7. Conclusion and future work
We introduce an efficient wave-based acoustic material design optimizer that is capable of handling multiple material segments and multiple target acoustic properties. We show that using the exact derivatives from Automatic Differentiation helps us converge faster on the target optimization result. Additionally, we take advantage of the performance and memory efficiency of the ARD solver compared to other standard acoustic wave solvers. Finally, we show how our system can be used in the application of designing concert halls or other acoustic spaces.In the future we would like to explore methods of applying discrete optimization techniques to the acoustic material optimization problem. While our method can take advantage of the sensitivity of ARD to drive continuous optimization, some of the materials produced may not be physically realistic materials. These continuous values can be discretized into material categories (for example concrete bricks have an absorption between (0.01 and 0.02)). However, discrete optimization approaches could take as input a library of acoustic materials that must be used rather than a continuous curve of absorption values. A further advantage to this approach could be the incorporation of other constraints on the optimization, including material cost or the structural feasibility of using a particular material in a specific location.