rendering bugs in log/log-base-2 histograms

[haberman]

My attempts to plot log/log-base-2 histograms were unsuccessful:

http://nbviewer.ipython.org/gist/haberman/4c7cc0697ab5c42568f5

Sorry if this is a dup.

[tacaswell]

Please send this to the mailing list or post in on stackoverflow. Ask to re-open if this you think this is actually a bug.

[haberman]

It seemed kind of self-evident that the first rendering isn't correct. Does that look like a histogram to you?

The second example in the notebook I linked has been identified as something that used to work but appears to have regressed in 1.3.1: http://stackoverflow.com/questions/16147827/histogram-with-base-2-logarithmic-y-scale-in-matplotlib

[tacaswell]

It was not clear from the issue that this was a bug-report, not a request for getting help to make code work. Can you also please paste code + images directly into the issue? The link to your notebook will rot eventually.

[tacaswell]

Actually, I am going to close this again, both blocks of code work as expected for me on close-to-master:

import random
import numpy
from matplotlib import pyplot, axes
import pylab
import math

pylab.rcParams['figure.figsize'] = (14.0, 8.0)

bins = numpy.logspace(-64, 64, base=2)

def bin_iter(bins):
    for i in range(len(bins) - 1):
        yield (bins[i], bins[i + 1])

x = [long(high - low) for low, high in bin_iter(bins)]
y = [1 for _ in bin_iter(bins)]

fig = pyplot.figure()
plot = fig.add_subplot(111)
plot.set_yscale("log", basey=2)
plot.set_xscale("log", basex=2)
plot.hist(x, bins, alpha=0.6, label='x', cumulative=True, histtype='stepfilled')
plot.legend(loc='upper right')
pyplot.show()

import numpy as np
import matplotlib.pyplot as plt
mu, sigma = 100, 15
fig, ax = plt.subplots()
x = mu + sigma * np.random.randn(10000)
ax.set_yscale('log', basey=2)
n, bins, histpatches = ax.hist(x, 50, facecolor='green', alpha=0.75)
plt.show()