Mimetic discretizations: new opportunities for pre- and post-processing

Date & time

1.30–2.30pm 11 December 2017

Location

John Dedman G35

Speakers

Alexander Pletzer, New Zealand eScience Infrastructure (NeSI) and National Institute for Water and Atmospheric research (NIWA)

Contacts

 Stephen Roberts

The need to preserve physical properties such as div curl = 0 has led to the development of "mimetic" discretization methods (e.g. DEC and FEEC), which are gaining popularity in the electromagnetics and computational fluid dynamics communities. The GungHo atmospheric dynamical core project developed by the UK Met Office is an example of a mimetic model, which employs mixed finite elements with non-collocated velocity field components to achieve energy and mass conservation. I will discuss the impact of mimetic methods on visualization, pre- and post-processing tools with particular emphasis on developing suitable interpolation and regridding techniques. The proposed interpolation methods for discretized scalar, vector, pseudo-vector and pseudo-scalar fields conserve line, surface flux and volumetric integrals depending on the field type, work for arbitrarily distorted hexahedral cells and handle grid singularities and masked regions out of the box.

Updated:  17 December 2017/Responsible Officer:  Director/Page Contact:  School Manager