An inverse design algorithm has been developed that predicts the necessary topographical template needed to direct the self-assembly of a diblock copolymer to produce a given complex target structure. The approach is optimized by varying the number of topographical posts, post size, and block copolymer volume fraction to yield a template solution that generates the target structure in a reproducible manner. The inverse algorithm is implemented computationally to predict post arrangements that will template two different target structures and the predicted templates are tested experimentally with a polydimethylsiloxane-b-polystyrene block copolymer. Simulated and experimental results show overall very good agreement despite the complexity of the patterns. The templates determined from the model can be used in developing simpler design rules for block copolymer directed self-assembly.