But it is rather hard to understand if we might use complex or real optical index sometimes... Examples:
1/ Snell-Descartes law in complex gives different results than with real indexes. So, which one is true ?
2/ Link between optical index and dielectric function. Usually, n=sqrt(epsilon). But epsilon is complex. Then index is complex. What about critical angle = arcsin(n2/n1), then ? arcsin(Re(sqrt(epsilon_2)/sqrt(epsilon_1)) ? Or Re(arcsin(sqrt(epsion_2)/sqrt(epsilon_1))) ? Different results.
There many relations like this, in which complex formulation makes everything harder to feel.