Random Walk Detectors

Fluidity's current detector advection algorithm only uses the velocity field to calculate the lagrangian detector path, but does not consider the displacement due turbulent motion, as parameterised by GLS. Many Random Walk schemes exist that are used to compute how individual particles move in response to turbulent diffusion. These usually depend on arbitrary fields, such as diffusitivity or turbulent kinetic energy (TKE), and this dependence should not be hard-coded into femtools (where the lagrangian detector advection lives). The proposed solution is to extend the Python interface to allow the user to define its own Random Walk scheme, which can optionally be added to the local velocity when advecting lagrangian detectors. The new Python field will provide the 'state' object, as well as a function to evaluate fields at the detector position, and will require the user to define a 'val' function that gets evaluated at each sub-timestep during lagrangian detector advection.