Abstract
Since the invention of sound diffusers four decades ago, several alternatives have been suggested. Sonic crystals have been demonstrated to be good candidates to substitute conventional diffusers in order to overcome the need for extremely deep structures when low
frequencies have to be scattered. In this work, we explore the possibility of optimizing such structures providing better performance over a large frequency range. For doing so, evolutionary algorithms have been used in combination with Finite-Difference Time-Domain (FDTD) algorithms that allow predicting the performance of a sound diffuser. Preliminary results
show that diffusion can be significantly increased.
