Skip to contents

Compare outcomes under different branch policies to quantify whether free branching is balanced.

Usage

mxo_branch_study(
  policy_factory,
  n_games = 30L,
  config = mxo_config_default(),
  seed = NULL
)

Arguments

policy_factory

A function function(branch_policy) mxo_policy(...) that returns a policy configured with the given branch_policy.

n_games

Integer scalar, games per branch policy.

config

A config list (see mxo_config_default()).

seed

Optional integer base seed.

Value

A tibble with one row per branch policy: branch_policy, x_win_rate, o_win_rate, draw_rate, mean_plies, mean_timelines, cross_timeline_fraction.

Examples

factory <- function(branch_policy) {
  mxo_policy("random", branch_policy = branch_policy)
}
mxo_branch_study(factory, n_games = 2L,
                 config = mxo_config_default(n = 3L, ply_cap = 6L),
                 seed = 1L)
#> # A tibble: 3 × 7
#>   branch_policy x_win_rate o_win_rate draw_rate mean_plies mean_timelines
#>   <chr>              <dbl>      <dbl>     <dbl>      <dbl>          <dbl>
#> 1 all                    0          0         1          6            3.5
#> 2 promising              0          0         1          6            4  
#> 3 none                   0          0         1          6            1  
#> # ℹ 1 more variable: cross_timeline_fraction <dbl>