Struct QuadraticProgrammingLayer 

pub struct QuadraticProgrammingLayer {
    pub n_vars: usize,
    pub n_eq: usize,
    pub n_ineq: usize,
}

Quadratic programming layer: min 0.5 x^T Q x + c^T x s.t. Ax = b, Gx ≤ h

Fields

n_vars: usize

Problem dimension

n_eq: usize

Number of equality constraints

n_ineq: usize

Number of inequality constraints

Implementations

impl QuadraticProgrammingLayer

pub fn new(n_vars: usize, n_eq: usize, n_ineq: usize) -> Self

pub fn forward( &self, q: &Tensor, c: &Tensor, a: &Tensor, b: &Tensor, g: &Tensor, h: &Tensor, config: &OptimizationConfig, ) -> Result<OptimizationSolution>

Forward pass: solve QP

pub fn backward( &self, solution: &OptimizationSolution, q: &Tensor, c: &Tensor, a: &Tensor, b: &Tensor, g: &Tensor, h: &Tensor, downstream_grad: &Tensor, config: &OptimizationConfig, ) -> Result<Vec<Tensor>>

Backward pass: differentiate through QP solution

Trait Implementations

impl DifferentiableOptimization for QuadraticProgrammingLayer

fn solve( &self, parameters: &[&Tensor], config: &OptimizationConfig, ) -> Result<OptimizationSolution>

Solve the optimization problem

fn differentiate( &self, solution: &OptimizationSolution, parameters: &[&Tensor], downstream_grad: &Tensor, config: &OptimizationConfig, ) -> Result<Vec<Tensor>>

Compute derivatives of the solution w.r.t. parameters

fn problem_type(&self) -> OptimizationProblem

Get the type of optimization problem