New materials with fast non-linear optical (NLO) response over broad spectral bandwidths, macromolecular structures for optical switching and advanced absorbing dyes, which are critical for laser eye and sensor protection, are explored. The design of such novel materials requires that material properties are carefully optimized, for example, to achieve large NLO coefficients, high conjugation, appropriate ground and excited state absorption, as well as exhibit defined structural motifs, folding patterns and properties, especially when derivatized with NLO chromophores. A comparative study of porphyrins with meso-acetylene substituents carried out with the local density functional approximation, the gradient-corrected DFT, and HF calculations, is discussed in detail. The calculated electronic structures clearly show that the acetylene group contributes to the it-electron conjugation along the porphyrin ring, being consistent with experimental results. Recent calculations on variety of porphyrins are also reported. For reliable properties predictions, a fundamental understanding of solvent effects is important, and large molecules have to be modeled from first principles. We report calculations of interactions with the effective-fragment potential (EFP) method, where solvent molecules are placed around a solute to generate correctly the first solvation cell within an ab initio framework, using the parallelized GAMESS code, providing insight and good agreement with experiment. For example, we correctly predict the change in the most stable isomer for glutamic acid from the neutral form (gas-phase) to the zwitterionic form (in solution), by using the EFP method with a larger number of solvent fragments. A recent implementation of simulated annealing within EFP to optimize the positions of the fragment molecular systems is discussed. The development of global optimization techniques for determining energy minima of complex macromolecules is described. In addition, simulations to evaluate the properties of a polymer dispersed liquid crystal system of interest are discussed. The utilization of this approach to the study of liquid crystalline materials is described. These challenge applications in computational chemistry and materials science enable us to derive crucial structure-to-property relationships for candidate materials and enhance the capability for 'real materials' design and atomic-scale control, all of which utilize extensively scalable high-performance computing.