Simulation

To simulate a model (or models), we will need to first perform a few steps. In the most simple forms, these are the steps:

1. Create the model objects
2. Create a simulation object
3. Simulate the simulation object.

Assume that you have two models named m1 and m2, that have been installed already.

Creating the model objects

The first step is to construct the model objects:

m1 = sund.load_model('m1') # Creates a model1 object, based on the model1 class. 
m2 = sund.load_model('m2') # Creates a model2 object, based on the model2 class. 

Creating a simulation object

Next, we use the model objects to construct a simulation object from the Simulation class, where we now use our two model objects:

sim = sund.Simulation(models = [m1, m2]) # Creates a simulation object, using our two model objects to define which models to simulate. 

Simulating the model

Finally, we simulate our simulation object.

sim.time_vector = times
sim.simulate() # Simulates the constructed simulation object

or by passing the time_vector argument (or the shorthand t):

sim.simulate(time_vector = times) # Simulates the constructed simulation object, for some time vector 'times'. 

After the simulation have been done, the values of the defined features is updated in the feature_data attribute, the values of the states (state_values) are updated to the values of the states in the final time point of the simulation. This means that you can extract these values as initial conditions to a future simulation, or that could potentially continue the simulation from where it stopped.

The simulation object does not reset between consecutive simulations

The simulation object does not reset between consecutive simulations if not explicitly asked for. Pass the keyword argument reset = True, or run sim.reset_states() between the simulations to actively reset the simulation before simulating. If you pass the state_values keyword argument, the reset is done automatically.

Note that the simulation is slightly faster when the time_vector is set before the simulation. There are some pre-calculations that are being done when the time is set, so by setting this as few times as possible, it is possible to speed up the simulation slightly. E.g. useful when doing parameter estimation. If you need to simulate multiple different times, we recommend creating multiple simulation objects, perhaps storing them in a dictionary:

sims = {
    "t1": sund.Simulation(models = [m1, m2], time_vector = t1)
    "t2": sund.Simulation(models = [m1, m2], time_vector = t2)
}

Using the results of the simulation

We can access the simulated values of the model features using the feature_data attribute of the simulation object.

print(sim.feature_data)

Let's assume that we have defined three features (F1, F2, and F3), and you want to get the values for F2, you can do either of the following:

# Use the position directly
values = sim.feature_data[:,1] # Remember, Python uses 0-indexing

# Use the name of the feature
idx = sim.feature_names.index("F2")
values = sim.feature_data[:,idx]

If you want to extract the final values of the simulation, e.g. to use as the initial conditions to another simulation, you can get the values from:

initial_conditions = sim.state_values

To get the feature_data as dictionary format, you can use the feature_data_as_dict() function. The keys of the dict will when be the feature_names

results_dict=sim.feature_data_as_dict() 
print(results_dict)

Summary

This has been the most basic steps to simulate one or more models. In practice, you would typically also want to include parameter_values or state_values to set the parameter/state values without having to change the default values set in the model file (m1 and m2 in this example).