Generate CI-focused control settings for batch-means inference

Convenience wrapper for control_opt() with defaults tailored for Xi-style batch-means confidence intervals.

Usage

control_opt_batch_ci(
  optimizer = sgd(stepsize = 0.03),
  burnin = 100,
  iterations = 2000,
  n_batch = 20,
  n_parallel_chain = 4,
  alpha = 0.501,
  t0 = 1,
  schedule_burnin_iter = 0,
  ...
)

Arguments

optimizer

optimizer object, default sgd(stepsize = 0.03).

burnin

burn-in iterations before optimization.

iterations

optimization iterations.

n_batch

number of checkpoints.

n_parallel_chain

number of parallel chains.

alpha

polynomial stepsize exponent for poly_decay(alpha, t0).

t0

non-negative schedule offset.

schedule_burnin_iter

non-negative integer. Initial optimization iterations where polynomial schedule scaling is disabled.

...

additional arguments forwarded to control_opt() to override defaults.

Value

object of class control_opt.

Details

This helper enforces trajectory-friendly defaults:

• store_traj = TRUE

• trend_std_conv_check = FALSE

• R_hat_conv_check = FALSE

• pflug_conv_check = FALSE

• stepsize_control = poly_decay(alpha, t0, schedule_burnin_iter)

Any of these can still be overridden through ....