Revisiting the moduli space of semistable \(G\)-bundles over elliptic curves (Q2073608)

Let \(E\) be an elliptic curve over an algebraically closed field \(k\) (of arbitrary characteristic) and \(G\) be a reductive group over \(k\). Denote by \(\mathcal{M}_G^d\) the moduli space of semi-stable \(G\)-bundles on \(E\) of topological type \(d \in \pi_1(G)\). In the paper under review, it is proved that there is an isomorphism \[ \mathcal{M}_G^d \simeq \mathcal{M}_{C_d}^{d'}/W_d \] where \(C_d\) is certain algebraic torus, \(d' \in \pi_1(C_d)\) and \(W_d\) is a certain Weyl group. Note that the RHS is isomorphic to certain power of the Jacobian of \(E\). This simultaneously generalizes the previous results \begin{itemize} \item for vector bundles, where \(\mathcal{M}_{\mathrm{GL}_r}^d \simeq E^m/\mathfrak{S}^m\) for \(m=\gcd(r,d)\). \item for \(G\)-bundles in the case \(k=\mathbb{C}\), where \(\mathcal{M}_G^0 \simeq \mathcal{M}_T^0/W\). See [\textit{Y. Laszlo}, Ann. Inst. Fourier 48, No. 2, 413--424 (1998; Zbl 0901.14019)]. \item etc.\ldots \end{itemize} The isomorphism is obtained via generalizing Laszlo's approach. Starting from Laszlo's result above, firstly one needs to find a replacement of the torus \(T\) and this is done in another paper of the author [\textit{D. Fratila}, Math. Ann. 365, No. 1--2, 401--421 (2016; Zbl 1345.14004)], which says that every polystable \(G\)-bundle is induced from a stable \(L_d\)-bundle of degree \(d'\) for some Levi subgroup \(L_d \subset G\) and \(d' \in \pi_1(L_d)\). This also gives the (relative) Weyl group \(W_d:=N_G(L_d)/L_d\) and induces a finite, \(W_d\)-invariant morphism \(\mathrm{ind}: \mathcal{M}_{L_d}^{d'} \to \mathcal{M}_G^d\). To show this is a quotient some extra work are required to deal with the subtleness appearing in positive characteristic. Relating \(\mathcal{M}_{L_d}^{d'}\) to line bundles is somehow natural by Atiyah's theorem. One takes the determinant map \(\det: L_d \to L_d/[L_d,L_d]=:C_d\) and this induces an isomorphism \(\mathcal{M}_{L_d}^{d'} \simeq \mathcal{M}_{C_d}^{\det(d')}\). This paper is well-written, the outline and the proofs are very clear. I enjoyed reading it.