`predicted_draws()` is missing `sim()`'s `vars` argument

[jonathan-g]

In the example on p. 143-144 of Rethinking, 2nd ed, we have

# d has columns A, M, and D
m5.3A <- quap(
  alist(
  D ~ dnorm(mu, sigma).
  mu <- a + bM * M + bA * A,
  a ~ dnorm(0, 0.2),
  bM ~ dnorm(0, 0.5).
  bA ~ dnorm(0, 0.5),
  sigma ~ dexp(1),
  M ~ dnorm(mu_M, sigma_M),
  mu_M <- aM + bAM * A,
  aM ~ dnorm(0, 0.2),
  bAM ~ dnorm(0, 0.5),
  sigma_M ~ dexp(1)
  ), data = d
)

A_seq <- seq(2, 2, length.out = 30)

# notice, no M in sim_dat
sim_dat <- data.frame(A = A_seq) 

# Use the model first to infer M from A and then to 
# infer D from M and A
s <- sim(m5.3A, data = sim_dat, vars = c("M", "D"))

I can't figure out a way to do this with add_linpred_draws(). When I try

s <- predicted_draws(m5.3A, sim_dat)

I get an error message about "M" not found.

More generally, is there a way to use predicted_draws to extract predictions of M from this model, given a data frame with values of A?

[jonathan-g]

If I call

s <- predicted_draws(m5.3A, sim_dat, vars = c("M", "D"))

I get an error

Error: `draws` must have exactly two dimensions. It has 0

[jonathan-g]

I've submitted PR #17 to fix this problem.