Hi,

To constrain hierarchical parameters arising from a bivariate normal distribution ( \beta_{1,i} ,\beta_{2,i}) \sim N( [\mu_{1}, \mu_{2}] , [\sigma_{1}, \sigma_{2}] ) such that \beta_{1,i} \ge \beta_{2,i} we can constrain the parameters such that:

z_{i,1} \ge (\mu_{2} - \mu_{1} + \sigma_{2}*z_{i,2}*\sqrt{1-\rho}) / (\sigma_{1} - \rho*\sigma_{2})

using off-centred parameterisations.

I am trying to do this in Stan by declaring:

```
parameters {
ordered[2] a1_m_raw[num_refs];
vector<lower=0>[2] sd[nt];
cholesky_factor_corr[2] L_Omega[nt];
vector<lower=(to_vector(a1_m_raw[ , 1]) - to_vector(a1_m_raw[ , 2]) + sd[1,2]*z2*L_Omega[1,2,2])/(sd[1,1] - L_Omega[1,1,2]*sd[1,2])>[NS] z1; // Se + Fp > 1
```

However, I am getting complication errors when attempting to compile the code. It is very long error message but I will copy-paste a screenshot of part of the code below:

Would anyone have any idea why this is happening? When I declare a similar constraint for 2 univariate normals (i.e. no L_Omegaâ€™s needed in the constraint) it works fine

Thanks