how do you generate two N(0,1)(standard normal distribution)random variables with corr rho if you have a random number generator for standard normal distribution

视频信息

答案文本 复制

视频字幕 复制

Today we'll learn how to generate two correlated standard normal random variables. Given a random number generator for standard normal distribution, we want to create two variables X and Y that follow N(0,1) and have a specific correlation coefficient rho. The method is straightforward and elegant. First, generate two independent standard normal random variables Z1 and Z2. Then set X equal to Z1. For the second variable, set Y equal to rho times Z1 plus the square root of one minus rho squared times Z2. This linear combination ensures that both X and Y are standard normal with the desired correlation rho. Let's verify why this method works mathematically. X equals Z1, so it's clearly standard normal. For Y, we need to check that it has mean zero and variance one. The variance of Y equals rho squared plus one minus rho squared, which equals one. The covariance between X and Y equals rho times the variance of Z1 plus zero, which gives us exactly rho. This confirms our method produces the desired correlation. Let's visualize how different correlation values affect the relationship between X and Y. Starting with strong negative correlation, we see the points form a downward sloping pattern. As we increase rho towards zero, the correlation weakens and points become more scattered. With positive correlation, points slope upward, and at rho equals one, they would form a perfect line. In summary, this elegant method uses a simple linear transformation to generate correlated normal variables. The algorithm is straightforward: generate two independent standard normals, set X equal to the first one, and set Y as a weighted combination. This technique is widely used in Monte Carlo simulations, financial risk modeling, and statistical applications where controlling correlation is essential.