Skip to content

Updating Plots

In bqplot almost all the attributes of plotting widgets (Figure, Mark, Scale, Axis) are implemented as traitlets, so the plots are responsive to data updates. All you need to do is update the attributes in place without having to recreate the figure and mark objects.

Updating the attributes of figure and marks automatically triggers a re-draw of the plots on the front-end.

Let's look at an example:

import numpy as np
import bqplot.pyplot as plt

x = np.linspace(-10, 10, 100)
y = np.sin(x)

fig = plt.figure()
curve = plt.plot(x, y)

Updating Single Data Attribute

Let's look at the correct (and incorrect) way of updating the plots:


Only update the figure and mark attributes like so:

curve.y = np.cos(x)
fig.title = "Cosine"


Do not create figure and mark objects again to update the plots!

fig = plt.figure(title="Cosine")
curve = plt.plot(x, np.cos(y))

Updating Multiple Data Attributes

We can update multiple attributes of the mark object simultaneously by using the hold_sync method like so. (This makes only one round trip from the python kernel to the front end)


Only one network trip

# use the hold_sync() context manager
with curve.hold_sync():
    curve.x = np.linspace(-20, 20, 200)
    curve.y = np.cos(x)


Makes two network trips

curve.x = np.linspace(-20, 20, 200)
curve.y = np.cos(x)


We can enable animated data updates by passing in animation_duration (in milliseconds) to the figure. Let's look at an example to update a scatter plot

x, y = np.random.rand(2, 10)

fig = plt.figure(animation_duration=1000) # 1000 ms or 1s
scat = plt.scatter(x=x, y=y)
# update the scatter plot to use animations (1s duration)
with scat.hold_sync():
    scat.x, scat.y = np.random.rand(2, 10)