗O. Similarly, Ψ(a⊗f)=a0⊗S∗(a1)f∈W⊗O ,
proving stability of W⊗O under Φ and Ψ.
(b)
By (1), I\sl:H=δ−1(Homk(H,I))=δ−1(I⊗O).
Putting I′=Ψ(I⊗O) and using the identity Φ∘ι=δ from (6),
we obtain I\sl:H=δ−1(I⊗O)=(ι)−1(I′)=I′∩A.
∎
1.3.2.
The action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA is certainly locally finite if the H-module algebra
A under consideration is generated as k-algebra by a finite-dimensional
H-stable subspace V⊆A. Assume this to be the case and
let ρ:H→Endk(V) denote
the algebra map given by the operation of H on V.
Fixing a k-basis (vi)1n of V and
letting (vi∗) denote the dual basis of V∗, we obtain the basis (vi⊗vj∗) of
Endk(V)≅V⊗V∗ and linear forms ρi,j such that
[TABLE]
The isomorphism Endk(V)≅Mn(k) given by our choice of basis for V
allows us to write ρ=(ρi,j):H→Mn(k): the scalar ⟨ρi,j,h⟩∈k
is the (i,j)-entry of the matrix ρh. Thus, ρi,j∈H∘, because
ρi,j vanishes on the cofinite ideal Kerρ, and
Δρi,j=∑kρi,k⊗ρk,j.
On the generating subspace V⊆A, the algebra map (10) takes the form
[TABLE]
and we may take O to be the Hopf subalgebra of H∘ that is generated by the
matrix coefficient functions ρi,j.
If H is involutory, then O is the k-subalgebra of H∘ that is generated
by all ρi,j and S∗ρi,j.
We will call O the coefficient Hopf algebra of the
representation V∈RepH; a more general situation is discussed in [14, Exercise 9.2.3].
2. The cocommutative case
2.1. The ⋅-action
2.1.1.
We recall some general ring-theoretic material that will be tacitly used
in the next proposition and throughout the remainder of this article.
A ring homomorphism f:R→S is called centralizing if the ring S is generated
by the image fR and its centralizer in S. If f makes S a free R-module having a basis
that centralizes fR, then f is called free centralizing.
Any centralizing homomorphism f restricts to a map of centers, ZR→ZS;
the assignment I↦(fI)S sends (two-sided) ideals of R to ideals of S; and
P↦f−1P gives a well-defined map SpecS→SpecR [11, §1.5].
2.1.2.
The proposition below shows that, for H cocommutative,
we may view B and various subalgebras of B as H-module algebras with
action (5), which we will refer to as the ⋅-action, rather than the
⇀-action (4). When using the latter H-action, we
will continue write (B,\text{\scriptsize\rightharpoonup}); otherwise, the ⋅-action is assumed.
Proposition 6**.**
Let H be cocommutative and let O⊆H∘ be a coefficient Hopf algebra
for the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureAfin . Then:
(a)
Φ,Ψ* are algebra automorphisms of B stabilizing the subalgebra Afin,O:=Afin⊗O.*
2. (b)
B∈HAlg* with the ⋅-action and BH=ΨA.*
3. (c)
AO* is an H-module subalgebra of B, with (AO)fin=Afin,O and
(AO)H=Ψ(Afin). The inclusion (AO)H⊆(AO)fin
is free centralizing.*
4. (d)
The following maps are bijections that are inverse to each other:
[TABLE]
Proof.
(a)
For b,b′∈B and h∈H, one computes using cocommutativity,
[TABLE]
Thus Φ is multiplicative and hence it is
an algebra automorphisms of B; likewise for Ψ=Φ−1. Stability of
the subalgebra Afin,O under Φ and Ψ
follows from Lemma 5(a) for Afin.
(b)
Since Ψ is an algebra automorphism of B,
the intertwining formula (7) together with the fact that
(B,\text{\scriptsize\rightharpoonup})\in{}_{H}\!\operatorname{\mathsf{Alg}} implies that
B=(B,⋅)∈HAlg as well. Also by (7), Ψ yields
an algebra isomorphism (B,\text{\scriptsize\rightharpoonup})^{H}\cong B^{H} and so
Lemma 3(a) gives the isomorphism
\Psi\circ\iota\colon A\mathrel{\leavevmode\hbox to27.15pt{\vbox to8.2pt{\pgfpicture\makeatletter\hbox{\hskip 13.57596pt\lower-3.33301pt\hbox to0.0pt{\pgfsys@beginscope\pgfsys@invoke{ }\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\pgfsys@setlinewidth{0.4pt}\pgfsys@invoke{ }\nullfont\hbox to0.0pt{\pgfsys@beginscope\pgfsys@invoke{ }{{}}
{}{{}}{}
{}{{}}{}{}{}{}{{}}{}
{{}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{-10.24295pt}{0.0pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
{{}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{10.24295pt}{0.0pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
{
{}{}{}}{}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{
{}{}{}}{}{
{}{}{}}
{{{{{}}{
{}{}}{}{}{{}{}}}}}{}{{{{{}}{
{}{}}{}{}{{}{}}}}}{{}}{}{}{}{}{}{{{}{}}}{}{{}}{}{}{}{}{{}}\pgfsys@moveto{-6.70996pt}{0.0pt}\pgfsys@lineto{6.24997pt}{0.0pt}\pgfsys@stroke\pgfsys@invoke{ }{{}{{}}{}{}{{}}{{{}}{{{}}{\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{6.24997pt}{0.0pt}\pgfsys@invoke{ }\pgfsys@invoke{ \lxSVG@closescope }\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}{{}}}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{-3.8889pt}{0.7pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{{\sim}}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope{{
{}{}{}{}{}}}{}{}\hss}\pgfsys@discardpath\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope\hss}}\lxSVG@closescope\endpgfpicture}}}(B,\text{\scriptsize\rightharpoonup})^{H}\mathrel{\leavevmode\hbox to27.15pt{\vbox to8.2pt{\pgfpicture\makeatletter\hbox{\hskip 13.57596pt\lower-3.33301pt\hbox to0.0pt{\pgfsys@beginscope\pgfsys@invoke{ }\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\pgfsys@setlinewidth{0.4pt}\pgfsys@invoke{ }\nullfont\hbox to0.0pt{\pgfsys@beginscope\pgfsys@invoke{ }{{}}
{}{{}}{}
{}{{}}{}{}{}{}{{}}{}
{{}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{-10.24295pt}{0.0pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
{{}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{10.24295pt}{0.0pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
{
{}{}{}}{}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{
{}{}{}}{}{
{}{}{}}
{{{{{}}{
{}{}}{}{}{{}{}}}}}{}{{{{{}}{
{}{}}{}{}{{}{}}}}}{{}}{}{}{}{}{}{{{}{}}}{}{{}}{}{}{}{}{{}}\pgfsys@moveto{-6.70996pt}{0.0pt}\pgfsys@lineto{6.24997pt}{0.0pt}\pgfsys@stroke\pgfsys@invoke{ }{{}{{}}{}{}{{}}{{{}}{{{}}{\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{6.24997pt}{0.0pt}\pgfsys@invoke{ }\pgfsys@invoke{ \lxSVG@closescope }\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}{{}}}}\hbox{\hbox{{\pgfsys@beginscope\pgfsys@invoke{ }{{}{}{{
{}{}}}{
{}{}}
{{}{{}}}{{}{}}{}{{}{}}
{
}{{{{}}\pgfsys@beginscope\pgfsys@invoke{ }\pgfsys@transformcm{1.0}{0.0}{0.0}{1.0}{-3.8889pt}{0.7pt}\pgfsys@invoke{ }\hbox{{\definecolor{pgfstrokecolor}{rgb}{0,0,0}\pgfsys@color@rgb@stroke{0}{0}{0}\pgfsys@invoke{ }\pgfsys@color@rgb@fill{0}{0}{0}\pgfsys@invoke{ }\hbox{{\sim}}
}}\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope}}}
\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope{{
{}{}{}{}{}}}{}{}\hss}\pgfsys@discardpath\pgfsys@invoke{\lxSVG@closescope }\pgfsys@endscope\hss}}\lxSVG@closescope\endpgfpicture}}}B^{H}. Identifying A with ιA=A⊗ε as usual,
we obtain the claimed equality BH=ΨA.
(c)
The subalgebra AO⊆B is stable under the ⋅-action by (9) and
so it is an H-module subalgebra of B. The equality (AO)fin=Afin,O
follows from [10, Corollary 6], because the H-action ⇀ on O is
locally finite. Since S is an involution of H, the formula
(Ψa)(h)=S(h1).a⟨ε,h2⟩=S(h).a(a∈A,h∈H)
shows that Ψa
vanishes on some cofinite ideal of H if and only if a∈Afin. In that case,
Ψa∈Afin,O⊆AO by (a). Thus, in view of part (b) and Lemma 3(b),
\Psi({A}_{\text{\rm fin}})=B^{H}\cap{(B,\text{\scriptsize\rightharpoonup})}_{\text{\rm fin}}=(A_{\mathcal{O}})^{H} as asserted.
Finally, since (AO)fin=Afin,O=Ψ(Afin,O) is generated by the commuting subalgebras
Ψ(Afin)=(AO)H and ΨO, with ΨO providing
an (AO)H-basis of (AO)fin, the inclusion
(AO)H⊆(AO)fin is free centralizing.
(d)
Let J be an ideal of (AO)H. Since the inclusion
(AO)H⊆(AO)fin=Afin,O is centralizing by (c), JAfin,O is an
ideal of Afin,O, clearly an H-ideal. Further,
JAfin,O∩(AO)H=J by freeness of Afin,O over (AO)H.
Now let L be any H-ideal of Afin,O. Then ΦL is an ideal of Afin,O that is
stable for the ⇀-action by (7). Thus, Lemma 4 implies that
ΦL=I⊗O for some ideal I of Afin.
Therefore, L=Ψ(I⊗O)=ΨIΨO=(ΨI)Afin,O
with ΨI being an ideal of (AO)H=Ψ(Afin). This proves that extension and contraction
give inverse bijections between the sets
of ideals of (AO)H and H-ideals of Afin,O. The bijection between the sets of ideals of
(AO)H and of Afin is a consequence of the equality Ψ(Afin)=(AO)H.
∎
2.2. Integral actions
2.2.1.
A locally finite action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA will be called integral if it has a coefficient Hopf algebra that is
a commutative integral domain. Even though H need not a priori be cocommutative,
any integral H-action factors through a cocommutative Hopf quotient of H.
Proposition 7**.**
Let H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA be integral and k algebraically closed. Then P\sl:H is prime
for every P∈SpecA. Furthermore, H-SpecA is the set of all prime H-ideals of A.
Proof.
Let O⊆H∘ be a coefficient Hopf algebra of the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA that is an integral domain,
and assume that H is cocommutative, as we may. Then Ψ
is an algebra automorphism of B that restricts to
an automorphism of the subalgebra AO (Proposition 6). For any
P∈SpecA, the ideal P⊗O of AO is prime [14, Lemma 11.19],
and hence
[TABLE]
By Lemma 5(b), P\sl:H=P′∩A, which is a prime ideal of A,
because A↪AO
is centralizing. The final assertion also follows, because
the map SpecA→H-SpecA , P↦P\sl:H, is surjective for any locally
finite action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA [14, Exercise 10.4.4].
∎
2.2.2.
Returning to the situation considered in §1.3.2, assume that
A is affine, generated by an
H-stable subspace V⊆A with n=dimkV<∞. We use our earlier notation, but we
now also assume that H is cocommutative and so involutory.
The k-subalgebra O⊆H∘ that is generated
by the functions ρi,j and S∗ρi,j is thus a coefficient Hopf algebra for
H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA.
Example 8** (Group algebras).**
Let H=kG be a group algebra. Then H∗ is the algebra kG of all functions G→k with pointwise
addition and multiplication. The subalgebra H∘ is commonly
referred to as the algebra of representative functions on G
and denoted by Rk(G) (e.g., [8, Chapter 1]).
For any g∈G, the determinant detρg is nonzero
and ⟨S∗ρi,j,g⟩=⟨ρi,j,g−1⟩=detρg1⟨cj,i,g⟩,
where ⟨cj,i,g⟩ denotes the (j,i)-cofactor of the matrix ρg∈GLn(k), a polynomial in the
entries of ρg. Thus, cj,i∈R:=k[ρi,j∣i,j=1,…,n] and
S∗ρi,j=detρcj,i∈R[(detρ)−1],
the subalgebra of H∘ that is generated by the
functions ρi,j and (detρ)−1=S∗detρ. We obtain
[TABLE]
Now let k be algebraically closed.
The group GLn(k) is affine algebraic, with associated Hopf algebra O(GLn) as in [14, Example 9.19].
The Hopf algebra O is the image of O(GLn) under restriction
to ρG≤GLn(k). Finally, O is a domain if and only if ρG is an irreducible subset
of GLn(k) in the Zariski topology or, equivalently, the closure ρG is a connected affine
algebraic group.
Example 9** (Enveloping algebras).**
Let g be a Lie k-algebra and H=Ug its enveloping algebra.
If chark=0, then H∗ is a (commutative) domain;
see [4, Chap. II, §1, no 5] or Section 4.3 below.
Therefore, the subalgebra H∘ and all coefficient Hopf algebras O are
integral domains as well in characteristic [math].
The situation is different for chark=p>0.
Indeed, in this case, ⟨fp,g⟩=0 for any f∈H∗. If f∈H∘ is grouplike
or, equivalently, an algebra map, then so is fp
and hence Kerfp is the ideal of H that is generated by g.
Therefore, fp=ε and so (f−ε)p=0. If g=[g,g] then we may choose f=ε, giving
a nonzero nilpotent element of H∘.
2.3. The symmetric ring of quotients and the extended center
2.3.1.
We briefly recall some background material
on the symmetric ring of quotients QR of an arbitrary ring R and
its center, CR:=Z(QR), the so-called extended center of R. See
[14, Appendix E] for details. The ring R is a subring of QR and CR
coincides with the centralizer of R in QR. In particular, ZR⊆CR.
If ZR=CR, then R is called centrally closed. In general, we may consider
the following subring of QR, possibly strictly larger than R:
[TABLE]
If R is semiprime, then R is a centrally closed ring
[2, Theorem 3.2], called the central closure of R.
If R is a k-algebra, then so is R, because ZR⊆CR=ZR.
2.3.2.
The next lemma concerns the extension of a given action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA to an action on QA.
Part (a), which summarizes known facts, shows that this is
always possible, in a unique way, in the situation that we are interested in. Indeed, any cocommutative
Hopf algebra over an algebraically closed field is pointed [19, Lemma 8.0.1]. However, local
finiteness of an action may be lost in the process. For instance, consider the action of the
group k× on the polynomial algebra A=k[x] that is determined by λ.x=λx
for λ∈k× and the extended action on
the field QA=FractA=k(x). If k is infinite, then k(x)fin=k[x±1]. Part (b) of the lemma focuses on
the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpicture(QA)fin rather than H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureQA.
Lemma 10**.**
Let H be pointed cocommutative. Then:
(a)
The H-action on A extends uniquely to an action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureQA and this action
stabilizes the subalgebras A=A(CA) and CA.
2. (b)
If the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA is locally finite with coefficient Hopf algebra O, then the extended action
H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpicture(QA)fin also has coefficient Hopf algebra O.
Proof.
(a)
Since H is pointed, [16, Cor. 3.5] tells us
that the H-action on A extends uniquely to an action on QA.
This action stabilizes the center, CA, because H is cocommutative [5, Prop. 4].
Therefore, A is stable as well.
(b)
Put R=(QA)fin. So A⊆R and the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureQA restricts to a locally finite
action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureR extending the H-action on A.
We must show that the map δ:R→R⊗H∘
(§1.3) has image in RO=R⊗O, given that δA⊆AO.
Fix r∈R and consider the subspace V=H.r⊆R, which is
finite dimensional and H-stable. Therefore,
the ideal I={a∈A∣aAV⊆A} of A has zero (left and) right annihilator in R
[14, Proposition E.1] and I is an H-ideal as is readily verified using the
following standard identity for H-module algebras:
[TABLE]
By Lemma 5(a), Φ(I⊗O)=I⊗O.
Furthermore, by Proposition 6(a), Φ gives an algebra automorphism of
R⊗H∘ stabilizing AO and satisfying Φ∘ι=δ by (6).
Therefore,
[TABLE]
Thus, (I⊗ε)δr⊆RO .
Since δr∈R⊗H∘ and I has zero right annihilator in R, it follows that
δr∈RO, as desired.
∎
3. Prime strata
3.1. Prime correspondences
The central closure R of a prime ring R is also prime and its center,
the extended center CR, is a field. Thus, for an arbitrary ring R, we may
associate to any P∈SpecR the field C(R/P),
called the heart of P.
Proposition 11**.**
Let R be a centrally closed prime k-algebra and put K=CR. Let S be any
K-algebra and put U=R⊗KS. Then we have the following bijections, which are inverse to each other:
[TABLE]
This correspondence preserves hearts: C(U/P)≅C(S/P∩S).
Moreover, if U∈HAlg with S being an H-module subalgebra,
then H-ideals are matched with H-ideals.
Proof.
Apart from the last assertion, involving an H-action, the proposition is identical with
[12, Proposition 5]. The bijections in the proposition,
contraction and extension of ideals, both evidently send H-ideals to H-ideals in the given situation.
∎
Proposition 12**.**
Assume that H is pointed cocommutative and let R,T∈HAlg, with
R being prime. Put V:=R⊗T⊆V:=R⊗T, where
R denotes the central closure of R. Then there is a bijection
[TABLE]
This bijection is an order isomorphism for ⊆ and it preserves hearts.
Moreover, viewing R∈HAlg with
the extended H-action (Lemma 10) and V,V∈HAlg with
the standard H-action on tensor products, the bijection gives a bijection on H-stable primes.
Proof.
Again, this proposition is covered by [12, Proposition 6] except for the statement about H-stability.
To justify this assertion, note that tensor products of H-module algebras are again H-module algebras,
with the standard H-action via Δ, because H is cocommutative.
Thus, V∈HAlg and all of R, R, T and V are H-module subalgebras of V.
If P is an H-ideal of V, then its contraction, P∩V, is clearly an H-ideal of V.
Conversely, let Q∈SpecV with Q∩R=0 be H-stable. In order to show that
the preimage of Q under the bijection in the
proposition is an H-ideal of V, we recall the construction of
the preimage from [12].
The canonical epimorphism
π:V↠W:=V/Q is a map in HAlg, and the restriction
\rho:=\pi\big{|}_{R}\colon R\hookrightarrow W is an embedding of prime algebras and a centralizing map in HAlg.
By [12, Lemma 4], ρ extends uniquely to an embedding of central closures,
ρ:R↪W, which is also centralizing.
Claim**.**
ρ is a map in HAlg for the extended action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureW (Lemma 10).
To prove the claim, let
r∈R and h∈H. Fix a nonzero ideal I of R such that (h.r)I and all r(S(h2).I)
are contained in R; this is possible by continuity of the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureR [16, Proposition 2.2].
Using H-equivariance of ρ one computes, with x∈I,
[TABLE]
where the last equality uses the identity (14).
Thus, (ρ(h.r)−h.ρr)ρI=0. Since
(ρI)W is a nonzero ideal of W, it follows that ρ(h.r)=h.ρr, proving the claim.
The image πT⊆W⊆W
centralizes ρR and hence also ρR=ρRρ(ZR),
because ρ(ZR)⊆ZW.
Therefore, ρ and \pi\big{|}_{T} give a homomorphism
π:V=R⊗T→W. The preimage of Q in Proposition 12 is Kerπ; see
[12, proof of Proposition 6].
Finally, since ρ and \pi\big{|}_{T} are maps in HAlg,
it follows that π is likewise. Hence Kerπ is an H-ideal, as desired.
∎
Let H be a cocommutative Hopf algebra over an algebraically closed field k and
let H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA be an integral action. Fix a coefficient Hopf algebra O⊆H∘
that is an integral domain and write AO=A⊗O as before.
Given I∈H-SpecA, our goal is to describe the stratum
SpecIA. Replacing A by A/I, we may assume that
A is prime (Proposition 7) and focus on the set Spec0A={P∈SpecA∣P\sl:H=0}.
3.2.1.
Put X:={Q∈SpecAO∣Q∩A=0}.
We first establish a bijection between Spec0A and the subset
XH⊆X consisting of all Q∈X that are stable for the ⋅-action (9).
Recall that Ψ gives
an automorphism of the algebra AO and,
for any P∈SpecA, we have P′:=Ψ(P⊗O)∈SpecAO;
see Proposition 6 and (13). If P∈Spec0A, then P′∈X
by Lemma 5(b). So we have a map Spec0A→X, P↦P′, which
is evidently injective. By Proposition 6(d), the image consists of H-ideals; so P′∈XH.
Conversely, let Q∈XH be given and put P=Φ(Q)∩A;
this is a prime ideal of A, because Φ(Q)∈SpecAO and the embedding
A↪AO is centralizing. Also, Lemma 5(b) gives P\sl:H=P′∩A⊆Q∩A=0; so P∈Spec0A. Finally, Q=P′ by Proposition 6(d).
Thus, we have the desired bijection:
[TABLE]
This bijection has the following properties, with (i) being evident and
(ii) resulting from the algebra isomorphism AO/P′≅AO/(P⊗O)≅(A/P)⊗O that comes from Ψ:
(i)
P1⊆P2 if and only if P1′⊆P2′, and
2. (ii)
C(AO/P′)≅C((A/P)⊗O).
3.2.2.
Now put K=CA and C=K⊗O; the former is a k-field and the latter a commutative
integral domain, identical to the algebra C0 in (3).
Let A=AK denote the central closure of A
and put AO=A⊗O≅A⊗KC. Then we have the following
isomorphisms of posets (for ⊆),
with d1 coming from Proposition 12 and d2 from Proposition 11:
Since i′ and both di are order isomorphisms and
preserve hearts, the same holds for c:
(i)
c(P1)⊆c(P2) if and only if P1⊆P2, and
2. (ii)
Fract(C/c(P))=C(C/c(P))≅C((A/P)⊗O).
Finally, Imc consists exactly of the H-ideals in SpecC. This amounts to
checking that Q∈X is an H-ideal if and only if d(Q)∈SpecC is an H-ideal.
But, by Lemma 10(a) and Proposition 6(c),
AO∈HAlg for the ⋅-action,
with AO and C=Z(AO) being H-module subalgebras.
By Propositions 11 and 12, we know that a given
P∈SpecAO with P∩A=0 is an H-ideal if and only if
d2(P)=P∩C∈SpecC is an H-ideal and if and only if
d1−1(P)=P∩AO∈X is an H-ideal.
This completes the proof of Theorem 1. ∎
3.3. Tensor algebras
As an application of Theorem 1, we offer the following result on the tensor algebra TV
of a finite-dimensional representation V∈RepH. The H-action on V extends uniquely to an
action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureTV [14, 10.4.2].
Corollary 13**.**
Let H be a cocommutative Hopf algebra over an algebraically closed field k and
let V be a representation of H with 2≤dimkV<∞ and such
that the coefficient Hopf algebra of V (§1.3.2) is a domain.
Then every nonzero prime ideal of the tensor algebra TV contains a nonzero H-stable prime ideal.
Proof.
The action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA=TV is integral and we may apply
Theorem 1 with I=0. By a result of Kharchenko [9], QA=A
and so CA=ZA=k. Therefore, the algebra C0 in Theorem 1 coincides
with O, the coefficient Hopf algebra of the representation V. Since C0=O is
H-simple by Lemma 4, we have SpecHC0={0} and Theorem 1
gives Spec0A={0}. Consequently, if P is a nonzero prime ideal of A, then P∈/Spec0A
and so P\sl:H=0. Finally, P\sl:H is an H-stable prime ideal by Proposition 7.
∎
3.4. Torus actions
For rational torus actions, the set SpecHCI in Theorem 1 has a simpler
description as SpecZI for an explicit commutative domain ZI. This was shown in [13],
but the presentation contains some inaccuracies which we will now repair.
To start with, let H be pointed cocommutative and consider an arbitrary locally finite action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureA
with coefficient Hopf subalgebra
O⊆H∘. Associated to any I∈H-SpecA,
we have the commutative algebra CI=C(A/I)⊗O∈HAlg
as in (3); the H-action on CI is (9) using the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureC(A/I) from
Lemma 10(a). This action need not be locally finite (§2.3.2).
Following [13], we consider the locally finite part and the invariants:
[TABLE]
The invariant algebra FI is in fact a field [15, Lemma 1.4].
By Lemma 10(b), the action H\leavevmodeto18.09pt\vboxto17.21pt\pgfpicture\makeatletter\lower-9.9692ptto0.0pt\pgfsys@beginscope\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@setlinewidth0.4pt\pgfsys@invoke\nullfontto0.0pt\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm1.00.00.01.00.0pt-1.36568pt\pgfsys@invoke\definecolorpgfstrokecolorrgb0,0,0\pgfsys@color@rgb@stroke000\pgfsys@invoke\pgfsys@color@rgb@fill000\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@beginscope\pgfsys@invoke\pgfsys@moveto-3.533pt-3.40517pt\pgfsys@curveto-14.55519pt-9.7692pt-14.55519pt7.03784pt-4.79736pt1.40381pt\pgfsys@stroke\pgfsys@invoke\pgfsys@beginscope\pgfsys@invoke\pgfsys@transformcm0.86601-0.500020.500020.86601-4.79738pt1.40381pt\pgfsys@invoke\pgfsys@invoke\lxSVG@closescope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\pgfsys@discardpath\pgfsys@invoke\lxSVG@closescope\pgfsys@endscope\hss\lxSVG@closescope\endpgfpictureZI has coefficient Hopf algebra O
and Proposition 6(c) gives the following isomorphism which
generalizes [13, Equation (2)]:
[TABLE]
Turning to torus actions, let k be algebraically closed and let
G be an algebraic k-torus acting rationally on the k-algebra A.
As was remarked in §0.4, the algebra of polynomial functions O=O(G) serves as a
coefficient Hopf algebra for this action. Moreover, O is a Laurent polynomial algebra over k, the algebra
CI is a Laurent polynomial algebra over the field C(A/I), and
the isomorphism ZI≅(CI)G in (15) realizes ZI as a
Laurent polynomial algebra over the field of G-invariants FI=C(A/I)G; see
the Stratification Theorem in [13] or [13, Equation (4)].
The set SpecHCI in Theorem 1 now is the set SpecGCI consisting of all
G-stable prime ideals of CI and Spec(CI)G≅SpecZI by (15).
By [13, Equation (5)], all G-stable ideals of S:=CI are generated
by their intersection with R:=SG. Therefore, the contraction
map SpecGS→SpecR, P↦P∩R, is injective.
If a is an ideal of R, then its extension a:=aS is a G-stable ideal of S.
Moreover, a∩R=a, because S is free over R by [13, 2.3]. Thus, a is the
only G-stable ideal of S that contracts to a. Now let p∈SpecR be given and
choose an ideal P of S maximal subject to the condition P∩R=p. Then P is prime
and hence so is its G-core by Proposition 7.
Thus, P\sl:G∈SpecGS and P\sl:G=p by the foregoing.
This proves surjectivity of the contraction map and that the inverse is given by extension.
The statements about G-equivariance and preservation of inclusions in the lemma are clear.
∎
4. Semiprimeness
4.1. Reformulations
We start this section by giving several reformulations, in terms of the semiprime radical
operator \sqrt{\,\text{\raisebox{-1.93747pt}{\bm{\cdot}}}\,\phantom{!}}, of the conclusion of Theorem 2,
which is equivalent to (ii) in the lemma below.
The semiprime radical of a subset X⊆A, by definition,
is the unique smallest semiprime ideal of A containing X:
[TABLE]
We continue to assume that A∈HAlg; the Hopf algebra H can be arbitrary for now.
Lemma 15**.**
The following are equivalent:
(i)
If J is an H-ideal of A, then so is J;
2. (ii)
for all ideals I of A, the H-core I\sl:H is semiprime;
3. (iii)
H.I⊆H.I* for any ideal I of A.*
Proof.
(i) ⇒ (ii).
We may assume that I is semiprime. Then I\sl:H⊆I=I, since
\sqrt{\,\text{\raisebox{-1.93747pt}{\bm{\cdot}}}\,\phantom{!}} preserves inclusions. In fact, I\sl:H⊆I\sl:H,
because I\sl:H is an H-ideal by (i).
The reverse inclusion being trivial, it follows that I\sl:H=I\sl:H is semiprime.
(ii) ⇒ (iii).
Let J denote the ideal of A that is generated by the subset H.I⊆A. Then
I⊆J=H.I and J
is easily seen to be an H-ideal. (If the antipode S is bijective, then J=H.I [14, Exercise 10.4.3].)
Thus, J=J\sl:H⊆J\sl:H and the latter ideal
is semiprime by (ii). It follows that J=J\sl:H⊆J\sl:H.
Again, the reverse inclusion is clear; so J=J\sl:H.
Therefore, H.I⊆H.J=J=H.I.
(iii) ⇒ (i). Specialize (iii) to the case where I=J is an H-ideal.
∎
4.2. Extending the base field
For the proof of Theorem 2, we may work over an algebraically closed base field.
This follows by taking K to be an algebraic closure of k in the argument below.
Let K/k be any field extension and put H′=H⊗K
and A′=A⊗K. Then A′∈H′Alg and
H′ is cocommutative if H is so. Assuming Theorem 2 to hold for A′, our goal is to show
that it also holds for A. So let I be a semiprime ideal of A.
Viewing A as being contained in A′ in the usual way, IA′ is an ideal of A′ satisfying
IA′∩A=I. By Zorn’s Lemma, we may choose an ideal I′ of A′ that is
maximal subject to the condition I′∩A=I. Then I′ is semiprime. For, if J is any ideal of A′
such that J⫌I′, then J∩A⫌I by maximality of I′, and so (J∩A)2⊈I by semiprimeness of I. Since (J∩A)2⊆J2∩A, it follows that J2⊈I′, proving that I′ is semiprime. Therefore, by our assumption, the core I′\sl:H′ is
semiprime. Since the extension A↪A′ is centralizing, it follows that (I′\sl:H′)∩A
is a semiprime ideal of A. Finally,
(I′\sl:H′)∩A={a∈A∣H′.a⊆I′}={a∈A∣H.a⊆I′∩A=I}=I\sl:H
by (1), giving the desired conclusion that I\sl:H is semiprime.
4.3. Enveloping algebras
4.3.1.
For any ring R, let R[[Xλ]]λ∈Λ
denote the ring of formal power series in the
commuting variables Xλ(λ∈Λ)
over R; see [3, Chap. III, §2, no 11].
Lemma 16**.**
Let R be a ring,
let Λ be any set, and let S
be a subring of R[[Xλ]]λ∈Λ
such that S maps onto R under the homomorphism
R[[Xλ]]λ∈Λ→R,
Xλ↦0. If R is prime (resp., semiprime, a domain) then so is S.
Proof.
We write monomials in the variables Xλ as
Xn=∏λXλn(λ)(n∈M), where
M=Z+(Λ) denotes the additive monoid of all functions
n:Λ→Z+ such that n(λ)=0 for almost all λ∈Λ.
Fix a total order < on M
having the following properties (e.g., [1, Example 2.5]):
every nonempty subset of M has a smallest element; the zero function 0
is the smallest element of M; and
n<m implies n+r<m+r for all n,m,r∈M.
For any 0=s=∑n∈MsnXn∈R[[Xλ]]λ∈Λ,
we may consider its lowest coefficient, smin:=sm with
m=min{n∈M∣sn=0}.
If R is prime and 0=s,t∈S are given, then
0=sminrtmin for some r∈R.
By assumption, there exists an element u∈S having the form
u=r+∑n=0unXn. It follows that sut=0, with
(sut)min=sminrtmin. This proves that S
is prime. For the assertions where R is semiprime or a domain, take s=t or r=1,
respectively.
∎
4.3.2.
Now let H=Ug be the enveloping algebra of an arbitrary Lie k-algebra g and assume that chark=0.
For the proof of the next proposition, we recall the structure of the convolution algebra Homk(H,R) for
an arbitrary k-algebra R. Let (eλ)λ∈Λ be a
k-basis of g and fix a total order of the index set Λ. Put
M=Z+(Λ) as in the proof of Lemma 16 and,
for each n∈M, put
en=∏λ<n(λ)!1eλn(λ)∈H, where the superscript <
indicates that the factors occur in the order of increasing λ. The elements en form
a k-basis of H by the Poincaré-Birkhoff-Witt Theorem, and the comultiplication of
H is given by
Δen=∑r+s=ner⊗es; see [14, Example 9.5].
Writing Xn=∏λXλn(λ) as in the proof of Lemma 16, we obtain
an isomorphism of k-algebras,
[TABLE]
Under this isomorphism, the algebra map
u∗:Homk(H,R)→R , f↦f(1) , coming from the
unit map u=uH:k→H translates into the map
R[[Xλ]]λ∈Λ↠R,
Xλ↦0, as considered in Lemma 16.
The proposition below is not new; it can be found in Dixmier’s book
[6, 3.3.2 and 3.8.8], albeit with a very different proof.
Recall that an ideal of a ring is said to be completely prime if the quotient
is a domain.
Proposition 17**.**
Let H=Ug be the enveloping algebra of a Lie k-algebra g,
let A∈HAlg, and let I be an ideal of A. Assume that chark=0. If
I is prime, semiprime or completely prime, then I\sl:H is
likewise.
Proof.
By (1), the core I\sl:H is identical to the kernel of the map
δI:A→Homk(H,A/I) that is given by δI(a)=(h↦h.a+I).
We need to show that the properties of being prime, semiprime, or a domain
all transfer from A/I to the subring δIA⊆Hom(H,A/I) or, equivalently, to the subring
S⊆(A/I)[[Xλ]]λ∈Λ that corresponds to δIA
under the above isomorphism φ. Consider the map u∗:Homk(H,A/I)→A/I ,
f↦f(1), and note that (u∗∘δI)(a)=a+I for a∈A.
Therefore, S maps onto A/I
under the map (A/I)[[Xλ]]λ∈Λ→A/I,
Xλ↦0. Now all assertions follow from Lemma 16.
∎
4.3.3.
For an arbitrary cocommutative Hopf algebra H, we cannot expect a result as strong as
Proposition 17: group algebras provide easy counterexamples
to the primeness and complete primeness assertions.
Indeed, let H=kG be the group k-algebra of the group G and let A∈HAlg.
Then I\sl:H=⋂g∈Gg.I for any ideal I of A. If I is
semiprime, then so are all g.I, because each g∈G acts on A by algebra automorphisms, and
hence ⋂g∈Gg.I will be semiprime also. However, primeness and complete primeness, while
inherited by each g.I, are generally
lost upon taking the intersection.
Let H be cocommutative Hopf and assume that chark=0 and that k is
algebraically closed, as we may by §4.2.
Then H has the structure of a smash product,
H≅U#V, where U is the enveloping algebra of the Lie algebra
of primitive elements of H and V is the group algebra of the group of grouplike
elements of H; see [19, §13.1] or [17, §15.3].
Thus, both U and V are Hopf subalgebras of H and
H=UV, the k-space spanned by all products uv with u∈U and v∈V.
Viewing A∈UAlg and A∈VAlg by restriction,
repeated application of (1) gives the following equality for any ideal I of A:
[TABLE]
If I is semiprime, then so is I\sl:U (Proposition 17). Our remarks
on group algebras in the first paragraph of this proof further give semiprimeness of (I\sl:U)\sl:V.
Thus, I\sl:H is semiprime and Theorem 2 is proved. ∎
Acknowledgment**.**
We thank the referee for pointing out reference [6] in
connection with Proposition 17 above.
Bibliography19
The reference list from the paper itself. Each links out to its DOI / PubMed record.
1[1] Matthias Aschenbrenner and Christopher J. Hillar, Finite generation of symmetric ideals , Trans. Amer. Math. Soc. 359 (2007), no. 11, 5171–5192. MR 2327026 (2008 g:13030)
2[2] W. E. Baxter and W. S. Martindale, III, Jordan homomorphisms of semiprime rings , J. Algebra 56 (1979), no. 2, 457–471. MR 528587 (80f:16008)
3[3] Nicolas Bourbaki, Algèbre. Chapitres 1 à 3 , Hermann, Paris, 1970. MR 43 #2
4[4] by same author, Groupes et algèbres de Lie. Chapitre II: Algèbres de Lie libres. Chapitre III: Groupes de Lie , Éléments de mathématique. Fasc. XXXVII., Hermann, Paris, 1972, Actualités Scientifiques et Industrielles, No. 1349. MR 58 #28083 a
6[6] Jacques Dixmier, Enveloping algebras , Graduate Studies in Mathematics, vol. 11, American Mathematical Society, Providence, RI, 1996, Revised reprint of the 1977 translation. MR 1393197 (97c:17010)
7[7] Kenneth R. Goodearl and Edward S. Letzter, The Dixmier-Moeglin equivalence in quantum coordinate rings and quantized Weyl algebras , Trans. Amer. Math. Soc. 352 (2000), no. 3, 1381–1403. MR 1615971 (2000 j:16040)
8[8] Gerhard P. Hochschild, Basic theory of algebraic groups and Lie algebras , Graduate Texts in Mathematics, vol. 75, Springer-Verlag, New York-Berlin, 1981. MR 620024 (82i:20002)