dune-copasi is the default PDE solver, which solves the PDE on a triangular mesh using finite element discretization methods. The mesh is automatically constructed from the geometry image, as described in mesh
.
The default settings should work well in most cases, but if desired they can be adjusted by going to Advanced->Simulation options
The simulation options that can be used to fine tune-the dune-copasi solver.- Discretization
- currently only 1st order FEM is supported
- other discretizations (such as 2nd order FEM) may be added in the future
- Integrator
- the Runge-Kutta integration scheme used for time integration
- a variety of implicit and explicit schemes of different orders are available
- the default is the 2nd order Alexander scheme, which is a Diagonally Implicit Runge Kutta method
- Initial timestep
- the timestep used at the start of a simulation
- Min timestep
- the minimum allowed timestep
- for a very stiff model it may be necessary to reduce this value
- Max timestep
- the maximum allowed timestep
- reducing this value may increase the accuracy of the solution but simulations will take longer to run
- Increase factor
- after a successful integration step, the timestep is multiplied by this factor
- this must be greater than or equal to 1
- if equal to 1, the timestep is never increased between integration steps
- the larger the value, the more the timestep is increased after successful integration steps
- Decrease factor
- if an integration step is unsuccessful, the timestep is multiplied by this factor and the step is repeated
- this must be less than 1
- the smaller the value, the more the timestep is decreased in the case of an unsuccessful integration step
- Output files
- VTK files of the species concentrations throughout the simulation can be generated
- these files can be viewed using ParaView
- Newton relative error
- the relative error where Newton iteration is considered to have converged
- currently this may need to be altered depending on the units and geometry size (see #315)
- Newton absolute error
- the absolute error where Newton iteration is considered to have converged
- currently this may need to be altered depending on the units and geometry size (see #315)
For more information see the dune-copasi documentation.