On a relative version of a theorem of Bogomolov over perfect fields and its applications (Q615821)

Consider a linear algebraic group \(G\) over an algebraically closed field. Of primary interest here are the notions of observable and quasiparabolic subgroups (and a related notion of subparabolic). A subgroup \(H\) of \(G\) is said to be observable if the homogeneous space \(G/H\) is a quasi-affine variety. A subgroup \(Q\) of \(G^0\) (the connected component at the identity of \(G\)) is said to be quasiparabolic if \(Q = G^0_v\) (the stabilizer in \(G^0\) of \(v\)) for a highest weight vector \(v\) in an absolutely irreducible \(G\)-module. Consider a finite dimensional \(G\)-module \(V\) and an element \(0 \neq v \in V\) which is unstable, that is, for which 0 lies in the closure of \(G\cdot v\). The title of the article references a theorem of Bogomolov which says that, for a connected reductive \(G\), the stabilizer \(G_v\) of \(v\) is contained in a quasiparabolic subgroup of \(G\), thus providing a method for showing the existence of quasiparabolic subgroups. The first main goal of this paper is to generalize this result to the case that \(G\) is defined over a not necessarily algebraically closed but perfect field \(k\). To do so, one must consider the notion of a subgroup being quasiparabolic over \(k\) (quasiparabolic and defined over \(k\)) and a stronger condition of being \(k\)-quasiparabolic. The authors give two proofs, one following arguments of Grosshans and the other arguments of Borel and Tits, of a generalization of Bogomolov's result to \(k\)-quasiparabolicity over a perfect field \(k\). This result is then used, in part, to give a characterization of the relationships between the notions of \(k\)-quasiparabolic, quasiparabolic over \(k\), observable over \(k\), \(k\)-subparabolic, and (strongly) subparabolic over \(k\) for a subgroup of a linear algebraic \(k\)-group with \(k\) perfect. From the definitions, quasiparabolic over \(k\) implies obesrvable over \(k\) but the converse is generally false. The main result is in part a generalization of a result of Sukhanov, that observable over \(k\) is equivalent to being \(k\)-subparabolic (or equivalently subparabolic over \(k\)). It is also shown that for a semisimple \(G\), quasiparabolic over \(k\) implies \(k\)-quasiparabolic.