A code for generation of Python objects and input for various MOOSE.
- Python version >=3.6
- NumPy
pytest(for testing)
From a terminal, run
pip install .To include packages for testsing, run
pip install .[test]First, import the package and load available syntax from a MOOSE application.
>>> from catbird import *
>>> factory=Factory('./heat_conduction-opt')This may take a few minutes, so don't panic if it hangs. The syntax first has to be dumped from the original application as json and then parsed; the factory then converts "available" syntax into python object constructors. The more syntax we enable the longer it takes, so if you want to speed things up you should only enable the syntax you intend to use. To write out what objects are currently enabled, try:
>>> factory.write_config("config.json")We could then edit that file and load it via an optional constructor argument to Factory:
>>> lightweight_factory=factory=Factory('./heat_conduction-opt', config="lighweight_config.json")However, modifying the config file directory is likely to be very laborious if done manually. Another option
is to derive your own Factory class and override the set_defaults method. To limit the enabled syntax, we can pass in dictionaries
of enabled syntax into the enable_syntax method, e.g.:
executioner_enable_dict={
"obj_type": ["Steady","Transient"]
}
self.enable_syntax("Mesh")
self.enable_syntax("Executioner", executioner_enable_dict)
With our factory built, all we have in practice is bunch of constructor objects. We now need to create the objects and assemble them as a Model:
>>> model=MooseModel(factory)This will have created a "boiler-plate" model with some sensible base objects to work with. We can now start to set their attributes in an object-oriented way.
>>> model.mesh.dim=2
To see the available attributes for a given object, just use "help" to obtain useful documentation e.g.:
>>> help(model.mesh)
The help will also print the type and valid options. Type checking of attributes is performed, and if the type (i) incompatible with the expected type and (ii) not castable as the expected type then a ValueError exception is raised, e.g.
>>> model.mesh.dim='2' # This is fine, MOOSE expects string
>>> model.mesh.dim=2 # This is fine, str(2) is valid
>>> model.mesh.nx=1 # This is fine, MOOSE expects in
>>> model.mesh.nx="hello" # This raises a ValueError as int("hello") is invalid
Many of the objects are Collection types, i.e. a collection MOOSE objects, for example Variables. To add a variable, we must call:
>>> model.add_variable("T", order="SECOND")
Notice the use of key-word arguments in this function call. We can also act on the created object directly, via
>>> model.add_variable("T")
>>> model.variables.objects["T"].order="SECOND"
Even if the root-level syntax hasn't been added to the model (but is available in the factory) we can still add it to our model through add_syntax calls:
model.add_syntax("VectorPostprocessors")
Since this is a Collection type, to add a specific VectorPostprocessor, we call the add_to_collection method:
model.add_to_collection("VectorPostprocessors","VectorPostprocessor","t_sampler")
It is generally expected that the user will develop their own class derived from MooseModel, overriding the method load_default_syntax.
See for example TransientModel in model.py.
Finally if we are happy with our model we can write to file:
>>> model.write("catbird_input.i")You can run that input from the command line as you normally would or launch a subprocess.
Note, if you inspect our file catbird_input.i you may not see all attributes written out. The default behaviour is that parameters that are unchanged relative to
their default value are suppressed when writing to file. To see all attributes, instead run
>>> model.write("catbird_input_full.i", print_default=True)Fully worked examples may be found in catbird/examples. They should be run from this directory.
- Minimal heat conduction example. Run with:
$ python thermal_example.py- Divertor monoblock thermal example. Run with:
$ python monoblock_thermal_example.py- Simple breeder pin tritium diffusion example. Run with:
$ python simple_breeder_example.py