The coupling of design and numerical analysis in computational design assumes a one-directional forward schema from input to output. Forward systems provide limited interpretability of how the design system maps to the engineering response. Subsequently, the difficulty to keep track of cause and effect relationships results in an opaque cyclic feedback loop between design-generation and evaluation. Therefore, how can we afford better control over physical behaviour when synthesizing architectural design with engineering numerical analysis?
In response, Zack’s research borrows methods from statistics and artificial intelligence to facilitate and promote knowledge-inference, in the human-computer relationship. His PhD thesis introduces ‘Bayesian inference’, borrowed from the field of probability, as a concept to facilitate the interpretability of engineering responses from numerical analysis, with focus on finite element analysis. The novelty of the approach presented in the research lies with the notion of bi-directional reasoning; in other words, the inverse facility to infer how the inputs generate an output response of interest. The research found that the capacity to reason about inverse scenarios within a probabilistic representation enables to narrow down a vague understanding of a design space into the meaningful regions of interest. In conclusion, the thesis demonstrated how the concept of Bayesian inference in a design space setting can serve as a powerful translational mechanism between the architectural design and engineering domains.