No label defined (Q4338613)

The paper consists of two parts. Let \(G\) be a linear algebraic group over an algebraically closed field \(K,\Gamma\) a finite group and let \(G\) act morphically on the \(K\)-variety \(X\). Then the morphism \(\eta: G\to X\), \(g\mapsto g\cdot x\), \(x\in X\), is called the orbit map attached to \(X\). This map is called separable if its differential at the neutral element \({\mathcal D}_e \eta:T_e G\to T_XX\) maps surjectively onto the tangent space \(T_XC\) of the orbit \(C= \{g\cdot x|g\in G\}\) of \(x\). The main results of the first part are the following. Theorem 1. Let \(H\) be a closed subgroup of \(G\) such that \((G,H)\) is a reductive pair, \(C \subset G^n\) the simultaneous \(G\)-conjugacy class of an element \(x\in H^n\). Assume that the associated orbit map \(\eta: G\to C\) is separable. Then \(C\cap H^n\) decomposes into finitely many simultaneous \(H\)-conjugacy classes. Each corresponding orbit map is again separable. Theorem 2. Assume that \(|\Gamma |\) is prime to \(\text{char} K\). Then there are only finitely many equivalence classes of homomorphisms of \(\Gamma\) into \(G\). \(\text{char} K\) is said to be ``good'' for \(G\) if (i) \(G^0\) contains a factor not of type \(A_n\), \(\text{char} K \neq 2\); (ii) \(G^0\) contains a factor of type \(E_6, E_7, E_8, F_4, G_2\), or a factor of type \({\mathcal D}_4\) on which \(G\) induces a triality, i.e. an outer automorphism of order 3, \(\text{char} K \neq 3\); (iii) \(G^0\) contains a factor of type \(E_8\), \(\text{char} K \neq 5\). Theorem 3. Let \(G\) be a reductive group over an algebraically closed field \(K\) of characteristic \(p>0\) such that \(p\) is good for \(G\). Let \(\Gamma\) be a finite group with Sylow \(p\)-subgroup \(\Gamma_p\) and fix a \(G\)-equivalence class of representations \(\Gamma_p \to G\). Then there are only finitely many \(G\)-equivalence classes of representations \(\rho: \Gamma\to G\) such that the restriction \(\rho |_{\Gamma_p}\) belongs to that fixed class. In the second part the author considers the following problem of B. Külshammer. (1) Let \(\text{char} K\) be prime to \(|\Gamma|\). Then are there only finitely many representations \(\rho: \Gamma\to G\) up to conjugation by \(G\)? (2) Let \(p= \text{char} K\) and \(\Gamma_p\) be a Sylow \(p\)-subgroup of \(\Gamma\). Fix a conjugacy class of representations \(\overline\rho: \Gamma_p \to G\). Are there, up to conjugation by \(G\), only finitely many representations \(p:\Gamma \to G\) whose restriction to \(\Gamma_p\) belongs to the given class? The answer to (1) is positive (A. Weil, 1964). Theorem 1 of the second part gives another positive solution of problem (1). For (2) there is Corollary 2 (to theorem 2). Assume \(G=GL(V)\) or that \(G\) is reductive and \(\text{char} K\) is good for \(G\). Then question (2) has a positive answer. Appendix (by G. Martin-Gram): The author shows that the answer to question (2) is negative. He constructs a counterexample for a prime field \(\mathbb{F}_2\). The finite group is \(\Gamma= SL_2(2)\), the algebraic group \(G\) has dimension 3 over \(\mathbb{F}_2\); it is a semidirect product of \(\Gamma\) with a Heisenberg-type algebraic group \(Q\) defined over \(\mathbb{F}_2\). For the entire collection see [Zbl 0857.00012].