Struct TreeLearning

pub struct TreeLearning<'a, 'b> {
    pub size: usize,
    root: SubProb<'a>,
    pub subproblems: Vec<SubProb<'a>>,
    limit: usize,
    pub conditions: &'b [(&'static Expr, Bits)],
    pub options: Vec<(&'static Expr, Bits)>,
    pub bump: &'a Bump,
    pub solved: bool,
}

A struct encapsulating the state and parameters for a decision tree learning process in string synthesis.

Fields

size: usize

root: SubProb<'a>

subproblems: Vec<SubProb<'a>>

limit: usize

conditions: &'b [(&'static Expr, Bits)]

options: Vec<(&'static Expr, Bits)>

bump: &'a Bump

solved: bool

Implementations

impl<'a, 'b> TreeLearning<'a, 'b>

pub fn new_in( size: usize, conditions: &'b [(&'static Expr, Bits)], options: Vec<(&'static Expr, Bits)>, bump: &'a Bump, limit: usize, ) -> Self

Creates a new instance with specified parameters including size, conditions, options, memory allocator, and limit.

pub fn entropy(&self, bits: &Bits) -> f32

Calculates the entropy of a given set of bits within the context of the TreeLearning algorithm’s options.

pub fn cond_entropy( &self, bits: &Bits, condition: &Bits, ) -> (f32, (Bits, f32), (Bits, f32))

Calculates the conditional entropy of a given set of bits based on a specified condition bitset.

pub fn select(&self, unsolved: &SubProblem<'a>) -> SelectResult

Determines the next action for an unsolved subproblem in the tree learning process.

pub fn run(&mut self) -> bool

Executes the learning algorithm by iterating over the subproblems within the decision tree.

fn fmt_recursive( &self, f: &mut Formatter<'_>, node: SubProb<'a>, indent: &mut String, ) -> Result

Facilitates the recursive formatting of the decision tree contained within the TreeLearning structure for display purposes.

fn size_recursive(&self, node: SubProb<'a>) -> usize

Determines the size of the decision tree by recursively traversing through its nodes.

fn cover_recursive(&self, node: SubProb<'a>) -> Bits

Covers a decision tree recursively starting from a given node and determining the set of bits covered by the tree structure.

fn expr_recursizve(&self, node: SubProb<'a>) -> &'static Expr

Returns the expression representation of a given node in the decision tree.

fn unsolved_recursive(&self, node: SubProb<'a>, result: &mut Vec<Box<[u128]>>)

Recursively traverses through a decision tree to collect bits from unsolved subproblems.

fn unsolved(&self) -> Vec<Box<[u128]>>

Returns a vector of boxed slices containing u128 values that represent unsolved components of a decision tree.

pub fn expr(&self) -> &'static Expr

Returns the expression associated with the root of the decision tree. This function utilizes a recursive approach by invoking expr_recursizve on the tree’s root node to retrieve the expression efficiently, leveraging the recursive structure to navigate through potentially complex tree configurations within the TreeLearning context.

pub fn result_size(&self) -> usize

Calculates the result size of a decision tree by recursively determining the size starting from the root node. This implementation utilizes the size_recursive function on the root to compute the cumulative size of the tree structure, which includes all subproblems, branches, and accepted solutions present in the tree.

Trait Implementations

impl<'a, 'b> Debug for TreeLearning<'a, 'b>

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

Formats the decision tree within the TreeLearning instance for display.