Struct CompiledExpr 

pub struct CompiledExpr {
    pub results: Vec<f64>,
    pub gradients: Vec<f64>,
    /* private fields */
}

A compiled version of an expression tree.

This struct represents a compiled version of an expression tree, which is faster to compute and backpropagate through than the regular expression tree.

Fields

results: Vec<f64>

The results of the operations in the expression.

gradients: Vec<f64>

The gradients of the operations in the expression.

Implementations

impl CompiledExpr

pub fn from_expr(expr: Expr) -> Self

Creates a new CompiledExpr from an expression.

This method consumes the expression and transforms it into a compiled form that is more efficient for computation and backpropagation.

Example:

use alpha_micrograd_rust::value::Expr;
use alpha_micrograd_rust::compiled::CompiledExpr;

let expr = Expr::new_leaf(1.0);
let expr2 = expr.tanh();
let compiled = CompiledExpr::from_expr(expr2);

pub fn recalculate(&mut self)

Recalculates the expression based on the current values.

This method recalculates the expression based on the current values of the parameters. It is more efficient than recalculating the expression tree, as it iterates through an array of operations instead of traversing a tree structure.

Example:

use alpha_micrograd_rust::value::Expr;
use alpha_micrograd_rust::compiled::CompiledExpr;

let expr = Expr::new_leaf_with_name(1.0, "x");
let expr2 = expr.tanh();
let mut compiled = CompiledExpr::from_expr(expr2);
assert_eq!(compiled.result(), 0.7615941559557649);
 
// Modify the value of "x"
compiled.set("x", 2.0);
compiled.recalculate();
 
assert_eq!(compiled.result(), 0.9640275800758169);

pub fn learn(&mut self, learning_rate: f64)

Performs one step of learning (backpropagation) on the compiled expression.

This function updates the values of the learnable parameters in the expression based on the gradients calculated during backpropagation.

Arguments

• learning_rate - The learning rate to use for updating the parameters.

Returns

Returns nothing. The results are updated in place.

Applies backpropagation to the expression, updating the values of the gradients and the expression itself.

This method will change the gradients based on the gradient of the last expression in the calculation graph. After adjusting the gradients, the method will update the values of the individual expression nodes (parameters) to minimize the loss function.

Example:

use alpha_micrograd_rust::value::Expr;
use alpha_micrograd_rust::compiled::CompiledExpr;

let expr = Expr::new_leaf(1.0);
let expr2 = expr.tanh();
let mut compiled = CompiledExpr::from_expr(expr2);
compiled.learn(1e-09);
compiled.recalculate();

pub fn result(&self) -> f64

Returns the final result of the compiled expression.

This function returns the last result in the results vector, which corresponds to the final output of the expression.

Returns

Returns the final result as a f64 value.

pub fn get_grad_by_name(&self, name: &str) -> Option<f64>

Gets the gradient of a learnable parameter by its name.

This function retrieves the gradient of a learnable parameter (e.g., a weight or bias) by looking up its name in the names-to-index mapping.

Arguments

• name - The name of the parameter to get the gradient for.

Returns

Returns an Option<f64> containing the gradient value if found, or None if not found.

pub fn set(&mut self, name: &str, value: f64)

Sets the value of a parameter by its name.

This method sets the value of a parameter in the compiled expression. It is used to modify the values of leaf nodes in the expression tree.

Example:

use alpha_micrograd_rust::value::Expr;
use alpha_micrograd_rust::compiled::CompiledExpr;

let expr = Expr::new_leaf_with_name(1.0, "x");
let expr2 = expr.tanh();
let mut compiled = CompiledExpr::from_expr(expr2);
 
compiled.set("x", 2.0);
compiled.recalculate();
 
assert_eq!(compiled.result(), 0.9640275800758169);