How to set initial (continuous) state before starting the simulation?

[FabiJa]

What is the preferred way to set the initial (continuous) state of an ODE before simulation?

model = pyfmi.load_fmu(fmu_path)
model.continuous_states= numpy.array((10.))
model.simulate()

is not working. Also I had no luck with set_real(). Or is it not possible to change the initial condition, when using ´simulate()´ and I have to do it low level with ´model.setup_experiment()´ etc. ?

Thx.

[PeterMeisrimelModelon]

The issue in your approach is that simulation includes initialization, which will overwrite your specified continuous states. You'll need to initialize the model, overwrite the continuous states and then simulate.

The following approach should work:

fmu = load_fmu(<your_fmu>)

fmu.initialize()
fmu.continuous_states = np.array([<your_states>])

opts = fmu.simulate_options()
opts['initialize'] = False

res = fmu.simulate(options=opts)

[PeterMeisrimelModelon]

Did this help/resolve the issue? @FabiJa

[FabiJa]

Sry for the late reply.

Is fmu.initialize() really necessary? It seems to work without and gives me an error for a CoSimulation FMU.

I am using this variant:

for var in initial_state:
   fmu.set(var, initial_state[var])

[PeterMeisrimelModelon]

Initialization of the FMU is a required step before simulation can be done.

In PyFMI this is done either via either

• an explicit fmu.initialize() call
• fmu.simulate(options=opts) with opts['initialize'] = True (default)

Having both will result in an error. Having neither should result in an error from the FMU.

The recommended approach for CS FMUs is:

fmu = load_fmu('<name>.fmu')

fmu.initialize()

fmu.set('<state>', <value>)

opts = fmu.simulate_options()
opts['initialize'] = False

res = fmu.simulate(options=opts)

It is possible that initialization may not overwrite set values, in which case it could be done before initialization. However, this depends on the FMU.

[PeterMeisrimelModelon]

Marking as resolved, please notify if this is still an open issue.