Enum SolveMethod 

#[non_exhaustive]
pub enum SolveMethod {
    Full,
    Sampled,
    External,
}

The method to use for finding approximate equilibria

When in doubt, you’ll probably want to default to External¹ as it generally converges faster than Full² or Sampled¹, which are mostly provided for completeness.

---

1. Lanctot, Marc, et al. “Monte Carlo sampling for regret minimization in extensive games.” Advances in neural information processing systems 22 (2009). ↩ ^1 2

2. Zinkevich, Martin, et al. “Regret minimization in games with incomplete information.” Advances in neural information processing systems 20 (2007). ↩

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

Full

This method indicates vanilla counterfactual regret minimization, which does no random sampling. This can be good for small games, espcially ones with very unlikely chance outcomes, but otherwise spends a lot of computation exploring unimportant areas of the game tree.

Sampled

This method indicates chance sampled counterfactual regret minimization, which samples outcomes at chance nodes, but fully explores player actions. This often performs better than full exploration, but may produce worse results if there are infrequent but very relevant chance outcomes.

Since this is sampled, there’s a chance that it terminates early with a small regret bound that’s slighly incorrect because it didn’t sample enough chance outcomes.

External

This method indicates external sampled counterfactual regret minimization, which alternates between players, and only fully explores the actions of one player, while sampling the actions of the other according to their current strategy. This often converges faster than the other methods because it doesn’t explore sections of the game tree with low value.

Since this is sampled, there’s a chance that it terminates early with a small regret bound that’s slighly incorrect because it didn’t sample enough chance outcomes.

Trait Implementations

impl Clone for SolveMethod

fn clone(&self) -> SolveMethod

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SolveMethod

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for SolveMethod

fn eq(&self, other: &SolveMethod) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for SolveMethod

impl Eq for SolveMethod

impl StructuralPartialEq for SolveMethod