Support simulations via FMI

[kdavies4]

Use PyFMI; see http://www.os.is/gogn/Skyrslur/OS-2012/OS-2012-06/OS-2012-06-03.pdf and http://nbviewer.ipython.org/gist/saroele/e6f176ac0ff391e8982a

[kdavies4]

Will be supported via a context manager: modelicares.exps.simulators.fmi

[arnoutaertgeerts]

Most of the implementation is done and works on Windows. I tried to stay as close as possible to your programming style of the dymosim class. 3f22fdc

If we need some other features, let me know!

[kdavies4]

I got the following error when I tried simulations.ipynb:
AttributeError: 'pyfmi.fmi.FMUModelCS1' object has no attribute 'setup_experiment

I got a similar error with using ME instead of CS:
AttributeError: 'pyfmi.fmi.FMUModelME1' object has no attribute 'setup_experiment'

When I commented out the self.fmu.setup_experiment() line of exps.simulators.fmi.load and generated a FMU with a binary for my platform, that IPython notebook worked. What do you think happened? Do we have different versions of PyFMI? I think I have version 2.0.

[kdavies4]

Otherwise, we mostly just need to clean things up (code, docstrings, examples). I need to do a sensitivity analysis (Morris method) using the exps module for my research. If that leads to any useful functions I'll add them. I'd like to release all this in a new version before the next Activity 1.4 mtg (mid to late Jan).

Good work!

[kdavies4]

Here's an idea: instead of internally storing memory_result, run() and continue_run() could return the results as SimRes instances. The best way might be to extend the pyfmi results class (ResultBase or JMResultBase) on SimRes, but that would require agreement from the PyFMI developers.

[arnoutaertgeerts]

The AttributeError was due to a different version of the FMU, not of PyFMI. I fixed this in 1e0027f.

[kdavies4]

Great!