Why must a multivariate-normal QMC transform account for covariance rather than only applying scalar normal quantiles?
MathematicsLevel: IntermediateSimulation design
Why must a multivariate-normal QMC transform account for covariance rather than only applying scalar normal quantiles?
Hints
- Marginals do not specify the joint law.
- Apply a covariance factor.
Answer
Scalar quantiles produce standard-normal marginals; a covariance factor is needed to impose dependence. Otherwise the simulated joint model has the wrong correlations.
Source and licence: SciPy — quasi-Monte Carlo designs · BSD-3-Clause · answer: Quantoros editorial