This is the situation: you have the band structure of a two-dimensional semiconductor E=E(k). Both, valence band and condcution band. You use the definition of effective mass: (m)^(-1)=(d2E/dkidkj), but both bands are in such a way that the 2x2 matrix that you obtain has zero determinant. So that matrix has no inverse and therefore you can not get the effective mass. Is this possible? Should there be an error in the expressions of the bands? What does it mean?