Recent Releases of AutoPDEx
AutoPDEx - v1.1.1
Adds the dae module. Enables solving coupled ODEs and DAEs with different time integrators and time step controlers. Can also be used to solve transient PDEs and PDAEs in the 'call pde' mode. See the notebook examples for more details.
Scientific Software - Peer-reviewed
- Python
Published by BodeTobias 11 months ago
AutoPDEx - v1.1.0
Breaking change:
In order to circumvent constant folding during just in time compilation, 'connectivity' and 'dirichlet dofs' has been moved from the static_settings to the settings. Further their entries do not have to be converted to FrozenDicts of tuples of tuples. Now, they have to have the same structure as the DOFs, i.e. either be a jnp.ndarray or a dict of jnp.ndarrays with the same keys as the DOFs.
Example:
Previous code:
python
static_settings = flax.core.FrozenDict({
...
"connectivity": (utility.jnp_to_tuple(connectivity)),
"dirichlet dofs": utility.jnp_to_tuple(dirichlet_dofs),
})
settings = {
...
}
New code:
python
static_settings = flax.core.FrozenDict({
...
})
settings = {
...
"connectivity": (connectivity),
"dirichlet dofs": dirichlet_dofs,
}
Scientific Software - Peer-reviewed
- Python
Published by BodeTobias about 1 year ago
AutoPDEx - v1.0.4
The degrees of freedom (dofs) can now be dictionaries of jnp.ndarrays with user-defined keywords. This enables the formulation of mixed problems with different solution spaces for each field and is first of all supported for the assembling modes 'user potential', 'user residual' and 'user element'. See the new notebook example mixed_problem.ipynb in the docs.
Scientific Software - Peer-reviewed
- Python
Published by BodeTobias over 1 year ago
AutoPDEx - First public version
Scientific Software - Peer-reviewed
- Python
Published by BodeTobias over 1 year ago