I have a complicated algorithm that updates 3 histograms that are stored in arrays. I want to debug my algorithm, so I was thinking of showing the arrays as histograms in a user interface. What is the easiest way to do this. (Rapid application development is more important than optimized code.)
I have some experience with Qt (in C++) and some experience with matplotlib.
(I'm going to leave this question open for a day or two because it's hard for me to evaluate the solutions without a lot more experience that I don't have. Hopefully, the community's votes will help choose the best answer.)
Edit: Nowadays, it is easier and better to use
import numpy as np import matplotlib.pyplot as plt import matplotlib.animation as animation def animate(frameno): x = mu + sigma * np.random.randn(10000) n, _ = np.histogram(x, bins, normed=True) for rect, h in zip(patches, n): rect.set_height(h) return patches mu, sigma = 100, 15 fig, ax = plt.subplots() x = mu + sigma * np.random.randn(10000) n, bins, patches = plt.hist(x, 50, normed=1, facecolor='green', alpha=0.75) ani = animation.FuncAnimation(fig, animate, blit=True, interval=10, repeat=True) plt.show()
There is an example of making an animated graph here. Building on this example, you might try something like:
import numpy as np import matplotlib.pyplot as plt plt.ion() mu, sigma = 100, 15 fig = plt.figure() x = mu + sigma*np.random.randn(10000) n, bins, patches = plt.hist(x, 50, normed=1, facecolor='green', alpha=0.75) for i in range(50): x = mu + sigma*np.random.randn(10000) n, bins = np.histogram(x, bins, normed=True) for rect,h in zip(patches,n): rect.set_height(h) fig.canvas.draw()
I can get about 14 frames per second this way, compared to 4 frames per second using the code I first posted. The trick is to avoid asking matplotlib to draw complete figures. Instead call
plt.hist once, then manipulate the existing
patches to update the histogram, and call
fig.canvas.draw() to make the updates visible.
For realtime plotting, I recommend trying Chaco, pyqtgraph, or any of the opengl-based libraries like glumpy or visvis. Matplotlib, wonderful as it is, is generally not suitable for this kind of application.
Edit: the developers of glumpy, visvis, galry, and pyqtgraph are all collaborating on a visualization library called vispy. It is still early in development, but promising and already quite powerful.