A coarse grain model and a Monte Carlo sampling formalism are proposed for simulations of self-assembly in block copolymer melts and nanoparticle-copolymer composites. Our approach relies on a particle-based representation of the system, it does not invoke a saddle point approximation, and it permits treatment of large three-dimensional systems. We provide a detailed description of the model and methods and discuss their relationship to results from self-consistent-field theory and single chain in mean field simulations. The validity of the proposed approach is addressed by applying it to study systems whose description within existing approaches would be demanding. In particular, we use it to examine the directed assembly of copolymer blends and nanoparticles on nanopatterned substrates. We show that results from Simulations are in good agreement with experiment, and we use our theoretical findings to help explain the experimental observations.