Pass model to solve as an alternative to sim_func

[valentinsulzer]

Description

Add a model keyword argument to lp.solve where a user can pass in a pybamm model (instead of via sim_func). This is slightly easier than having to write an entire sim_func function, if all you want to change is the model

Motivation

Pro: easier to change the model and leave everything else the same
Con: more ways of doing the same thing

Possible Implementation

lp.solve(
    netlist=netlist,
    parameter_values=param,
    experiment=experiment,
    initial_soc=initial_soc,
    model=model,
)

Additional context

No response

[TomTranter]

The reason I have set it up as a sim_func is that the function gets run on every sub-process which is less expensive and memory intensive than passing objects to each sub-process. I agree it's not as easy as passing a model and probably for small problems would work fine but for large scale problems I was getting a lot of issues making pybamm objects and passing them around.

[valentinsulzer]

I was thinking that if the user passes a model instead of a sim_func it would automatically create a basic sim_func using that model. Something like

def get_basic_simulation_function(model=None):
    # Create the pybamm model
    if model is None:
        model = pybamm.lithium_ion.SPM()
        
    def basic_simulation(parameter_values=None):
        """
        Create a Basic PyBaMM simulation set up for integration with liionpack

        Args:
            parameter_values (pybamm.ParameterValues):
                The default is None.

        Returns:
            pybamm.Simulation:
                A simulation that can be solved individually or passed into the
                liionpack solve method

        """
        # Add events to the model
        model = lp.add_events_to_model(model)

        # Set up parameter values
        if parameter_values is None:
            param = pybamm.pybamm.ParameterValues("Chen2020")
        else:
            param = parameter_values.copy()

        # Set up solver and simulation
        solver = pybamm.CasadiSolver(mode="safe")
        sim = pybamm.Simulation(
            model=model,
            parameter_values=param,
            solver=solver,
        )
        return sim

    return basic_simulation

[TomTranter]

@tinosulzer we can set it up and see what the speed is like for a large problem, maybe we could actually get away with passing an unbuilt simulation