Accessing traced variables within a @gen macro

[sdwfrost]

I have a problem where I draw a vector of random numbers.

# Priors
beta_mean ~ uniform(0.0, 1.0)
beta_sd ~ uniform(0.0, 0.1)
for i in 1:nsteps
      {(:βi,i)} ~ normal(beta_mean, beta_sd)
 end

I want to pass this series of addresses as a vector to another function in order to get a derived variable within the same @gen macro (so I don't have access to the trace) - how can this be done?

[ztangent]

You can do assignment in addition to sampling. The ~ syntax in a @gen function is just syntactic sugar (as explained here: https://www.gen.dev/docs/dev/ref/modeling/dml/#Tilde-syntax)

For your example, you can do:

# Priors
beta_mean ~ uniform(0.0, 1.0)
beta_sd ~ uniform(0.0, 0.1)
beta = zeros(nsteps)
for i in 1:nsteps
      beta[i] = {(:βi,i)} ~ normal(beta_mean, beta_sd)
 end

[sdwfrost]

Thanks! This now works great.