default_selection_score

Function default_selection_score 

Source
pub fn default_selection_score<const N: usize>(
    value: f64,
    policy: f64,
    n_visits: f64,
    parent_n_visits: f64,
    exploration_coef: f64,
) -> f64
Expand description

A selection function that combines value, visit count, and initial policy.

This function prioritizes unvisited nodes. For visited nodes, it balances exploitation (node’s value) with an exploration term that incorporates the initial policy probability.

§Parameters

  • value: The mean value of the node.
  • policy: The initial policy probability for the action leading to this node.
  • n_visits: Number of visits to the current node.
  • parent_n_visits: Number of visits to the parent node.
  • exploration_coef: The exploration coefficient.

§Returns

The calculated selection score for the node.