Given a multivariate normal distribution with an autoregressive covariance structure, for instance N(0,sigmaG(rho)), where sigma is the variance and G is a block diagonal matrix parametrized by rho (i.e. serial correlation parameter). The diagonal components of G say G_{i} are the autoregressive covariance matrices.

I am considering a transformation of this distribution to symmetric mixtures of two multivariate normal distributions having the same covariance structure, for instance

0.5*N(-0.2,sigma_{1}G(rho_{1})) + 0.5*N(0.2,sigma_{2}G(rho_{2}))

Suppose sigma=0.3 and rho=0.5. My question is how can we find the numerical values of the variance and the serial correlation parameters in the mixture distribution.

I greatly appreciate your help.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.