Inference in nonlinear ordinary differential equations (ODEs) is challenging due to the high computational complexity of the numerical integration. In the present paper, we explore an emulation-based approach for approximating the likelihood, based on Gaussian processes, with the objective to reduce the number of numerical integration steps. We assess the viability of the scheme on the Lotka-Volterra model of predator-prey interactions.
|Title of host publication||Proceedings of the 30th International Workshop on Statistical Modelling (IWSM)|
|Place of Publication||Linz, Austria|
|Publication status||Published - 1 Jul 2015|