07-11, 10:20–10:30 (Europe/Amsterdam), Function (4.1)
VoronoiFVM.jl solves nonlinear coupled multiphysics diffusion-reaction-convection partial differential equation systems in one, two an three space dimensions using the the Voronoi finite volume method, leveraging ForwardDifff.jl for the construction of the Jacobi matrices used by the nonlinear solver.
The talk will explain the basic ideas of the package API and give an overview on the developments since the first presentation at JuliaCon 2020, and highlight the integrative potential of Julia.
After the first presentation at JuliaCon 2020, the package has undergone a number of improvements. These include
- Integration with DifferentialEquations.jl ODE solvers
- Integration with LinearSolve.jl providing easy access to different linear solvers
- New preconditioners for linear systems inlcuding AMGCLWrap.jl
- Differentiability through Sparspak.jl
- Mesh generation capabilities in accompanying packages.
- Visualization capabilities in accompanying packages based on Makie.jl and PlutoVista.jl, among others
- Multithreading (work in progress)
The package is used as the core solver in several research projects in fields like electrochemistry and semiconductor modeling.
As the contribution is planned as a lightning talk, presentation time will be short. I however intend to present with the help of a Pluto notebook, which (or an extended version thereof) will be made available online at the time of the talk, giving interested participants the possibility to try out computational examples showcasing the package's capabilities.
Deputy head of numerical mathematics and scientific computing at Weierstrass Instute for Applied Analysis and Stochastics Berlin. Teaching at TU Berlin. 35 years of experience in PDE numerics and visualization. Moved from C++/python to Julia since v1.0 in 2018.