探索规范化

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()