1.2. Genration of training data

1.2.1. Creating 1D Kuramoto-Sivashinsky Data

To generate training data run

python3 generate_ks_data.py <PARAMETER_YML_FILENAME> [--path <PATH_TO_PARAMETER_YML>]

The script comes with a command line help, execute python3 generate_ks_data.py -h. Currently the default value for --path is Data/1D_KuramotoSivashinsky/, if the config-YAML is in that directory a path may be omitted. If a path is given, it must be given relative to the git root! The script finds the local path to the git root by utilising GitPython.

We assume the Kuramoto-Sivashinsky equation in the following form

\[\partial_t u(x,t) = - \frac{1}{2} \nabla\left[ u^2(x,t)\right] - \nabla^2 u(x,t) - \nu \nabla^4 u(x,t) \,.\]

The config-YAML configures the following attributes:

Date: "25.07.2027"      # only for convenience, i.e. not used in code

System:
  Name: "KuramotoSivashinsky"
  Parameters:
    # nu: 1
    L: 60
    nx: 128
    t0: 0
    tN: 30000
    dt: 0.25
  Method: "spectral"  # spectral or py-pde

Saving:
  # this must be absolute paths with respect to the repository root
  Name: "KS_t"
  PlotPath: "Figures/"

Jobscript:
  Name: "Example-2023"
  Datadir: "~"  # this has to be an absolute path on the executing system
  Cores: 4

ParamScan:  # this is optional and will generate a parameter sweep
  nu: [0.1, 1.1, 0.2]

1.2.2. Creating 2D Data

The creation of Data (along with an animation of the Data) for implemented models is preformed in ./2D_ExcitableMedia/ by

python3 CreateAnimation.py PATH_TO_PARAMETER_YML

where PATH_TO_PARAMETER_YML is a path to the Parameter .yml files. These are currently stored at ../Data/2D_MODEL_NAME/. The script expects the absolute path with respect to the git root, i.e. in this specific example that would be Data/2D_MODEL_NAME/.

Model parameters or parameters describing the dataset (to be saved) can be changed by modification of the parameter files.

Please report all bugs to @l.fleddermann or fix them by yourself. Bugs will most likely occur if parameters will be changed considerably.

Creation of FitzHugh-Nagumo and Aliev-Panfilov data is implemented and can easily be created. If other models are desired, ./2D_ExcitableMedia/SES.py needs to be modified.

If you need more help you can contact L. Fleddermann.