Generating Correlated Normals From Independent Ones

Reported in interviews at Barclays

Problem statement

How would you generate correlated normal variables from independent standard normal draws? You have a random number generator for the independent draws and want a specified correlation or covariance. Give a construction, prove its moments, and extend it to a mean vector and covariance matrix.

Show hint

Which family of transformations keeps normal variables normal, and how does it change a covariance matrix?

Relevant roles: Data Scientist, Applied Scientist, Research Scientist and ML Engineer.

Related free problems