UC Merced

Soft corals in the Xeniidae family pulse in an energetically expensive manner to generate fluid flow, which benefits them biologically. This flow has been modeled using computationally intensive numerical methods, such as the Immersed Boundary Method, which limits the ability to explore a wide parameter space. To address this challenge, we introduce Gaussian Process Regression as a surrogate model to capture the kinematic and flow outcomes of these simulations. By training on a limited set of simulations, the surrogates enable accurate predictions across a broader range of parameters than those originally sampled. These surrogate models can then be optimized to identify the parameter combinations which maximize fluid flow. The resulting optimal parameters can further guide additional numerical simulations, creating an iterative framework for targeted exploration of the parameter space. This approach offers new insights into how the coral polyp musculature and local flow conditions influence the resulting kinematics and fluid transport while substantially reducing computational cost.