Frequently asked questions: general

Why is my run in the lost state?

The Simvue client sends a heartbeat to the server every minute. A run goes into the lost state if there are no heartbeats for over 3 minutes. Because of this, it is preferable to use the context manager for the Run() object, for example:

from simvue import Run

if __name__ == "__main__":
    with Run() as run:
        run.init()
        ...

If the user's code exits with an exception the state of the run will be set to failed and an event will be created with details about this exception. Without the context manager, i.e.

from simvue import Run

if __name__ == "__main__":
    run = Run()
    run.init()
    ...

the run automatically be set to the lost state if there is an exception and no information is available as to what happened.

Interactive plots from artifacts

If Matplotlib and Plotly plots are saved directly as artifacts (rather than saved as files first) they can be opened in the web UI as interactive plots. This makes use of plotly.js - see here for documentation for plotly.js.

Here is a simple but complete example creating a Matplotlib plot (based on an example from the Matplotlib documentation, seen here) and saving it as an artifact. Note the use of the gcf() method to get the current figure, as we need to pass a matplotlib.figure.Figure instance to Simvue.

import numpy as np
import matplotlib.pyplot as plt
from simvue import Run

def f(t):
    return np.exp(-t) * np.cos(2*np.pi*t)

if __name__ == "__main__":
    t1 = np.arange(0.0, 5.0, 0.1)
    t2 = np.arange(0.0, 5.0, 0.02)

    plt.figure()
    plt.subplot(211)
    plt.plot(t1, f(t1), 'bo', t2, f(t2), 'k')

    plt.subplot(212)
    plt.plot(t2, np.cos(2*np.pi*t2), 'r--')

    with Run() as run:
        run.init()
        run.save_object(plt.gcf(), 'output', name='plot')

Note that we have provided a name for the artifact using the name argument, which is essential when creating artifacts directly from Python objects. This name is completely arbitrary and up to the user.

If you get an error like:

Aw. Snap! You're gonna have to hold off on the selfies for now. Plotly can't import images from matplotlib yet!

or:

AttributeError: 'PathCollection' object has no attribute 'get_offset_position

it means that your plot is not compatible with the Plotly conversion function (see documentation for the Plotly conversion function). In this situation the only option currently is to try using Plotly rather than Matplotlib to create the plot. If you are unfamiliar with this, view the full Plotly documentation.