The scientific community often runs iterative code, often in the form of
simulation. It's often useful to see the results after each iteration.
Accordingly, MATLAB® has a nice feature that allows you to update the
figure, drawnow. This repo brings the same feature to Python's
matplotlib, with some extras.
Example:
This is shown with imshow, but python-drawnow allows updates of any
figure.
Usage:
# complete implementation of script found in test/test.py
from pylab import *
from drawnow import drawnow, figure
# if global namespace, import plt.figure before drawnow.figure
def approx(x, k):
"""Approximate x with k singular values"""
...
figure(figsize=(7, 7/2))
def draw_fig():
subplot(1, 2, 1)
imshow(x)
subplot(1, 2, 2)
imshow(x_hat)
#show()
x = imread('test-data/mandrill.png').mean(axis=2)
k_values = around(logspace(0, 2, num=10))
for k in k_values:
x_hat = approx(x, k)
drawnow(draw_fig)If you want to wait for confirmation after update or the option to drop
into a debugger, call drawnow(function_to_draw_figure, confirm=True).
If you only want to show the figure once, call
drawnow(function_to_draw_figure, show_once=True). The full
documentation is included in the doc strings. Use drawnow? or
help(drawnow) to see these docs.
Try running the folloowing code in a Jupyter input cell/in the console/etc:
%matplotlib
This will disable the Matplotlin inline mode and use the default plotting backend. For more detail, see the IPython plotting documentation.
Two options:
- Run
pip install drawnow. - Download this repository and run
python setup.py install.
Option 2 assumes a working Python installation with pip. I suggest
Anaconda's distribution: https://www.anaconda.com/distribution/ For
other options, see https://realpython.com/installing-python/.
This does require small changes to your code. All it should really
amount to is moving figure(); plot(...); show() inside a function; not
much.