return value of optimizing omits log_p and log_g

Summary:

The documentation for rstan::optimizing says that the return value includes components log_p and log_g if draws>0. It does not.

Description:

See above.

Reproducible Steps:

Run the following lines:

m <- rstan::stan_model(model_code = 'data { int N; vector[N] y; } parameters {real mu; real<lower=0> sigma; } transformed parameters { real mu1 = 2.0 * mu; } model {y ~ normal(mu,sigma); }')
f <- rstan::optimizing(m, hessian = TRUE, draws=10, data=list(N=1000, y=rnorm(1000, 0.5, 1)))

Current Output:

The return value f contains only par, value, return_code, hessian, and theta_tilde.

Expected Output:

There should be components log_p, described in the documentation as "a vector of length draws that contains the value of the log-posterior evaluated at each row of theta_tilde", and log_g, described in the documentation as "a vector of length draws that contains the value of the logarithm of the multivariate normal density evaluated at each row of theta_tilde."

RStan Version:

2.18.1

R Version:

R version 3.4.3 (2017-11-30)

Operating System:

macOS High Sierra (Version 10.13.6)

Run the reproducible R example with rstan::optimizing and inspect the returned components of f. Compare those components with the documented log_p and log_g descriptions, then update the implementation or documentation so the documented return value and actual result agree; verify the behavior with a focused test or reproduction.