Efficient Parameter Estimation in a PDEs Model of the Pulmonary Circulation
Umberto Noe (University of Glasgow)
Friday 28th October, 2016 15:00-16:00 Maths 203
As mathematicians gain more insight and knowledge into the mechanistic models of biological systems, their models will inherently become more complex and perhaps involve systems of nonlinear differential equations. Solving these models will then require high computational time and cost, even for a single simulation from them. In this setting, the problem of estimating model’s parameters intrinsically requires multiple function evaluations, and this means that we can not perform inference in the traditional way as we might end up waiting for months or years when an immediate action/decision is needed.
Recent methods involving emulators of objective functions have been shown to substantially reduce the computational time required to minimize them. In this talk we will show how these emulator-based methods can lead to a faster estimation process.