Hi Gregor,
thank you so much for your prompt and clear response! I see the point, I think the only way to sort it out is to perform a non-self-consistent calc. with fleur on the whole BZ, not only the irreducible one, that should be possible, isn't it? Then I'll write by myself the routine. I know that in principle the IBZ is sufficient as I can re-construct the eigenvalues in the whole BZ with symmetry operation, but I'm afraid to fall down in some silly errors of mine...
The bxsf file is in the reciprocal space, not in the direct one, and is basically a "rolled" 3D matrix per band. What I need is to represent the band structure on the whole BZ like a 4D matrix, three dimensions for the reciprocal space dimensions and one for the eigenvalue index, in a way that the nearest neighbour in the matrix (by numerical index) are also nearest neighbour in the physical reciprocal space, for transport calc. with a BTE code named ElecTra. I attach a file that hopefully illustrates the purpose.