“…If C(R) = ∅, the story gets more complicated, and we will not discuss the details here. Let us just indicate that, as explained in [19], Section 6, to define the appropriate moduli space and the spectral problem on it, we need to fix a real form G i of G in the inner class s for each component (oval) C i of C(R), and the eigenvalues of Hecke operators are conjecturally parametrized by a certain kind of "real" opers corresponding to this data, i.e., opers with real coefficients satisfying appropriate reality conditions on the monodromy of the corresponding G ∨ -connection. Furthermore, in the tamely ramified case, when we also have a collection of marked points D on C defined over R, to define the most general version of our spectral problem, we need to fix a unitary representation π i of the real group G i for every marked point c ∈ D on C i and a unitary representation of the complex group G C for every pair of complex conjugate marked points c, c ∈ D not belonging to C(R).…”