使用numpy生成服从多元正态分布（multivariate normal distribution）的样本

代码

import numpy as np
mean = [0, 0]
sigam1 = 1
sigam2 = 1

rho = 0  # 相关系数
cov12 = rho*sigam1*sigam2  # 协方差
cov = [
    [sigam1**2, cov12],
    [cov12, sigam2**2]
]  # 协方差矩阵
x1, y1 = np.random.multivariate_normal(
    mean,
    cov,
    5000
).T

rho = 0.5
cov12 = rho*sigam1*sigam2
cov = [
    [sigam1**2, cov12],
    [cov12, sigam2**2]
]
x2, y2 = np.random.multivariate_normal(
    mean,
    cov,
    5000
).T

rho = 1
cov12 = rho*sigam1*sigam2
cov = [
    [sigam1**2, cov12],
    [cov12, sigam2**2]
]
x3, y3 = np.random.multivariate_normal(
    mean,
    cov,
    5000
).T

import matplotlib.pyplot as plt
plt.plot(x1, y1, '.', label='rho=0')
plt.plot(x2, y2, 'x', label='rho=0.5')
plt.plot(x3, y3, '1', label='rho=1')
plt.axis('equal')
plt.legend()
plt.show()

结果图示

see also

1. https://numpy.org/doc/stable/reference/random/generated/numpy.random.multivariate_normal.html