Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

The underlying variety of a linear elgebraic group (say, over an algebraically closed field) is affine, so doesn't have nontrivial (infinitesimal) deformations. I'm curious to know whether it's possible to deform the group structure on a fixed variety (that admits at least a structure of an algebraic group).

Edit: I should have added, though when I wrote the question I mistakenly didn't expect reductivity made any difference, that I was mostly interested in reductive groups (in which case the comment of user54268 shows my question was not that naive after all..).

share|improve this question
The group law on a unipotent algebraic group deforms non-trivially in general. For instance by scaling the symplectic form you get a one parameter deformation of the three dimennsional Heisenberg group (upper triangular $3\times 3$ matrices with ones on the diagonal) to the additive group of a $3$ dimensional vector space. –  Tony Pantev Aug 28 '14 at 1:48
@Tony Pantev Huh. I guess I just wrote down this. For some reason your comment didn't show up while I was writing my answer... –  Stephen Griffeth Aug 28 '14 at 1:57
If $G$ and $H$ are smooth affine group schemes with connected reductive fibers over an artin local ring $R$ (with arbitrary residue field) and their geometric special fibers are isomorphic then the functor ${\rm{Isom}}(G,H)$ is represented by a disjoint union of smooth affine $R$-schemes, as can be checked over a finite etale cover that splits $G$ and $H$ (and then using etale descent, which is avoids some effectivity problems since Spec($R$) is a fat point). Thus, an isomorphism of special fibers (if one exists) always lifts to an $R$-isomorphism. –  user54268 Aug 28 '14 at 2:00
@user54268 So, you are saying that reductive groups are rigid, yes? –  Stephen Griffeth Aug 28 '14 at 2:04
@StephenGriffeth: Yes, though I'd prefer to say "infinitesimally rigid" (and be explicit about the connectivity assumption; hmm, perhaps fibral connectivity can be dropped, but I haven't thought it through). The proof rests on quite a bit of the theory of reductive group schemes in SGA3. –  user54268 Aug 28 '14 at 3:43

2 Answers 2

A most excellent example of "non-rigidity" beyond the reductive case (depending on how loose one wants to be about the meaning of "rigidity") is given in 5.2--5.10 of Exp. XIX of SGA3: a smooth affine group scheme $G$ over $k[t]$ for any field $k$ of characteristic 0 such that $G|_{t \ne 0}$ is a form of ${\rm{PGL}}_2$ (reductive!) but the fiber $G_0$ is solvable with 2 geometric connected components. By definition, $G$ is the automorphism scheme of the Lie algebra over $k[t]$ whose underlying $k[t]$-module is free on a basis $\{X,Y,H\}$ that satisfy $$[H,X]=X,\,\,\,[H,Y]=-Y,\,\,\,[X,Y]=2tH.$$ (It is clear that $G|_{t\ne 0}$ becomes ${\rm{Aut}}_{\mathfrak{sl}_2} = {\rm{PGL}}_2$ over the degree-2 finite etale cover defined by $\sqrt{t}$, but proving that $G$ is $k[t]$-smooth requires some cleverness.)

share|improve this answer
Nice! It's never a bad idea to read SGA. Do you know if it is possible to see this deformation as a commutative deformation living within a larger family also containing quantum $\mathfrak{sl}_2$? –  Stephen Griffeth Aug 28 '14 at 14:43
@Stephen Griffeth: I know less than nothing about quantum-math, so I have nothing meaningful to say about your question; sorry. –  user27920 Aug 29 '14 at 16:25
I don't really know the general definition of "reductive", but is this $G$ reductive "over $k[t]$"? There are many examples of reductive groups degenerating to solvable groups, of which this is a particularly nice one. –  Theo Johnson-Freyd Sep 2 '14 at 22:45

How about this: for each $t \in F$, define a group structure on $F^3$ by $$(a,b,c) \cdot (d,e,f)=(a+d,b+e,c+f+tae).$$ When $t=0$ this is just $F^3$ with coordinate-wise addition, while for $t \neq 0$ it is $3$ by $3$ unipotent matrices.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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