Reparameterization of correlation parameter for improved computation?

Modeling
2

Yeah these \sqrt{1 - \rho^2} terms can cause problems.

Can you change:

chi_raw[ik] ~ normal(rhochi*chi_raw[FwdConnectionSeqID[ik,1]],sqrt((1-rhochi^2)));

and:

chi[ik]=chi_raw[ik] * sigma;

to

chi_raw[ik] ~ normal(0, 1);

and

chi[ik] = rhochi*chi[FwdConnectionSeqID[ik,1]] + chi_raw[ik] * sqrt((1-rhochi^2)) * sigma;

and similarly for lambda? Is that possible here?