Matplotlib is hiring a Research Software Engineering Fellow! See discourse for details. Apply by January 3, 2020

Version 3.1.1
matplotlib
Fork me on GitHub

目录

Related Topics

探索规范化

多元正态分布的各种归一化。

import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
import numpy as np
from numpy.random import multivariate_normal

data = np.vstack([
    multivariate_normal([10, 10], [[3, 2], [2, 3]], size=100000),
    multivariate_normal([30, 20], [[2, 3], [1, 3]], size=1000)
])

gammas = [0.8, 0.5, 0.3]

fig, axes = plt.subplots(nrows=2, ncols=2)

axes[0, 0].set_title('Linear normalization')
axes[0, 0].hist2d(data[:, 0], data[:, 1], bins=100)

for ax, gamma in zip(axes.flat[1:], gammas):
    ax.set_title(r'Power law $(\gamma=%1.1f)$' % gamma)
    ax.hist2d(data[:, 0], data[:, 1],
              bins=100, norm=mcolors.PowerNorm(gamma))

fig.tight_layout()

plt.show()
探索规范化