@ Shlomo, thanks for your answer. But, no it does not answer my question. Bruggeman's effective medium approach works for physical mixtures and is not ab-initio.

Thanks Craig. I am thinking of a very fundamental approach where you only know the molecular structures of the components and do not know (or do not have to) define the final structure of the composite (or arrangement of pseudo-potentials for that matter). Packages such as VASP (materials design, studio etc etc) all employ very similar methodologies in computing optical properties (or as you have mentioned electromagnetic properties) and cannot (or do not) predict based on fundamental components. Particularly polymer composites are generally very large systems and formation of them is a very complicated process. One of the approaches that has been used successfully is defining bond potential of various atoms and use reverse annealing in systems like LAMMPS (+NAMD) that can handle large MD simulations and form the much larger composites. In reverse annealing you start at high temperature where everything is gas and then bring down the temperature, bonds are formed according to the bond potential. This is also possible in VASP and this approach works well for systems like say networked silica or condensation reactions leading to silica formation. But if you want to condense 3 different silicon derivatives and form a derivative of silica, such a method never converges to a particular solutions because of the proximity of various bond potentials. But optical properties, such as refractive indices can be very different. I am sure someone here will correct me if I am wrong.

For the composite materials, the refractive indices may be some mixing of each components. Thus first of all, you need to determine the refractive property of the simple unit first. First-principles calculations of the whole hybrid system seems impossible up to now.

You may refer to the following paper for calculating the refractive properties directly from first-principles methods.

Infrared reflectance spectrum of BN calculated from first principles, Solid State Communications 141 (2007) 262–266

Thanks Yongqing.