-
Notifications
You must be signed in to change notification settings - Fork 29
Thetis meeting agenda 2021 05 14
Tuomas Kärnä edited this page Jun 8, 2021
·
5 revisions
Date/time of meeting 2021-05-14 11:00 BST
Present: Stephan, David, Simon, Lucas, Mariana, Lawrence, Anastasia, Tuomas (minutes)
Adds a new term in the 2D sediment model, Tuomas will take a look.
We discussed the thetis_how_to repository and it's structure. Some comments
- Some concerns that the repository duplicates some aspects of Thetis interface. E.g. it would be preferable to use the same options in the how_to_repository.
- For maintainability it would be preferable not to duplicate code in the input/tools files.
- If there's need, options could be read in from a python/yaml/json file. But that might clash with Thetis'
options.pywhich already defines the default values. The doc pages have a summary page of all the options and their default values. We could make this page nicer for the users for example. - The main script could still be a python script the defining
solver_objectand setting it's options? There could be helper functions to create, say, a viscosity sponge field which could then be assigned tosolver_object.options.horizontal_viscosity. Same for all other frequently used settings. - More generic file I/O could be implemented in Thetis to facilitate the use cases needed in the examples. Would be useful to have examples of desired functionality, e.g. as new Thetis issues.
Discussion to be continued online.
The netCDF4 lib built by PETSc does not have parallel mode enabled; can be fixed by passing appropriate config args to PETSc.
The netcdf4-python package however needs to be built manually with the right flags to enable parallel mode. Plausibly it's possible to pass in the right agrs with pip install. Tuomas will look into it.